diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,21222 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 3532,
+  "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": 2.0402,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.773584905660378e-07,
+      "loss": 2.0465,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.660377358490567e-07,
+      "loss": 2.0787,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.547169811320755e-07,
+      "loss": 1.9903,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.433962264150944e-07,
+      "loss": 2.0705,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1320754716981133e-06,
+      "loss": 2.0782,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3207547169811322e-06,
+      "loss": 2.1609,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.509433962264151e-06,
+      "loss": 2.0711,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6981132075471698e-06,
+      "loss": 1.9196,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8867924528301889e-06,
+      "loss": 1.9863,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.075471698113208e-06,
+      "loss": 1.7778,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2641509433962266e-06,
+      "loss": 1.8818,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4528301886792453e-06,
+      "loss": 1.7063,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6415094339622644e-06,
+      "loss": 1.8075,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 1.6259,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.018867924528302e-06,
+      "loss": 1.4204,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.207547169811321e-06,
+      "loss": 1.3659,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3962264150943395e-06,
+      "loss": 1.3107,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5849056603773586e-06,
+      "loss": 1.3484,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7735849056603777e-06,
+      "loss": 1.0751,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.962264150943396e-06,
+      "loss": 1.0267,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.150943396226416e-06,
+      "loss": 1.073,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.339622641509435e-06,
+      "loss": 0.9984,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.528301886792453e-06,
+      "loss": 0.832,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.716981132075472e-06,
+      "loss": 0.8829,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.905660377358491e-06,
+      "loss": 0.8634,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.09433962264151e-06,
+      "loss": 0.8371,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.283018867924529e-06,
+      "loss": 0.775,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4716981132075475e-06,
+      "loss": 0.8543,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 0.8085,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.849056603773585e-06,
+      "loss": 0.7077,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.037735849056604e-06,
+      "loss": 0.6228,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.226415094339623e-06,
+      "loss": 0.6408,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.415094339622642e-06,
+      "loss": 0.6416,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.60377358490566e-06,
+      "loss": 0.5874,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.792452830188679e-06,
+      "loss": 0.7124,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.981132075471699e-06,
+      "loss": 0.6173,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.169811320754717e-06,
+      "loss": 0.674,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.358490566037736e-06,
+      "loss": 0.5855,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5471698113207555e-06,
+      "loss": 0.6024,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.735849056603775e-06,
+      "loss": 0.5438,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.924528301886793e-06,
+      "loss": 0.4874,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.113207547169812e-06,
+      "loss": 0.6156,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.301886792452832e-06,
+      "loss": 0.6169,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.49056603773585e-06,
+      "loss": 0.6958,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.67924528301887e-06,
+      "loss": 0.4946,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.867924528301887e-06,
+      "loss": 0.5401,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.056603773584907e-06,
+      "loss": 0.5046,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.245283018867926e-06,
+      "loss": 0.4916,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 0.5366,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.622641509433963e-06,
+      "loss": 0.4358,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.811320754716981e-06,
+      "loss": 0.6337,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.504,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.018867924528302e-05,
+      "loss": 0.5924,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0377358490566038e-05,
+      "loss": 0.431,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0566037735849058e-05,
+      "loss": 0.4596,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0754716981132076e-05,
+      "loss": 0.4523,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0943396226415095e-05,
+      "loss": 0.4389,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1132075471698115e-05,
+      "loss": 0.5669,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 0.4935,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1509433962264152e-05,
+      "loss": 0.5047,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.169811320754717e-05,
+      "loss": 0.4703,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.188679245283019e-05,
+      "loss": 0.5094,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2075471698113209e-05,
+      "loss": 0.5125,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2264150943396227e-05,
+      "loss": 0.411,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2452830188679246e-05,
+      "loss": 0.4512,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2641509433962264e-05,
+      "loss": 0.5447,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2830188679245283e-05,
+      "loss": 0.5489,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3018867924528303e-05,
+      "loss": 0.4921,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.320754716981132e-05,
+      "loss": 0.4281,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.339622641509434e-05,
+      "loss": 0.4727,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3584905660377358e-05,
+      "loss": 0.4285,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3773584905660378e-05,
+      "loss": 0.4434,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3962264150943397e-05,
+      "loss": 0.3933,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4150943396226415e-05,
+      "loss": 0.5299,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4339622641509435e-05,
+      "loss": 0.5554,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4528301886792452e-05,
+      "loss": 0.4687,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4716981132075472e-05,
+      "loss": 0.4361,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4905660377358491e-05,
+      "loss": 0.4303,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5094339622641511e-05,
+      "loss": 0.4253,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5283018867924532e-05,
+      "loss": 0.4609,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.547169811320755e-05,
+      "loss": 0.503,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5660377358490568e-05,
+      "loss": 0.4637,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5849056603773586e-05,
+      "loss": 0.4278,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6037735849056607e-05,
+      "loss": 0.4215,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6226415094339625e-05,
+      "loss": 0.4197,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6415094339622643e-05,
+      "loss": 0.4076,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6603773584905664e-05,
+      "loss": 0.4071,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.679245283018868e-05,
+      "loss": 0.4412,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 0.4835,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.716981132075472e-05,
+      "loss": 0.4735,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.735849056603774e-05,
+      "loss": 0.4185,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7547169811320756e-05,
+      "loss": 0.4103,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7735849056603774e-05,
+      "loss": 0.449,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7924528301886795e-05,
+      "loss": 0.3959,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8113207547169813e-05,
+      "loss": 0.4117,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.830188679245283e-05,
+      "loss": 0.4672,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8490566037735852e-05,
+      "loss": 0.4283,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.867924528301887e-05,
+      "loss": 0.4693,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 0.4437,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.905660377358491e-05,
+      "loss": 0.3636,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9245283018867927e-05,
+      "loss": 0.3859,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9433962264150945e-05,
+      "loss": 0.445,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9622641509433963e-05,
+      "loss": 0.4204,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9811320754716984e-05,
+      "loss": 0.4943,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.356,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999995795687373e-05,
+      "loss": 0.5222,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999983182753032e-05,
+      "loss": 0.3811,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999996216120758e-05,
+      "loss": 0.4473,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999932731068688e-05,
+      "loss": 0.4229,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999894892361107e-05,
+      "loss": 0.3659,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999848645116656e-05,
+      "loss": 0.4103,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999979398937422e-05,
+      "loss": 0.437,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999973092517976e-05,
+      "loss": 0.5038,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99996594525863e-05,
+      "loss": 0.3931,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999579571653944e-05,
+      "loss": 0.385,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999491282449858e-05,
+      "loss": 0.3757,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999939458504828e-05,
+      "loss": 0.3918,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999928947953052e-05,
+      "loss": 0.4246,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999175965984957e-05,
+      "loss": 0.4217,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999054044507042e-05,
+      "loss": 0.394,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999892371519929e-05,
+      "loss": 0.4884,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998784978171295e-05,
+      "loss": 0.3898,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998637833539716e-05,
+      "loss": 0.3885,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998482281428275e-05,
+      "loss": 0.3304,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998318321967774e-05,
+      "loss": 0.3608,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998145955296085e-05,
+      "loss": 0.4128,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997965181558136e-05,
+      "loss": 0.4305,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999777600090594e-05,
+      "loss": 0.4072,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997578413498567e-05,
+      "loss": 0.4022,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999737241950216e-05,
+      "loss": 0.4187,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999715801908994e-05,
+      "loss": 0.3722,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996935212442183e-05,
+      "loss": 0.4045,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996703999746234e-05,
+      "loss": 0.3783,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996464381196518e-05,
+      "loss": 0.3738,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996216356994518e-05,
+      "loss": 0.4074,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999595992734879e-05,
+      "loss": 0.4133,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995695092474953e-05,
+      "loss": 0.4293,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99954218525957e-05,
+      "loss": 0.3815,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995140207940785e-05,
+      "loss": 0.3844,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999485015874704e-05,
+      "loss": 0.3718,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994551705258344e-05,
+      "loss": 0.366,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994244847725667e-05,
+      "loss": 0.3936,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999392958640703e-05,
+      "loss": 0.3715,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999360592156752e-05,
+      "loss": 0.4351,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993273853479298e-05,
+      "loss": 0.3746,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992933382421593e-05,
+      "loss": 0.3691,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992584508680685e-05,
+      "loss": 0.3471,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992227232549933e-05,
+      "loss": 0.3603,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999186155432976e-05,
+      "loss": 0.3896,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991487474327647e-05,
+      "loss": 0.4416,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991104992858146e-05,
+      "loss": 0.4104,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999071411024287e-05,
+      "loss": 0.4145,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99903148268105e-05,
+      "loss": 0.3847,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989907142896775e-05,
+      "loss": 0.4058,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99894910588445e-05,
+      "loss": 0.3589,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998906657500355e-05,
+      "loss": 0.3602,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988633691730856e-05,
+      "loss": 0.3775,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998819240939041e-05,
+      "loss": 0.399,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987742728353274e-05,
+      "loss": 0.5006,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987284648997566e-05,
+      "loss": 0.3703,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998681817170846e-05,
+      "loss": 0.3715,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986343296878217e-05,
+      "loss": 0.3875,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985860024906128e-05,
+      "loss": 0.3676,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985368356198556e-05,
+      "loss": 0.3977,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984868291168943e-05,
+      "loss": 0.394,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984359830237757e-05,
+      "loss": 0.4455,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998384297383255e-05,
+      "loss": 0.4015,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998331772238793e-05,
+      "loss": 0.4296,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982784076345564e-05,
+      "loss": 0.3696,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982242036154166e-05,
+      "loss": 0.3736,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998169160226952e-05,
+      "loss": 0.3602,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981132775154472e-05,
+      "loss": 0.3906,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998056555527891e-05,
+      "loss": 0.4537,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979989943119794e-05,
+      "loss": 0.4218,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997940593916113e-05,
+      "loss": 0.3569,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997881354389399e-05,
+      "loss": 0.3758,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978212757816492e-05,
+      "loss": 0.4019,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977603581433818e-05,
+      "loss": 0.3787,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99769860152582e-05,
+      "loss": 0.3569,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976360059808927e-05,
+      "loss": 0.3767,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975725715612344e-05,
+      "loss": 0.4503,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997508298320184e-05,
+      "loss": 0.4285,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997443186311787e-05,
+      "loss": 0.4394,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997377235590794e-05,
+      "loss": 0.3588,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973104462126595e-05,
+      "loss": 0.3905,
+      "step": 186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997242818233545e-05,
+      "loss": 0.3707,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997174351710316e-05,
+      "loss": 0.3565,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971050467005435e-05,
+      "loss": 0.3666,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970349032625037e-05,
+      "loss": 0.3718,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996963921455177e-05,
+      "loss": 0.4592,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99689210133825e-05,
+      "loss": 0.3947,
+      "step": 192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968194429721132e-05,
+      "loss": 0.3764,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996745946417862e-05,
+      "loss": 0.3475,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966716117372976e-05,
+      "loss": 0.3619,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965964389929252e-05,
+      "loss": 0.3541,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996520428247954e-05,
+      "loss": 0.3952,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964435795662988e-05,
+      "loss": 0.4318,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963658930125797e-05,
+      "loss": 0.3407,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962873686521194e-05,
+      "loss": 0.3543,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962080065509462e-05,
+      "loss": 0.3912,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996127806775793e-05,
+      "loss": 0.3839,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9960467693940967e-05,
+      "loss": 0.3417,
+      "step": 203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995964894473999e-05,
+      "loss": 0.3567,
+      "step": 204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958821820843448e-05,
+      "loss": 0.3619,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957986322946842e-05,
+      "loss": 0.4749,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995714245175271e-05,
+      "loss": 0.3367,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995629020797063e-05,
+      "loss": 0.3683,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9955429592317222e-05,
+      "loss": 0.3276,
+      "step": 209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995456060551615e-05,
+      "loss": 0.387,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995368324829811e-05,
+      "loss": 0.3544,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9952797521400836e-05,
+      "loss": 0.3457,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951903425569105e-05,
+      "loss": 0.4216,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951000961554728e-05,
+      "loss": 0.3818,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950090130116556e-05,
+      "loss": 0.4163,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949170932020472e-05,
+      "loss": 0.3412,
+      "step": 216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994824336803939e-05,
+      "loss": 0.372,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994730743895327e-05,
+      "loss": 0.3695,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946363145549098e-05,
+      "loss": 0.3511,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994541048862089e-05,
+      "loss": 0.4267,
+      "step": 220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944449468969707e-05,
+      "loss": 0.3924,
+      "step": 221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943480087403634e-05,
+      "loss": 0.3511,
+      "step": 222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994250234473778e-05,
+      "loss": 0.3567,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941516241794295e-05,
+      "loss": 0.3826,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9940521779402365e-05,
+      "loss": 0.3209,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9939518958398187e-05,
+      "loss": 0.354,
+      "step": 226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9938507779624996e-05,
+      "loss": 0.3687,
+      "step": 227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937488243933055e-05,
+      "loss": 0.436,
+      "step": 228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9936460352179655e-05,
+      "loss": 0.4078,
+      "step": 229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9935424105229117e-05,
+      "loss": 0.4037,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993437950395277e-05,
+      "loss": 0.3443,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993332654922899e-05,
+      "loss": 0.3161,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993226524194316e-05,
+      "loss": 0.3707,
+      "step": 233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931195582987698e-05,
+      "loss": 0.3805,
+      "step": 234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930117573262043e-05,
+      "loss": 0.4079,
+      "step": 235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9929031213672646e-05,
+      "loss": 0.3514,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992793650513299e-05,
+      "loss": 0.4261,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9926833448563574e-05,
+      "loss": 0.3418,
+      "step": 238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9925722044891914e-05,
+      "loss": 0.3529,
+      "step": 239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9924602295052554e-05,
+      "loss": 0.3211,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923474199987046e-05,
+      "loss": 0.3506,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992233776064396e-05,
+      "loss": 0.3497,
+      "step": 242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992119297797889e-05,
+      "loss": 0.3366,
+      "step": 243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9920039852954435e-05,
+      "loss": 0.403,
+      "step": 244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9918878386540226e-05,
+      "loss": 0.4247,
+      "step": 245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9917708579712882e-05,
+      "loss": 0.3378,
+      "step": 246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9916530433456063e-05,
+      "loss": 0.3323,
+      "step": 247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9915343948760418e-05,
+      "loss": 0.3541,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9914149126623623e-05,
+      "loss": 0.3531,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912945968050357e-05,
+      "loss": 0.3468,
+      "step": 250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991173447405231e-05,
+      "loss": 0.4209,
+      "step": 251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9910514645648188e-05,
+      "loss": 0.3766,
+      "step": 252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990928648386369e-05,
+      "loss": 0.3863,
+      "step": 253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908049989731538e-05,
+      "loss": 0.3258,
+      "step": 254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9906805164291453e-05,
+      "loss": 0.3397,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9905552008590154e-05,
+      "loss": 0.3762,
+      "step": 256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9904290523681384e-05,
+      "loss": 0.3371,
+      "step": 257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9903020710625874e-05,
+      "loss": 0.3482,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990174257049136e-05,
+      "loss": 0.3791,
+      "step": 259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9900456104352582e-05,
+      "loss": 0.426,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9899161313291284e-05,
+      "loss": 0.4099,
+      "step": 261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897858198396207e-05,
+      "loss": 0.3232,
+      "step": 262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989654676076309e-05,
+      "loss": 0.3507,
+      "step": 263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9895227001494674e-05,
+      "loss": 0.3328,
+      "step": 264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989389892170069e-05,
+      "loss": 0.378,
+      "step": 265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9892562522497877e-05,
+      "loss": 0.3931,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989121780500996e-05,
+      "loss": 0.347,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9889864770367662e-05,
+      "loss": 0.3848,
+      "step": 268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888503419708697e-05,
+      "loss": 0.3457,
+      "step": 269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9887133754177775e-05,
+      "loss": 0.3556,
+      "step": 270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.98857557749266e-05,
+      "loss": 0.3267,
+      "step": 271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9884369483113858e-05,
+      "loss": 0.3487,
+      "step": 272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988297487990523e-05,
+      "loss": 0.3009,
+      "step": 273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9881571966473392e-05,
+      "loss": 0.3557,
+      "step": 274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9880160743997993e-05,
+      "loss": 0.4104,
+      "step": 275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9878741213665683e-05,
+      "loss": 0.3536,
+      "step": 276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987731337667009e-05,
+      "loss": 0.3252,
+      "step": 277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9875877234211828e-05,
+      "loss": 0.3232,
+      "step": 278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9874432787498493e-05,
+      "loss": 0.355,
+      "step": 279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9872980037744672e-05,
+      "loss": 0.3407,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9871518986171923e-05,
+      "loss": 0.3781,
+      "step": 281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987004963400879e-05,
+      "loss": 0.3878,
+      "step": 282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9868571982490797e-05,
+      "loss": 0.3742,
+      "step": 283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986708603286045e-05,
+      "loss": 0.3856,
+      "step": 284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986559178636722e-05,
+      "loss": 0.3493,
+      "step": 285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986408924426757e-05,
+      "loss": 0.3575,
+      "step": 286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986257840782493e-05,
+      "loss": 0.3213,
+      "step": 287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9861059278309703e-05,
+      "loss": 0.3327,
+      "step": 288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985953185699927e-05,
+      "loss": 0.3637,
+      "step": 289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985799614517798e-05,
+      "loss": 0.4191,
+      "step": 290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9856452144137158e-05,
+      "loss": 0.419,
+      "step": 291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9854899855175098e-05,
+      "loss": 0.3294,
+      "step": 292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985333927959706e-05,
+      "loss": 0.3225,
+      "step": 293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985177041871527e-05,
+      "loss": 0.3326,
+      "step": 294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985019327384893e-05,
+      "loss": 0.3312,
+      "step": 295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848607846324195e-05,
+      "loss": 0.3232,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.98470141374742e-05,
+      "loss": 0.4266,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845412148639033e-05,
+      "loss": 0.4033,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984380188116574e-05,
+      "loss": 0.4052,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9842183336408342e-05,
+      "loss": 0.3286,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9840556515727808e-05,
+      "loss": 0.3266,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9838921420492073e-05,
+      "loss": 0.321,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9837278052076027e-05,
+      "loss": 0.3348,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9835626411861515e-05,
+      "loss": 0.3976,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9833966501237344e-05,
+      "loss": 0.347,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9832298321599265e-05,
+      "loss": 0.3657,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983062187434999e-05,
+      "loss": 0.3634,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.982893716089918e-05,
+      "loss": 0.36,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.982724418266345e-05,
+      "loss": 0.3425,
+      "step": 309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982554294106636e-05,
+      "loss": 0.2985,
+      "step": 310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982383343753842e-05,
+      "loss": 0.3202,
+      "step": 311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982211567351709e-05,
+      "loss": 0.3936,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982038965044677e-05,
+      "loss": 0.4131,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981865536977881e-05,
+      "loss": 0.3715,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.98169128329715e-05,
+      "loss": 0.3176,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981516204149008e-05,
+      "loss": 0.3315,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9813402996806714e-05,
+      "loss": 0.3493,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9811635700400523e-05,
+      "loss": 0.3354,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980986015375756e-05,
+      "loss": 0.3141,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980807635837082e-05,
+      "loss": 0.3793,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980628431574022e-05,
+      "loss": 0.4136,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980448402737263e-05,
+      "loss": 0.4139,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980267549478184e-05,
+      "loss": 0.3538,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980085871948858e-05,
+      "loss": 0.3219,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9799033703020512e-05,
+      "loss": 0.3436,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9797200446912216e-05,
+      "loss": 0.3808,
+      "step": 326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979535895270521e-05,
+      "loss": 0.3276,
+      "step": 327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979350922194794e-05,
+      "loss": 0.4174,
+      "step": 328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9791651256195772e-05,
+      "loss": 0.3538,
+      "step": 329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9789785057011006e-05,
+      "loss": 0.4222,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9787910625962855e-05,
+      "loss": 0.3132,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9786027964627457e-05,
+      "loss": 0.3254,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9784137074587872e-05,
+      "loss": 0.3163,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978223795743408e-05,
+      "loss": 0.3175,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9780330614762977e-05,
+      "loss": 0.3832,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9778415048178375e-05,
+      "loss": 0.3957,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9776491259291003e-05,
+      "loss": 0.3526,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97745592497185e-05,
+      "loss": 0.3168,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9772619021085425e-05,
+      "loss": 0.3317,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9770670575023243e-05,
+      "loss": 0.327,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9768713913170325e-05,
+      "loss": 0.3429,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976674903717196e-05,
+      "loss": 0.3621,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9764775948680332e-05,
+      "loss": 0.3672,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9762794649354546e-05,
+      "loss": 0.3828,
+      "step": 344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9760805140860596e-05,
+      "loss": 0.4029,
+      "step": 345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9758807424871385e-05,
+      "loss": 0.3325,
+      "step": 346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975680150306672e-05,
+      "loss": 0.3158,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9754787377133306e-05,
+      "loss": 0.3375,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9752765048764743e-05,
+      "loss": 0.3333,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9750734519661534e-05,
+      "loss": 0.3586,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9748695791531075e-05,
+      "loss": 0.3568,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974664886608765e-05,
+      "loss": 0.3585,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9744593745052452e-05,
+      "loss": 0.3722,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9742530430153548e-05,
+      "loss": 0.3547,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97404589231259e-05,
+      "loss": 0.3311,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973837922571137e-05,
+      "loss": 0.3197,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9736291339658692e-05,
+      "loss": 0.3755,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973419526672349e-05,
+      "loss": 0.3662,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9732091008668277e-05,
+      "loss": 0.4181,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.972997856726244e-05,
+      "loss": 0.3399,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9727857944282254e-05,
+      "loss": 0.3756,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9725729141510873e-05,
+      "loss": 0.3207,
+      "step": 362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9723592160738322e-05,
+      "loss": 0.3585,
+      "step": 363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9721447003761517e-05,
+      "loss": 0.3462,
+      "step": 364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9719293672384237e-05,
+      "loss": 0.3213,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9717132168417132e-05,
+      "loss": 0.3392,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9714962493677733e-05,
+      "loss": 0.3755,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9712784649990442e-05,
+      "loss": 0.376,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9710598639186517e-05,
+      "loss": 0.3167,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9708404463104104e-05,
+      "loss": 0.2889,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.97062021235882e-05,
+      "loss": 0.3298,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970399162249066e-05,
+      "loss": 0.3733,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9701772961670225e-05,
+      "loss": 0.3332,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9699546142992476e-05,
+      "loss": 0.3776,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9697311168329864e-05,
+      "loss": 0.3652,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9695068039561697e-05,
+      "loss": 0.3519,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969281675857413e-05,
+      "loss": 0.2999,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969055732726019e-05,
+      "loss": 0.3419,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9688289747519748e-05,
+      "loss": 0.3277,
+      "step": 379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968601402125952e-05,
+      "loss": 0.318,
+      "step": 380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968373015039308e-05,
+      "loss": 0.3506,
+      "step": 381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9681438136840854e-05,
+      "loss": 0.3539,
+      "step": 382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9679137982530108e-05,
+      "loss": 0.4287,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9676829689394956e-05,
+      "loss": 0.2964,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9674513259376354e-05,
+      "loss": 0.3507,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9672188694422105e-05,
+      "loss": 0.3248,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9669855996486845e-05,
+      "loss": 0.3455,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966751516753205e-05,
+      "loss": 0.3239,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966516620952604e-05,
+      "loss": 0.3365,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9662809124443965e-05,
+      "loss": 0.3888,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9660443914267807e-05,
+      "loss": 0.412,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9658070580986382e-05,
+      "loss": 0.3937,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9655689126595344e-05,
+      "loss": 0.2729,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965329955309716e-05,
+      "loss": 0.3435,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9650901862501136e-05,
+      "loss": 0.3319,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9648496056823404e-05,
+      "loss": 0.3422,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9646082138086915e-05,
+      "loss": 0.3428,
+      "step": 397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9643660108321436e-05,
+      "loss": 0.3556,
+      "step": 398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9641229969563568e-05,
+      "loss": 0.3688,
+      "step": 399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963879172385672e-05,
+      "loss": 0.3488,
+      "step": 400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9636345373251123e-05,
+      "loss": 0.3187,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963389091980382e-05,
+      "loss": 0.304,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963142836557867e-05,
+      "loss": 0.3299,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9628957712646346e-05,
+      "loss": 0.3691,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962647896308432e-05,
+      "loss": 0.3467,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962399211897689e-05,
+      "loss": 0.3747,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9621497182415135e-05,
+      "loss": 0.3224,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961899415549697e-05,
+      "loss": 0.315,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9616483040327084e-05,
+      "loss": 0.3556,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9613963839016988e-05,
+      "loss": 0.3001,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961143655368498e-05,
+      "loss": 0.2948,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9608901186456157e-05,
+      "loss": 0.2954,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960635773946242e-05,
+      "loss": 0.366,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9603806214842457e-05,
+      "loss": 0.4226,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9601246614741748e-05,
+      "loss": 0.33,
+      "step": 415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9598678941312562e-05,
+      "loss": 0.3297,
+      "step": 416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959610319671396e-05,
+      "loss": 0.3147,
+      "step": 417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95935193831118e-05,
+      "loss": 0.3069,
+      "step": 418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95909275026787e-05,
+      "loss": 0.3517,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958832755759408e-05,
+      "loss": 0.3486,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9585719550044138e-05,
+      "loss": 0.3409,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9583103482221848e-05,
+      "loss": 0.3569,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9580479356326963e-05,
+      "loss": 0.3377,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9577847174566016e-05,
+      "loss": 0.3324,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957520693915231e-05,
+      "loss": 0.3151,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9572558652305912e-05,
+      "loss": 0.2923,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9569902316253672e-05,
+      "loss": 0.3101,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9567237933229208e-05,
+      "loss": 0.3722,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956456550547289e-05,
+      "loss": 0.4104,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9561885035231873e-05,
+      "loss": 0.2884,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955919652476006e-05,
+      "loss": 0.3328,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9556499976318116e-05,
+      "loss": 0.3274,
+      "step": 432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9553795392173468e-05,
+      "loss": 0.3283,
+      "step": 433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9551082774600304e-05,
+      "loss": 0.3397,
+      "step": 434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9548362125879555e-05,
+      "loss": 0.3486,
+      "step": 435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9545633448298918e-05,
+      "loss": 0.4073,
+      "step": 436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9542896744152835e-05,
+      "loss": 0.3348,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95401520157425e-05,
+      "loss": 0.3508,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9537399265375843e-05,
+      "loss": 0.2955,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9534638495367563e-05,
+      "loss": 0.3287,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9531869708039076e-05,
+      "loss": 0.3306,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9529092905718555e-05,
+      "loss": 0.3331,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9526308090740913e-05,
+      "loss": 0.3554,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952351526544779e-05,
+      "loss": 0.2884,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9520714432187576e-05,
+      "loss": 0.3876,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951790559331538e-05,
+      "loss": 0.3257,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9515088751193052e-05,
+      "loss": 0.3374,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9512263908189165e-05,
+      "loss": 0.3163,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950943106667903e-05,
+      "loss": 0.2761,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9506590229044678e-05,
+      "loss": 0.3274,
+      "step": 450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9503741397674857e-05,
+      "loss": 0.3644,
+      "step": 451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9500884574965044e-05,
+      "loss": 0.3835,
+      "step": 452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9498019763317436e-05,
+      "loss": 0.3391,
+      "step": 453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9495146965140943e-05,
+      "loss": 0.3159,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9492266182851194e-05,
+      "loss": 0.3217,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9489377418870536e-05,
+      "loss": 0.3277,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948648067562802e-05,
+      "loss": 0.3337,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9483575955559405e-05,
+      "loss": 0.2992,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9480663261107163e-05,
+      "loss": 0.4012,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9477742594720467e-05,
+      "loss": 0.3781,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9474813958855202e-05,
+      "loss": 0.3273,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9471877355973947e-05,
+      "loss": 0.309,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9468932788545976e-05,
+      "loss": 0.297,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946598025904727e-05,
+      "loss": 0.3305,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9463019769960503e-05,
+      "loss": 0.3335,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9460051323775033e-05,
+      "loss": 0.3548,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945707492298692e-05,
+      "loss": 0.3258,
+      "step": 467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.94540905700989e-05,
+      "loss": 0.3559,
+      "step": 468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.945109826762041e-05,
+      "loss": 0.3476,
+      "step": 469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944809801806757e-05,
+      "loss": 0.339,
+      "step": 470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944508982396317e-05,
+      "loss": 0.3305,
+      "step": 471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9442073687836687e-05,
+      "loss": 0.2842,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943904961222428e-05,
+      "loss": 0.3174,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943601759966878e-05,
+      "loss": 0.3917,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9432977652719695e-05,
+      "loss": 0.3321,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9429929773933195e-05,
+      "loss": 0.3158,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9426873965872135e-05,
+      "loss": 0.2941,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9423810231106027e-05,
+      "loss": 0.3435,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942073857221105e-05,
+      "loss": 0.3178,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9417658991770048e-05,
+      "loss": 0.3372,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9414571492372523e-05,
+      "loss": 0.33,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941147607661464e-05,
+      "loss": 0.3774,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9408372747099216e-05,
+      "loss": 0.3481,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940526150643573e-05,
+      "loss": 0.3801,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.94021423572403e-05,
+      "loss": 0.2817,
+      "step": 485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.93990153021357e-05,
+      "loss": 0.3085,
+      "step": 486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9395880343751364e-05,
+      "loss": 0.3176,
+      "step": 487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939273748472335e-05,
+      "loss": 0.3074,
+      "step": 488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938958672769438e-05,
+      "loss": 0.3732,
+      "step": 489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.93864280753138e-05,
+      "loss": 0.3005,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938326153023761e-05,
+      "loss": 0.3372,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938008709512843e-05,
+      "loss": 0.3255,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9376904772655528e-05,
+      "loss": 0.3455,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9373714565494804e-05,
+      "loss": 0.2941,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9370516476328775e-05,
+      "loss": 0.2811,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.93673105078466e-05,
+      "loss": 0.3633,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936409666274406e-05,
+      "loss": 0.3531,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936087494372355e-05,
+      "loss": 0.3817,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9357645353494095e-05,
+      "loss": 0.2918,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935440789477134e-05,
+      "loss": 0.3142,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9351162570277537e-05,
+      "loss": 0.311,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9347909382741564e-05,
+      "loss": 0.339,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.93446483348989e-05,
+      "loss": 0.334,
+      "step": 503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9341379429491644e-05,
+      "loss": 0.2963,
+      "step": 504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9338102669268486e-05,
+      "loss": 0.3198,
+      "step": 505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933481805698474e-05,
+      "loss": 0.3975,
+      "step": 506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9331525595402306e-05,
+      "loss": 0.2884,
+      "step": 507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93282252872897e-05,
+      "loss": 0.3089,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932491713542202e-05,
+      "loss": 0.3036,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932160114258097e-05,
+      "loss": 0.3129,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9318277311554844e-05,
+      "loss": 0.3211,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9314945645138524e-05,
+      "loss": 0.3714,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931160614613349e-05,
+      "loss": 0.3629,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9308258817347795e-05,
+      "loss": 0.3362,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9304903661596087e-05,
+      "loss": 0.3405,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.930154068169959e-05,
+      "loss": 0.323,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9298169880486103e-05,
+      "loss": 0.3003,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9294791260790012e-05,
+      "loss": 0.2885,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9291404825452273e-05,
+      "loss": 0.3269,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9288010577320408e-05,
+      "loss": 0.4042,
+      "step": 520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928460851924851e-05,
+      "loss": 0.3685,
+      "step": 521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928119865409725e-05,
+      "loss": 0.3356,
+      "step": 522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927778098473385e-05,
+      "loss": 0.3072,
+      "step": 523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9274355514032105e-05,
+      "loss": 0.3229,
+      "step": 524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9270922244872362e-05,
+      "loss": 0.3209,
+      "step": 525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9267481180141525e-05,
+      "loss": 0.3325,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9264032322733063e-05,
+      "loss": 0.3544,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9260575675546985e-05,
+      "loss": 0.3036,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9257111241489865e-05,
+      "loss": 0.3604,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9253639023474805e-05,
+      "loss": 0.3338,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.925015902442147e-05,
+      "loss": 0.3007,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924667124725606e-05,
+      "loss": 0.294,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9243175694911313e-05,
+      "loss": 0.3416,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.923967237032651e-05,
+      "loss": 0.3475,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9236161276447465e-05,
+      "loss": 0.3769,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9232642416226526e-05,
+      "loss": 0.3345,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.922911579262257e-05,
+      "loss": 0.3296,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9225581408601002e-05,
+      "loss": 0.3154,
+      "step": 538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.922203926713375e-05,
+      "loss": 0.3307,
+      "step": 539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921848937119928e-05,
+      "loss": 0.3017,
+      "step": 540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921493172378255e-05,
+      "loss": 0.2903,
+      "step": 541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9211366327875067e-05,
+      "loss": 0.3404,
+      "step": 542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9207793186474832e-05,
+      "loss": 0.3352,
+      "step": 543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9204212302586363e-05,
+      "loss": 0.3721,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9200623679220694e-05,
+      "loss": 0.3096,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.919702731939537e-05,
+      "loss": 0.3226,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9193423226134423e-05,
+      "loss": 0.3063,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918981140246841e-05,
+      "loss": 0.3157,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9186191851434372e-05,
+      "loss": 0.3034,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918256457607586e-05,
+      "loss": 0.3107,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9178929579442906e-05,
+      "loss": 0.3784,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917528686459205e-05,
+      "loss": 0.3705,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917163643458631e-05,
+      "loss": 0.3433,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.916797829249519e-05,
+      "loss": 0.3273,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9164312441394705e-05,
+      "loss": 0.2618,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9160638884367308e-05,
+      "loss": 0.2962,
+      "step": 556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9156957624501968e-05,
+      "loss": 0.3138,
+      "step": 557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915326866489412e-05,
+      "loss": 0.3405,
+      "step": 558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9149572008645662e-05,
+      "loss": 0.3402,
+      "step": 559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9145867658864983e-05,
+      "loss": 0.3251,
+      "step": 560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914215561866693e-05,
+      "loss": 0.3293,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9138435891172812e-05,
+      "loss": 0.299,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.913470847951042e-05,
+      "loss": 0.3002,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9130973386813987e-05,
+      "loss": 0.285,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9127230616224212e-05,
+      "loss": 0.2921,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9123480170888253e-05,
+      "loss": 0.3632,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.911972205395972e-05,
+      "loss": 0.3701,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9115956268598665e-05,
+      "loss": 0.3344,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9112182817971606e-05,
+      "loss": 0.2967,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9108401705251495e-05,
+      "loss": 0.3187,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9104612933617722e-05,
+      "loss": 0.3182,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9100816506256125e-05,
+      "loss": 0.3059,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9097012426358984e-05,
+      "loss": 0.3047,
+      "step": 573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9093200697124997e-05,
+      "loss": 0.2848,
+      "step": 574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9089381321759307e-05,
+      "loss": 0.3588,
+      "step": 575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9085554303473494e-05,
+      "loss": 0.349,
+      "step": 576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908171964548554e-05,
+      "loss": 0.3286,
+      "step": 577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9077877351019872e-05,
+      "loss": 0.2911,
+      "step": 578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907402742330733e-05,
+      "loss": 0.3172,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9070169865585175e-05,
+      "loss": 0.3541,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9066304681097084e-05,
+      "loss": 0.3472,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9062431873093143e-05,
+      "loss": 0.3642,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9058551444829852e-05,
+      "loss": 0.3306,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9054663399570118e-05,
+      "loss": 0.3308,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9050767740583254e-05,
+      "loss": 0.351,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904686447114497e-05,
+      "loss": 0.3085,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904295359453738e-05,
+      "loss": 0.3078,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9039035114048996e-05,
+      "loss": 0.3194,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9035109032974723e-05,
+      "loss": 0.3958,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903117535461585e-05,
+      "loss": 0.3304,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.902723408228006e-05,
+      "loss": 0.3615,
+      "step": 591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9023285219281428e-05,
+      "loss": 0.2836,
+      "step": 592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.90193287689404e-05,
+      "loss": 0.3019,
+      "step": 593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9015364734583803e-05,
+      "loss": 0.3447,
+      "step": 594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9011393119544853e-05,
+      "loss": 0.2948,
+      "step": 595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9007413927163124e-05,
+      "loss": 0.3096,
+      "step": 596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900342716078458e-05,
+      "loss": 0.3407,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8999432823761534e-05,
+      "loss": 0.347,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899543091945268e-05,
+      "loss": 0.3494,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8991421451223067e-05,
+      "loss": 0.3133,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8987404422444104e-05,
+      "loss": 0.2903,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8983379836493565e-05,
+      "loss": 0.3051,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8979347696755572e-05,
+      "loss": 0.3137,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.89753080066206e-05,
+      "loss": 0.33,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8971260769485468e-05,
+      "loss": 0.3024,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8967205988753357e-05,
+      "loss": 0.336,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8963143667833768e-05,
+      "loss": 0.3073,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8959073810142564e-05,
+      "loss": 0.3199,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.895499641910193e-05,
+      "loss": 0.3163,
+      "step": 609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8950911498140397e-05,
+      "loss": 0.3191,
+      "step": 610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8946819050692816e-05,
+      "loss": 0.3411,
+      "step": 611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8942719080200372e-05,
+      "loss": 0.3494,
+      "step": 612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8938611590110584e-05,
+      "loss": 0.3537,
+      "step": 613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.893449658387728e-05,
+      "loss": 0.3232,
+      "step": 614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8930374064960617e-05,
+      "loss": 0.2752,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8926244036827067e-05,
+      "loss": 0.3148,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8922106502949414e-05,
+      "loss": 0.3188,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8917961466806756e-05,
+      "loss": 0.314,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.89138089318845e-05,
+      "loss": 0.3046,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8909648901674352e-05,
+      "loss": 0.3146,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8905481379674335e-05,
+      "loss": 0.3525,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8901306369388755e-05,
+      "loss": 0.3264,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8897123874328223e-05,
+      "loss": 0.3157,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889293389800964e-05,
+      "loss": 0.2891,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8888736443956205e-05,
+      "loss": 0.3051,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.88845315156974e-05,
+      "loss": 0.298,
+      "step": 626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8880319116768982e-05,
+      "loss": 0.3503,
+      "step": 627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8876099250713012e-05,
+      "loss": 0.3322,
+      "step": 628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8871871921077808e-05,
+      "loss": 0.3305,
+      "step": 629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8867637131417978e-05,
+      "loss": 0.3576,
+      "step": 630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8863394885294398e-05,
+      "loss": 0.3051,
+      "step": 631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8859145186274214e-05,
+      "loss": 0.3347,
+      "step": 632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8854888037930832e-05,
+      "loss": 0.3135,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8850623443843935e-05,
+      "loss": 0.2977,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.884635140759946e-05,
+      "loss": 0.3678,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.88420719327896e-05,
+      "loss": 0.3034,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8837785023012808e-05,
+      "loss": 0.3506,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8833490681873785e-05,
+      "loss": 0.2734,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8829188912983478e-05,
+      "loss": 0.3341,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8824879719959085e-05,
+      "loss": 0.2918,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8820563106424048e-05,
+      "loss": 0.3341,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8816239076008042e-05,
+      "loss": 0.3202,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8811907632346983e-05,
+      "loss": 0.3262,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.880756877908302e-05,
+      "loss": 0.3658,
+      "step": 644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8803222519864533e-05,
+      "loss": 0.3491,
+      "step": 645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8798868858346126e-05,
+      "loss": 0.3019,
+      "step": 646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879450779818863e-05,
+      "loss": 0.2831,
+      "step": 647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8790139343059103e-05,
+      "loss": 0.2903,
+      "step": 648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8785763496630807e-05,
+      "loss": 0.3161,
+      "step": 649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.878138026258323e-05,
+      "loss": 0.3518,
+      "step": 650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877698964460207e-05,
+      "loss": 0.3523,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877259164637924e-05,
+      "loss": 0.28,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876818627161284e-05,
+      "loss": 0.3066,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876377352400719e-05,
+      "loss": 0.3094,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8759353407272804e-05,
+      "loss": 0.3031,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.875492592512639e-05,
+      "loss": 0.3011,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8750491081290858e-05,
+      "loss": 0.2867,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8746048879495296e-05,
+      "loss": 0.3258,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874159932347499e-05,
+      "loss": 0.3753,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8737142416971398e-05,
+      "loss": 0.3102,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873267816373217e-05,
+      "loss": 0.2935,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8728206567511134e-05,
+      "loss": 0.2809,
+      "step": 662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872372763206828e-05,
+      "loss": 0.3073,
+      "step": 663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8719241361169777e-05,
+      "loss": 0.3025,
+      "step": 664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8714747758587967e-05,
+      "loss": 0.3675,
+      "step": 665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.871024682810135e-05,
+      "loss": 0.2868,
+      "step": 666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870573857349459e-05,
+      "loss": 0.3649,
+      "step": 667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8701222998558505e-05,
+      "loss": 0.3206,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.869670010709008e-05,
+      "loss": 0.3165,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.869216990289244e-05,
+      "loss": 0.2859,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8687632389774867e-05,
+      "loss": 0.2789,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8683087571552786e-05,
+      "loss": 0.3128,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.867853545204776e-05,
+      "loss": 0.3142,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.86739760350875e-05,
+      "loss": 0.3562,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8669409324505843e-05,
+      "loss": 0.2889,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8664835324142772e-05,
+      "loss": 0.307,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.3066,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8655665469462928e-05,
+      "loss": 0.3034,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.865106962285674e-05,
+      "loss": 0.2966,
+      "step": 679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.86464665018903e-05,
+      "loss": 0.3028,
+      "step": 680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.864185611043421e-05,
+      "loss": 0.347,
+      "step": 681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863723845236516e-05,
+      "loss": 0.3524,
+      "step": 682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863261353156598e-05,
+      "loss": 0.3625,
+      "step": 683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8627981351925584e-05,
+      "loss": 0.3157,
+      "step": 684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8623341917339003e-05,
+      "loss": 0.3104,
+      "step": 685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.861869523170736e-05,
+      "loss": 0.3094,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8614041298937877e-05,
+      "loss": 0.3291,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8609380122943875e-05,
+      "loss": 0.34,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8604711707644763e-05,
+      "loss": 0.3052,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860003605696604e-05,
+      "loss": 0.3875,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8595353174839275e-05,
+      "loss": 0.3329,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8590663065202137e-05,
+      "loss": 0.3159,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8585965731998357e-05,
+      "loss": 0.2892,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8581261179177753e-05,
+      "loss": 0.3064,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8576549410696205e-05,
+      "loss": 0.3126,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8571830430515655e-05,
+      "loss": 0.3338,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.856710424260413e-05,
+      "loss": 0.3707,
+      "step": 697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8562370850935697e-05,
+      "loss": 0.3081,
+      "step": 698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8557630259490483e-05,
+      "loss": 0.2896,
+      "step": 699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8552882472254686e-05,
+      "loss": 0.3167,
+      "step": 700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.854812749322053e-05,
+      "loss": 0.3185,
+      "step": 701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8543365326386305e-05,
+      "loss": 0.3234,
+      "step": 702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8538595975756336e-05,
+      "loss": 0.2827,
+      "step": 703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8533819445340993e-05,
+      "loss": 0.3229,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8529035739156685e-05,
+      "loss": 0.3828,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8524244861225843e-05,
+      "loss": 0.3229,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8519446815576942e-05,
+      "loss": 0.3177,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.851464160624448e-05,
+      "loss": 0.2698,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8509829237268975e-05,
+      "loss": 0.2883,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8505009712696965e-05,
+      "loss": 0.3333,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.850018303658101e-05,
+      "loss": 0.3118,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8495349212979685e-05,
+      "loss": 0.3194,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.849050824595757e-05,
+      "loss": 0.3224,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8485660139585245e-05,
+      "loss": 0.3531,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.848080489793931e-05,
+      "loss": 0.3033,
+      "step": 715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.847594252510235e-05,
+      "loss": 0.3003,
+      "step": 716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8471073025162956e-05,
+      "loss": 0.2674,
+      "step": 717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8466196402215708e-05,
+      "loss": 0.302,
+      "step": 718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8461312660361172e-05,
+      "loss": 0.359,
+      "step": 719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8456421803705907e-05,
+      "loss": 0.3364,
+      "step": 720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8451523836362447e-05,
+      "loss": 0.3405,
+      "step": 721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8446618762449314e-05,
+      "loss": 0.2819,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8441706586090993e-05,
+      "loss": 0.3325,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8436787311417962e-05,
+      "loss": 0.2953,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8431860942566645e-05,
+      "loss": 0.2978,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8426927483679444e-05,
+      "loss": 0.2884,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8421986938904715e-05,
+      "loss": 0.3041,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8417039312396783e-05,
+      "loss": 0.3346,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.841208460831592e-05,
+      "loss": 0.3667,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840712283082835e-05,
+      "loss": 0.3157,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840215398410625e-05,
+      "loss": 0.2711,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8397178072327727e-05,
+      "loss": 0.3143,
+      "step": 732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8392195099676845e-05,
+      "loss": 0.2985,
+      "step": 733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8387205070343602e-05,
+      "loss": 0.3241,
+      "step": 734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8382207988523927e-05,
+      "loss": 0.3054,
+      "step": 735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.837720385841967e-05,
+      "loss": 0.3583,
+      "step": 736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8372192684238627e-05,
+      "loss": 0.3311,
+      "step": 737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83671744701945e-05,
+      "loss": 0.3204,
+      "step": 738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8362149220506914e-05,
+      "loss": 0.2847,
+      "step": 739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8357116939401424e-05,
+      "loss": 0.2795,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8352077631109474e-05,
+      "loss": 0.2923,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8347031299868444e-05,
+      "loss": 0.3197,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.834197794992159e-05,
+      "loss": 0.3703,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8336917585518095e-05,
+      "loss": 0.3194,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8331850210913026e-05,
+      "loss": 0.3049,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8326775830367347e-05,
+      "loss": 0.3213,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832169444814792e-05,
+      "loss": 0.3197,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831660606852748e-05,
+      "loss": 0.3002,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831151069578466e-05,
+      "loss": 0.2676,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8306408334203968e-05,
+      "loss": 0.371,
+      "step": 750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.830129898807579e-05,
+      "loss": 0.3284,
+      "step": 751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8296182661696382e-05,
+      "loss": 0.3407,
+      "step": 752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829105935936787e-05,
+      "loss": 0.3002,
+      "step": 753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8285929085398248e-05,
+      "loss": 0.3226,
+      "step": 754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8280791844101375e-05,
+      "loss": 0.3099,
+      "step": 755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8275647639796955e-05,
+      "loss": 0.3157,
+      "step": 756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8270496476810563e-05,
+      "loss": 0.2976,
+      "step": 757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826533835947362e-05,
+      "loss": 0.3053,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8260173292123393e-05,
+      "loss": 0.2983,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8255001279102994e-05,
+      "loss": 0.3373,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.824982232476137e-05,
+      "loss": 0.2892,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8244636433453315e-05,
+      "loss": 0.3304,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8239443609539445e-05,
+      "loss": 0.2876,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823424385738622e-05,
+      "loss": 0.3018,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.822903718136591e-05,
+      "loss": 0.3605,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8223823585856616e-05,
+      "loss": 0.3223,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.821860307524225e-05,
+      "loss": 0.354,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.821337565391255e-05,
+      "loss": 0.2894,
+      "step": 768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.820814132626306e-05,
+      "loss": 0.3219,
+      "step": 769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8202900096695125e-05,
+      "loss": 0.3122,
+      "step": 770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8197651969615897e-05,
+      "loss": 0.3191,
+      "step": 771
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8192396949438337e-05,
+      "loss": 0.2997,
+      "step": 772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818713504058119e-05,
+      "loss": 0.3183,
+      "step": 773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8181866247468997e-05,
+      "loss": 0.3915,
+      "step": 774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8176590574532085e-05,
+      "loss": 0.3022,
+      "step": 775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.817130802620658e-05,
+      "loss": 0.3011,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816601860693437e-05,
+      "loss": 0.2986,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8160722321163137e-05,
+      "loss": 0.3142,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8155419173346324e-05,
+      "loss": 0.3133,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.815010916794315e-05,
+      "loss": 0.2974,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.81447923094186e-05,
+      "loss": 0.3079,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813946860224342e-05,
+      "loss": 0.3355,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8134138050894117e-05,
+      "loss": 0.3284,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8128800659852952e-05,
+      "loss": 0.31,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.812345643360794e-05,
+      "loss": 0.2965,
+      "step": 785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8118105376652833e-05,
+      "loss": 0.2724,
+      "step": 786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811274749348714e-05,
+      "loss": 0.2973,
+      "step": 787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81073827886161e-05,
+      "loss": 0.3148,
+      "step": 788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81020112665507e-05,
+      "loss": 0.3747,
+      "step": 789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8096632931807647e-05,
+      "loss": 0.3096,
+      "step": 790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.809124778890938e-05,
+      "loss": 0.3019,
+      "step": 791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.808585584238406e-05,
+      "loss": 0.3027,
+      "step": 792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.808045709676558e-05,
+      "loss": 0.3004,
+      "step": 793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8075051556593544e-05,
+      "loss": 0.273,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8069639226413263e-05,
+      "loss": 0.2912,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8064220110775762e-05,
+      "loss": 0.2931,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8058794214237774e-05,
+      "loss": 0.3368,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8053361541361734e-05,
+      "loss": 0.3286,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.804792209671577e-05,
+      "loss": 0.3047,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.804247588487371e-05,
+      "loss": 0.2706,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8037022910415068e-05,
+      "loss": 0.3205,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8031563177925048e-05,
+      "loss": 0.3012,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8026096691994533e-05,
+      "loss": 0.3249,
+      "step": 803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8020623457220082e-05,
+      "loss": 0.2936,
+      "step": 804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8015143478203937e-05,
+      "loss": 0.2958,
+      "step": 805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8009656759554008e-05,
+      "loss": 0.3513,
+      "step": 806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800416330588387e-05,
+      "loss": 0.2952,
+      "step": 807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7998663121812763e-05,
+      "loss": 0.3136,
+      "step": 808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7993156211965582e-05,
+      "loss": 0.2857,
+      "step": 809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7987642580972884e-05,
+      "loss": 0.2733,
+      "step": 810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7982122233470875e-05,
+      "loss": 0.3568,
+      "step": 811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797659517410141e-05,
+      "loss": 0.327,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7971061407511984e-05,
+      "loss": 0.3328,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.796552093835573e-05,
+      "loss": 0.2809,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7959973771291424e-05,
+      "loss": 0.2901,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7954419910983478e-05,
+      "loss": 0.292,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7948859362101914e-05,
+      "loss": 0.3013,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7943292129322395e-05,
+      "loss": 0.2787,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79377182173262e-05,
+      "loss": 0.3079,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7932137630800217e-05,
+      "loss": 0.3641,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7926550374436954e-05,
+      "loss": 0.3071,
+      "step": 821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7920956452934527e-05,
+      "loss": 0.3197,
+      "step": 822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7915355870996656e-05,
+      "loss": 0.2502,
+      "step": 823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.790974863333266e-05,
+      "loss": 0.2835,
+      "step": 824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7904134744657452e-05,
+      "loss": 0.3099,
+      "step": 825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7898514209691544e-05,
+      "loss": 0.2921,
+      "step": 826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7892887033161032e-05,
+      "loss": 0.3446,
+      "step": 827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7887253219797595e-05,
+      "loss": 0.3145,
+      "step": 828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.78816127743385e-05,
+      "loss": 0.36,
+      "step": 829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7875965701526586e-05,
+      "loss": 0.2783,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.787031200611026e-05,
+      "loss": 0.2996,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.786465169284351e-05,
+      "loss": 0.2865,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7858984766485873e-05,
+      "loss": 0.2904,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7853311231802462e-05,
+      "loss": 0.3292,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7847631093563937e-05,
+      "loss": 0.3569,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7841944356546515e-05,
+      "loss": 0.329,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7836251025531955e-05,
+      "loss": 0.2642,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7830551105307572e-05,
+      "loss": 0.2982,
+      "step": 838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7824844600666208e-05,
+      "loss": 0.2959,
+      "step": 839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.781913151640626e-05,
+      "loss": 0.2993,
+      "step": 840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7813411857331636e-05,
+      "loss": 0.2734,
+      "step": 841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7807685628251782e-05,
+      "loss": 0.3111,
+      "step": 842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7801952833981683e-05,
+      "loss": 0.3424,
+      "step": 843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7796213479341816e-05,
+      "loss": 0.3357,
+      "step": 844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7790467569158197e-05,
+      "loss": 0.3123,
+      "step": 845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7784715108262344e-05,
+      "loss": 0.3119,
+      "step": 846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777895610149129e-05,
+      "loss": 0.2919,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7773190553687562e-05,
+      "loss": 0.3244,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7767418469699195e-05,
+      "loss": 0.2816,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.776163985437972e-05,
+      "loss": 0.3037,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.775585471258816e-05,
+      "loss": 0.2934,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7750063049189016e-05,
+      "loss": 0.3538,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7744264869052287e-05,
+      "loss": 0.318,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7738460177053447e-05,
+      "loss": 0.2982,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.773264897807344e-05,
+      "loss": 0.2837,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7726831276998682e-05,
+      "loss": 0.2842,
+      "step": 856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7721007078721068e-05,
+      "loss": 0.3034,
+      "step": 857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.771517638813794e-05,
+      "loss": 0.3507,
+      "step": 858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7709339210152117e-05,
+      "loss": 0.3506,
+      "step": 859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7703495549671854e-05,
+      "loss": 0.3066,
+      "step": 860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.769764541161087e-05,
+      "loss": 0.2854,
+      "step": 861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7691788800888326e-05,
+      "loss": 0.2993,
+      "step": 862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7685925722428832e-05,
+      "loss": 0.3015,
+      "step": 863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7680056181162424e-05,
+      "loss": 0.2824,
+      "step": 864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7674180182024578e-05,
+      "loss": 0.3074,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.766829772995621e-05,
+      "loss": 0.349,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7662408829903645e-05,
+      "loss": 0.3395,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.765651348681864e-05,
+      "loss": 0.3123,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7650611705658375e-05,
+      "loss": 0.2751,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7644703491385428e-05,
+      "loss": 0.2942,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7638788848967796e-05,
+      "loss": 0.3107,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7632867783378887e-05,
+      "loss": 0.3061,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7626940299597497e-05,
+      "loss": 0.3515,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7621006402607827e-05,
+      "loss": 0.2637,
+      "step": 874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.761506609739947e-05,
+      "loss": 0.3653,
+      "step": 875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.76091193889674e-05,
+      "loss": 0.2748,
+      "step": 876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7603166282311987e-05,
+      "loss": 0.2973,
+      "step": 877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7597206782438978e-05,
+      "loss": 0.2897,
+      "step": 878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7591240894359485e-05,
+      "loss": 0.2831,
+      "step": 879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7585268623090004e-05,
+      "loss": 0.3348,
+      "step": 880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7579289973652395e-05,
+      "loss": 0.3175,
+      "step": 881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7573304951073876e-05,
+      "loss": 0.3723,
+      "step": 882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.756731356038704e-05,
+      "loss": 0.2415,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7561315806629803e-05,
+      "loss": 0.324,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7555311694845468e-05,
+      "loss": 0.3048,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7549301230082664e-05,
+      "loss": 0.2955,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754328441739536e-05,
+      "loss": 0.3018,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7537261261842874e-05,
+      "loss": 0.3052,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.753123176848985e-05,
+      "loss": 0.3519,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.752519594240626e-05,
+      "loss": 0.3457,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.751915378866741e-05,
+      "loss": 0.3039,
+      "step": 891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.751310531235392e-05,
+      "loss": 0.2724,
+      "step": 892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.750705051855172e-05,
+      "loss": 0.2951,
+      "step": 893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.750098941235206e-05,
+      "loss": 0.2895,
+      "step": 894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.749492199885151e-05,
+      "loss": 0.3188,
+      "step": 895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.748884828315192e-05,
+      "loss": 0.3307,
+      "step": 896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7482768270360446e-05,
+      "loss": 0.2449,
+      "step": 897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.747668196558955e-05,
+      "loss": 0.3381,
+      "step": 898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7470589373956977e-05,
+      "loss": 0.2907,
+      "step": 899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7464490500585756e-05,
+      "loss": 0.3068,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7458385350604207e-05,
+      "loss": 0.2925,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7452273929145915e-05,
+      "loss": 0.2931,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7446156241349748e-05,
+      "loss": 0.29,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7440032292359842e-05,
+      "loss": 0.3197,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.743390208732559e-05,
+      "loss": 0.337,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7427765631401663e-05,
+      "loss": 0.2929,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7421622929747967e-05,
+      "loss": 0.28,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.741547398752967e-05,
+      "loss": 0.3096,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7409318809917192e-05,
+      "loss": 0.289,
+      "step": 909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7403157402086193e-05,
+      "loss": 0.309,
+      "step": 910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.739698976921756e-05,
+      "loss": 0.2653,
+      "step": 911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7390815916497432e-05,
+      "loss": 0.3532,
+      "step": 912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7384635849117172e-05,
+      "loss": 0.3378,
+      "step": 913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7378449572273362e-05,
+      "loss": 0.3056,
+      "step": 914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.737225709116781e-05,
+      "loss": 0.3,
+      "step": 915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7366058411007543e-05,
+      "loss": 0.2838,
+      "step": 916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7359853537004805e-05,
+      "loss": 0.2942,
+      "step": 917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7353642474377032e-05,
+      "loss": 0.2952,
+      "step": 918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.734742522834688e-05,
+      "loss": 0.3227,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7341201804142193e-05,
+      "loss": 0.3024,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733497220699602e-05,
+      "loss": 0.3391,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7328736442146595e-05,
+      "loss": 0.278,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.732249451483734e-05,
+      "loss": 0.3045,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7316246430316855e-05,
+      "loss": 0.2943,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.730999219383892e-05,
+      "loss": 0.2793,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.730373181066249e-05,
+      "loss": 0.2995,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7297465286051686e-05,
+      "loss": 0.321,
+      "step": 927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.729119262527579e-05,
+      "loss": 0.3593,
+      "step": 928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7284913833609257e-05,
+      "loss": 0.284,
+      "step": 929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7278628916331678e-05,
+      "loss": 0.31,
+      "step": 930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.727233787872781e-05,
+      "loss": 0.2797,
+      "step": 931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.726604072608755e-05,
+      "loss": 0.3064,
+      "step": 932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7259737463705932e-05,
+      "loss": 0.2807,
+      "step": 933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.725342809688314e-05,
+      "loss": 0.2579,
+      "step": 934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.724711263092448e-05,
+      "loss": 0.3577,
+      "step": 935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7240791071140398e-05,
+      "loss": 0.3045,
+      "step": 936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7234463422846445e-05,
+      "loss": 0.3297,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7228129691363312e-05,
+      "loss": 0.2601,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7221789882016792e-05,
+      "loss": 0.3013,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7215444000137797e-05,
+      "loss": 0.2843,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7209092051062337e-05,
+      "loss": 0.3101,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.720273404013153e-05,
+      "loss": 0.3392,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7196369972691595e-05,
+      "loss": 0.2926,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7189999854093824e-05,
+      "loss": 0.3352,
+      "step": 944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.718362368969462e-05,
+      "loss": 0.274,
+      "step": 945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7177241484855467e-05,
+      "loss": 0.2938,
+      "step": 946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7170853244942907e-05,
+      "loss": 0.3053,
+      "step": 947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7164458975328586e-05,
+      "loss": 0.2611,
+      "step": 948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7158058681389193e-05,
+      "loss": 0.3123,
+      "step": 949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7151652368506508e-05,
+      "loss": 0.351,
+      "step": 950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7145240042067345e-05,
+      "loss": 0.3431,
+      "step": 951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7138821707463598e-05,
+      "loss": 0.2878,
+      "step": 952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7132397370092206e-05,
+      "loss": 0.285,
+      "step": 953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.712596703535515e-05,
+      "loss": 0.288,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7119530708659456e-05,
+      "loss": 0.3177,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7113088395417193e-05,
+      "loss": 0.2909,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7106640101045456e-05,
+      "loss": 0.2639,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.710018583096638e-05,
+      "loss": 0.3702,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7093725590607116e-05,
+      "loss": 0.3287,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.708725938539984e-05,
+      "loss": 0.3,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7080787220781742e-05,
+      "loss": 0.3024,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7074309102195018e-05,
+      "loss": 0.2554,
+      "step": 962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7067825035086874e-05,
+      "loss": 0.2936,
+      "step": 963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.706133502490953e-05,
+      "loss": 0.2811,
+      "step": 964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7054839077120176e-05,
+      "loss": 0.3238,
+      "step": 965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.704833719718103e-05,
+      "loss": 0.3514,
+      "step": 966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7041829390559265e-05,
+      "loss": 0.3038,
+      "step": 967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7035315662727055e-05,
+      "loss": 0.3157,
+      "step": 968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7028796019161557e-05,
+      "loss": 0.2902,
+      "step": 969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7022270465344884e-05,
+      "loss": 0.3082,
+      "step": 970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7015739006764134e-05,
+      "loss": 0.2655,
+      "step": 971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7009201648911367e-05,
+      "loss": 0.3133,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7002658397283603e-05,
+      "loss": 0.3387,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6996109257382812e-05,
+      "loss": 0.3341,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6989554234715928e-05,
+      "loss": 0.3089,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6982993334794818e-05,
+      "loss": 0.2761,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.69764265631363e-05,
+      "loss": 0.281,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.696985392526212e-05,
+      "loss": 0.2987,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6963275426698972e-05,
+      "loss": 0.283,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6956691072978463e-05,
+      "loss": 0.3042,
+      "step": 980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.695010086963713e-05,
+      "loss": 0.2969,
+      "step": 981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6943504822216433e-05,
+      "loss": 0.3243,
+      "step": 982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6936902936262734e-05,
+      "loss": 0.3204,
+      "step": 983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6930295217327315e-05,
+      "loss": 0.2532,
+      "step": 984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6923681670966358e-05,
+      "loss": 0.2872,
+      "step": 985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691706230274095e-05,
+      "loss": 0.3353,
+      "step": 986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6910437118217066e-05,
+      "loss": 0.2784,
+      "step": 987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6903806122965574e-05,
+      "loss": 0.356,
+      "step": 988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6897169322562232e-05,
+      "loss": 0.2743,
+      "step": 989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6890526722587678e-05,
+      "loss": 0.3809,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.688387832862742e-05,
+      "loss": 0.2688,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.687722414627185e-05,
+      "loss": 0.2985,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.687056418111622e-05,
+      "loss": 0.2973,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6863898438760638e-05,
+      "loss": 0.2558,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6857226924810084e-05,
+      "loss": 0.3072,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685054964487438e-05,
+      "loss": 0.2949,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6843866604568204e-05,
+      "loss": 0.3708,
+      "step": 997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.683717780951107e-05,
+      "loss": 0.2819,
+      "step": 998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6830483265327342e-05,
+      "loss": 0.2595,
+      "step": 999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6823782977646204e-05,
+      "loss": 0.2881,
+      "step": 1000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6817076952101682e-05,
+      "loss": 0.2927,
+      "step": 1001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6810365194332616e-05,
+      "loss": 0.3235,
+      "step": 1002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6803647709982676e-05,
+      "loss": 0.3065,
+      "step": 1003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.679692450470034e-05,
+      "loss": 0.3144,
+      "step": 1004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.67901955841389e-05,
+      "loss": 0.33,
+      "step": 1005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6783460953956457e-05,
+      "loss": 0.3117,
+      "step": 1006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6776720619815908e-05,
+      "loss": 0.3028,
+      "step": 1007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6769974587384944e-05,
+      "loss": 0.2553,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6763222862336054e-05,
+      "loss": 0.3147,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6756465450346505e-05,
+      "loss": 0.327,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6749702357098356e-05,
+      "loss": 0.3416,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.674293358827844e-05,
+      "loss": 0.2864,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673615914957836e-05,
+      "loss": 0.2699,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6729379046694486e-05,
+      "loss": 0.3295,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6722593285327953e-05,
+      "loss": 0.2826,
+      "step": 1015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6715801871184655e-05,
+      "loss": 0.2877,
+      "step": 1016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6709004809975235e-05,
+      "loss": 0.2902,
+      "step": 1017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.670220210741509e-05,
+      "loss": 0.2985,
+      "step": 1018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6695393769224355e-05,
+      "loss": 0.3286,
+      "step": 1019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.668857980112791e-05,
+      "loss": 0.3099,
+      "step": 1020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.668176020885536e-05,
+      "loss": 0.3313,
+      "step": 1021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6674934998141048e-05,
+      "loss": 0.2626,
+      "step": 1022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.666810417472404e-05,
+      "loss": 0.2821,
+      "step": 1023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6661267744348114e-05,
+      "loss": 0.286,
+      "step": 1024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6654425712761772e-05,
+      "loss": 0.2794,
+      "step": 1025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6647578085718223e-05,
+      "loss": 0.3321,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6640724868975377e-05,
+      "loss": 0.304,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6633866068295847e-05,
+      "loss": 0.3155,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662700168944694e-05,
+      "loss": 0.3213,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6620131738200664e-05,
+      "loss": 0.3211,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6613256220333692e-05,
+      "loss": 0.3037,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6606375141627394e-05,
+      "loss": 0.2848,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6599488507867816e-05,
+      "loss": 0.2718,
+      "step": 1033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6592596324845663e-05,
+      "loss": 0.3228,
+      "step": 1034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6585698598356316e-05,
+      "loss": 0.301,
+      "step": 1035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6578795334199818e-05,
+      "loss": 0.3364,
+      "step": 1036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.657188653818086e-05,
+      "loss": 0.2768,
+      "step": 1037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656497221610879e-05,
+      "loss": 0.3285,
+      "step": 1038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.655805237379761e-05,
+      "loss": 0.2684,
+      "step": 1039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6551127017065944e-05,
+      "loss": 0.2798,
+      "step": 1040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6544196151737075e-05,
+      "loss": 0.2979,
+      "step": 1041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.65372597836389e-05,
+      "loss": 0.3367,
+      "step": 1042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6530317918603958e-05,
+      "loss": 0.3571,
+      "step": 1043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.65233705624694e-05,
+      "loss": 0.2792,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6516417721076996e-05,
+      "loss": 0.2937,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.650945940027313e-05,
+      "loss": 0.2737,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6502495605908803e-05,
+      "loss": 0.2811,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.649552634383959e-05,
+      "loss": 0.2918,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.64885516199257e-05,
+      "loss": 0.2919,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6481571440031907e-05,
+      "loss": 0.3159,
+      "step": 1050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6474585810027584e-05,
+      "loss": 0.3142,
+      "step": 1051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.646759473578669e-05,
+      "loss": 0.3296,
+      "step": 1052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.646059822318775e-05,
+      "loss": 0.3098,
+      "step": 1053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6453596278113877e-05,
+      "loss": 0.2427,
+      "step": 1054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644658890645274e-05,
+      "loss": 0.2642,
+      "step": 1055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6439576114096573e-05,
+      "loss": 0.2977,
+      "step": 1056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6432557906942176e-05,
+      "loss": 0.3247,
+      "step": 1057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.642553429089089e-05,
+      "loss": 0.3157,
+      "step": 1058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.641850527184862e-05,
+      "loss": 0.3072,
+      "step": 1059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6411470855725794e-05,
+      "loss": 0.3123,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6404431048437396e-05,
+      "loss": 0.2763,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6397385855902933e-05,
+      "loss": 0.2875,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.639033528404644e-05,
+      "loss": 0.2661,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.638327933879649e-05,
+      "loss": 0.275,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.637621802608615e-05,
+      "loss": 0.3158,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6369151351853022e-05,
+      "loss": 0.3187,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6362079322039206e-05,
+      "loss": 0.3148,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6355001942591302e-05,
+      "loss": 0.2565,
+      "step": 1068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6347919219460418e-05,
+      "loss": 0.3171,
+      "step": 1069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634083115860215e-05,
+      "loss": 0.2924,
+      "step": 1070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.633373776597658e-05,
+      "loss": 0.281,
+      "step": 1071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.632663904754828e-05,
+      "loss": 0.2896,
+      "step": 1072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.631953500928629e-05,
+      "loss": 0.3324,
+      "step": 1073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6312425657164137e-05,
+      "loss": 0.3409,
+      "step": 1074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6305310997159805e-05,
+      "loss": 0.301,
+      "step": 1075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6298191035255745e-05,
+      "loss": 0.3102,
+      "step": 1076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6291065777438865e-05,
+      "loss": 0.2995,
+      "step": 1077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628393522970053e-05,
+      "loss": 0.2641,
+      "step": 1078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.627679939803655e-05,
+      "loss": 0.2684,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.626965828844718e-05,
+      "loss": 0.3138,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6262511906937105e-05,
+      "loss": 0.3382,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6255360259515453e-05,
+      "loss": 0.2929,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.624820335219578e-05,
+      "loss": 0.3161,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6241041190996058e-05,
+      "loss": 0.2849,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.623387378193868e-05,
+      "loss": 0.2785,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.622670113105045e-05,
+      "loss": 0.2858,
+      "step": 1086
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.621952324436258e-05,
+      "loss": 0.2759,
+      "step": 1087
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.621234012791069e-05,
+      "loss": 0.3321,
+      "step": 1088
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6205151787734798e-05,
+      "loss": 0.3508,
+      "step": 1089
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.61979582298793e-05,
+      "loss": 0.2955,
+      "step": 1090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.619075946039299e-05,
+      "loss": 0.3079,
+      "step": 1091
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.618355548532905e-05,
+      "loss": 0.254,
+      "step": 1092
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6176346310745033e-05,
+      "loss": 0.2956,
+      "step": 1093
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6169131942702854e-05,
+      "loss": 0.2864,
+      "step": 1094
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6161912387268818e-05,
+      "loss": 0.2583,
+      "step": 1095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6154687650513566e-05,
+      "loss": 0.3449,
+      "step": 1096
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.614745773851212e-05,
+      "loss": 0.296,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.614022265734383e-05,
+      "loss": 0.3548,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6132982413092412e-05,
+      "loss": 0.2597,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6125737011845917e-05,
+      "loss": 0.2823,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6118486459696724e-05,
+      "loss": 0.2644,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6111230762741563e-05,
+      "loss": 0.2823,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6103969927081463e-05,
+      "loss": 0.3248,
+      "step": 1103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6096703958821804e-05,
+      "loss": 0.3143,
+      "step": 1104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608943286407225e-05,
+      "loss": 0.2749,
+      "step": 1105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6082156648946808e-05,
+      "loss": 0.2915,
+      "step": 1106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607487531956376e-05,
+      "loss": 0.3025,
+      "step": 1107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6067588882045714e-05,
+      "loss": 0.2751,
+      "step": 1108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6060297342519562e-05,
+      "loss": 0.2854,
+      "step": 1109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.605300070711648e-05,
+      "loss": 0.2602,
+      "step": 1110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6045698981971932e-05,
+      "loss": 0.3232,
+      "step": 1111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6038392173225678e-05,
+      "loss": 0.3145,
+      "step": 1112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6031080287021734e-05,
+      "loss": 0.3283,
+      "step": 1113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6023763329508386e-05,
+      "loss": 0.2924,
+      "step": 1114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6016441306838194e-05,
+      "loss": 0.2791,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6009114225167972e-05,
+      "loss": 0.2893,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6001782090658787e-05,
+      "loss": 0.3053,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.599444490947596e-05,
+      "loss": 0.333,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5987102687789046e-05,
+      "loss": 0.2624,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.597975543177185e-05,
+      "loss": 0.3419,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5972403147602397e-05,
+      "loss": 0.3001,
+      "step": 1121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5965045841462954e-05,
+      "loss": 0.2973,
+      "step": 1122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.595768351954e-05,
+      "loss": 0.2654,
+      "step": 1123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5950316188024237e-05,
+      "loss": 0.276,
+      "step": 1124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.594294385311058e-05,
+      "loss": 0.3045,
+      "step": 1125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5935566520998145e-05,
+      "loss": 0.3213,
+      "step": 1126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5928184197890258e-05,
+      "loss": 0.3408,
+      "step": 1127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.592079688999444e-05,
+      "loss": 0.3268,
+      "step": 1128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591340460352239e-05,
+      "loss": 0.2634,
+      "step": 1129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5906007344690018e-05,
+      "loss": 0.2632,
+      "step": 1130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.589860511971739e-05,
+      "loss": 0.2714,
+      "step": 1131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5891197934828773e-05,
+      "loss": 0.318,
+      "step": 1132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5883785796252578e-05,
+      "loss": 0.2717,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5876368710221404e-05,
+      "loss": 0.3306,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5868946682971995e-05,
+      "loss": 0.3013,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5861519720745257e-05,
+      "loss": 0.3374,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5854087829786247e-05,
+      "loss": 0.2782,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.584665101634416e-05,
+      "loss": 0.2781,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.583920928667233e-05,
+      "loss": 0.2859,
+      "step": 1139
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.583176264702824e-05,
+      "loss": 0.2983,
+      "step": 1140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5824311103673486e-05,
+      "loss": 0.2988,
+      "step": 1141
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5816854662873783e-05,
+      "loss": 0.2629,
+      "step": 1142
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5809393330898983e-05,
+      "loss": 0.3234,
+      "step": 1143
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5801927114023038e-05,
+      "loss": 0.3276,
+      "step": 1144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5794456018524006e-05,
+      "loss": 0.2905,
+      "step": 1145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5786980050684055e-05,
+      "loss": 0.2825,
+      "step": 1146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.577949921678944e-05,
+      "loss": 0.2646,
+      "step": 1147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5772013523130526e-05,
+      "loss": 0.2666,
+      "step": 1148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.576452297600174e-05,
+      "loss": 0.3367,
+      "step": 1149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.575702758170161e-05,
+      "loss": 0.2884,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.574952734653273e-05,
+      "loss": 0.2914,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5742022276801768e-05,
+      "loss": 0.2707,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.573451237881945e-05,
+      "loss": 0.2979,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5726997658900572e-05,
+      "loss": 0.2787,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5719478123363983e-05,
+      "loss": 0.2824,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5711953778532576e-05,
+      "loss": 0.2654,
+      "step": 1156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.570442463073329e-05,
+      "loss": 0.3022,
+      "step": 1157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5696890686297103e-05,
+      "loss": 0.3526,
+      "step": 1158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5689351951559027e-05,
+      "loss": 0.3343,
+      "step": 1159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5681808432858107e-05,
+      "loss": 0.2814,
+      "step": 1160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5674260136537395e-05,
+      "loss": 0.2804,
+      "step": 1161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5666707068943975e-05,
+      "loss": 0.2939,
+      "step": 1162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.565914923642894e-05,
+      "loss": 0.2872,
+      "step": 1163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.565158664534739e-05,
+      "loss": 0.3076,
+      "step": 1164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.564401930205842e-05,
+      "loss": 0.265,
+      "step": 1165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5636447212925126e-05,
+      "loss": 0.3067,
+      "step": 1166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.56288703843146e-05,
+      "loss": 0.316,
+      "step": 1167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5621288822597903e-05,
+      "loss": 0.2898,
+      "step": 1168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.561370253415009e-05,
+      "loss": 0.2766,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5606111525350193e-05,
+      "loss": 0.2645,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5598515802581194e-05,
+      "loss": 0.2845,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.559091537223007e-05,
+      "loss": 0.3167,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.558331024068772e-05,
+      "loss": 0.3324,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5575700414349023e-05,
+      "loss": 0.3071,
+      "step": 1174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5568085899612802e-05,
+      "loss": 0.2858,
+      "step": 1175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5560466702881805e-05,
+      "loss": 0.2725,
+      "step": 1176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5552842830562737e-05,
+      "loss": 0.2799,
+      "step": 1177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.554521428906622e-05,
+      "loss": 0.2873,
+      "step": 1178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5537581084806816e-05,
+      "loss": 0.2512,
+      "step": 1179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5529943224202993e-05,
+      "loss": 0.3124,
+      "step": 1180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5522300713677153e-05,
+      "loss": 0.3593,
+      "step": 1181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5514653559655584e-05,
+      "loss": 0.3381,
+      "step": 1182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5507001768568503e-05,
+      "loss": 0.2759,
+      "step": 1183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5499345346850003e-05,
+      "loss": 0.268,
+      "step": 1184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5491684300938086e-05,
+      "loss": 0.2642,
+      "step": 1185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5484018637274645e-05,
+      "loss": 0.2838,
+      "step": 1186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.547634836230544e-05,
+      "loss": 0.2862,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546867348248012e-05,
+      "loss": 0.3152,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546099400425221e-05,
+      "loss": 0.3021,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5453309934079092e-05,
+      "loss": 0.2997,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.544562127842201e-05,
+      "loss": 0.2855,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.543792804374607e-05,
+      "loss": 0.3075,
+      "step": 1192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.543023023652022e-05,
+      "loss": 0.2485,
+      "step": 1193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5422527863217264e-05,
+      "loss": 0.2754,
+      "step": 1194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.541482093031383e-05,
+      "loss": 0.3078,
+      "step": 1195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5407109444290402e-05,
+      "loss": 0.3335,
+      "step": 1196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5399393411631268e-05,
+      "loss": 0.298,
+      "step": 1197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5391672838824555e-05,
+      "loss": 0.2743,
+      "step": 1198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5383947732362205e-05,
+      "loss": 0.3172,
+      "step": 1199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.537621809873997e-05,
+      "loss": 0.2857,
+      "step": 1200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5368483944457403e-05,
+      "loss": 0.2812,
+      "step": 1201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5360745276017873e-05,
+      "loss": 0.2816,
+      "step": 1202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5353002099928535e-05,
+      "loss": 0.2999,
+      "step": 1203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5345254422700328e-05,
+      "loss": 0.3432,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5337502250847993e-05,
+      "loss": 0.3025,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5329745590890038e-05,
+      "loss": 0.2784,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5321984449348742e-05,
+      "loss": 0.2804,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.531421883275017e-05,
+      "loss": 0.2753,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.530644874762412e-05,
+      "loss": 0.2824,
+      "step": 1209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5298674200504183e-05,
+      "loss": 0.2738,
+      "step": 1210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5290895197927675e-05,
+      "loss": 0.303,
+      "step": 1211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5283111746435672e-05,
+      "loss": 0.273,
+      "step": 1212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5275323852572985e-05,
+      "loss": 0.3162,
+      "step": 1213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.526753152288816e-05,
+      "loss": 0.2756,
+      "step": 1214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5259734763933478e-05,
+      "loss": 0.308,
+      "step": 1215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5251933582264938e-05,
+      "loss": 0.2652,
+      "step": 1216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5244127984442263e-05,
+      "loss": 0.2757,
+      "step": 1217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5236317977028891e-05,
+      "loss": 0.2993,
+      "step": 1218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5228503566591961e-05,
+      "loss": 0.3056,
+      "step": 1219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5220684759702318e-05,
+      "loss": 0.3096,
+      "step": 1220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5212861562934507e-05,
+      "loss": 0.2407,
+      "step": 1221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5205033982866752e-05,
+      "loss": 0.2852,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5197202026080984e-05,
+      "loss": 0.2917,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5189365699162795e-05,
+      "loss": 0.2699,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5181525008701458e-05,
+      "loss": 0.2786,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.517367996128992e-05,
+      "loss": 0.2893,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5165830563524782e-05,
+      "loss": 0.3045,
+      "step": 1227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5157976822006312e-05,
+      "loss": 0.3164,
+      "step": 1228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5150118743338429e-05,
+      "loss": 0.3065,
+      "step": 1229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5142256334128694e-05,
+      "loss": 0.2374,
+      "step": 1230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5134389600988308e-05,
+      "loss": 0.2877,
+      "step": 1231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.512651855053212e-05,
+      "loss": 0.2717,
+      "step": 1232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5118643189378597e-05,
+      "loss": 0.2847,
+      "step": 1233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5110763524149836e-05,
+      "loss": 0.3183,
+      "step": 1234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.510287956147155e-05,
+      "loss": 0.2941,
+      "step": 1235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5094991307973073e-05,
+      "loss": 0.3163,
+      "step": 1236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5087098770287333e-05,
+      "loss": 0.3119,
+      "step": 1237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5079201955050882e-05,
+      "loss": 0.273,
+      "step": 1238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5071300868903847e-05,
+      "loss": 0.2557,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5063395518489957e-05,
+      "loss": 0.2727,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5055485910456522e-05,
+      "loss": 0.307,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.504757205145444e-05,
+      "loss": 0.3396,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5039653948138175e-05,
+      "loss": 0.301,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5031731607165766e-05,
+      "loss": 0.2761,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5023805035198814e-05,
+      "loss": 0.2818,
+      "step": 1245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5015874238902468e-05,
+      "loss": 0.2807,
+      "step": 1246
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5007939224945446e-05,
+      "loss": 0.2915,
+      "step": 1247
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2461,
+      "step": 1248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4992056570741931e-05,
+      "loss": 0.3008,
+      "step": 1249
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4984108943850564e-05,
+      "loss": 0.307,
+      "step": 1250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4976157126008763e-05,
+      "loss": 0.315,
+      "step": 1251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4968201123902919e-05,
+      "loss": 0.3028,
+      "step": 1252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4960240944222933e-05,
+      "loss": 0.2487,
+      "step": 1253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4952276593662218e-05,
+      "loss": 0.2755,
+      "step": 1254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4944308078917703e-05,
+      "loss": 0.2811,
+      "step": 1255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4936335406689805e-05,
+      "loss": 0.2954,
+      "step": 1256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4928358583682458e-05,
+      "loss": 0.3034,
+      "step": 1257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4920377616603064e-05,
+      "loss": 0.2937,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4912392512162523e-05,
+      "loss": 0.343,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.490440327707521e-05,
+      "loss": 0.2695,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4896409918058969e-05,
+      "loss": 0.2937,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4888412441835122e-05,
+      "loss": 0.2456,
+      "step": 1262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4880410855128446e-05,
+      "loss": 0.2747,
+      "step": 1263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4872405164667175e-05,
+      "loss": 0.306,
+      "step": 1264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4864395377182988e-05,
+      "loss": 0.2805,
+      "step": 1265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4856381499411024e-05,
+      "loss": 0.3629,
+      "step": 1266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4848363538089847e-05,
+      "loss": 0.2328,
+      "step": 1267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4840341499961464e-05,
+      "loss": 0.3086,
+      "step": 1268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4832315391771301e-05,
+      "loss": 0.2888,
+      "step": 1269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.482428522026822e-05,
+      "loss": 0.3091,
+      "step": 1270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4816250992204481e-05,
+      "loss": 0.3114,
+      "step": 1271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4808212714335772e-05,
+      "loss": 0.283,
+      "step": 1272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.480017039342118e-05,
+      "loss": 0.3136,
+      "step": 1273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4792124036223187e-05,
+      "loss": 0.3246,
+      "step": 1274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4784073649507677e-05,
+      "loss": 0.2791,
+      "step": 1275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4776019240043917e-05,
+      "loss": 0.2459,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4767960814604559e-05,
+      "loss": 0.2631,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4759898379965628e-05,
+      "loss": 0.2854,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4751831942906528e-05,
+      "loss": 0.2971,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4743761510210016e-05,
+      "loss": 0.2989,
+      "step": 1280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4735687088662225e-05,
+      "loss": 0.2884,
+      "step": 1281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.472760868505263e-05,
+      "loss": 0.3232,
+      "step": 1282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4719526306174055e-05,
+      "loss": 0.2925,
+      "step": 1283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4711439958822675e-05,
+      "loss": 0.2723,
+      "step": 1284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4703349649797991e-05,
+      "loss": 0.2582,
+      "step": 1285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4695255385902844e-05,
+      "loss": 0.2745,
+      "step": 1286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4687157173943397e-05,
+      "loss": 0.2874,
+      "step": 1287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4679055020729134e-05,
+      "loss": 0.3143,
+      "step": 1288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.467094893307285e-05,
+      "loss": 0.3159,
+      "step": 1289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4662838917790651e-05,
+      "loss": 0.275,
+      "step": 1290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4654724981701942e-05,
+      "loss": 0.272,
+      "step": 1291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4646607131629434e-05,
+      "loss": 0.3025,
+      "step": 1292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4638485374399116e-05,
+      "loss": 0.2548,
+      "step": 1293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4630359716840276e-05,
+      "loss": 0.2511,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4622230165785467e-05,
+      "loss": 0.2833,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4614096728070533e-05,
+      "loss": 0.29,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4605959410534572e-05,
+      "loss": 0.3575,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4597818220019947e-05,
+      "loss": 0.2699,
+      "step": 1298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4589673163372284e-05,
+      "loss": 0.3092,
+      "step": 1299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4581524247440455e-05,
+      "loss": 0.2823,
+      "step": 1300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4573371479076577e-05,
+      "loss": 0.2872,
+      "step": 1301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.456521486513601e-05,
+      "loss": 0.2786,
+      "step": 1302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4557054412477343e-05,
+      "loss": 0.2888,
+      "step": 1303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4548890127962395e-05,
+      "loss": 0.2763,
+      "step": 1304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4540722018456207e-05,
+      "loss": 0.2819,
+      "step": 1305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4532550090827033e-05,
+      "loss": 0.2879,
+      "step": 1306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4524374351946348e-05,
+      "loss": 0.2758,
+      "step": 1307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.451619480868882e-05,
+      "loss": 0.2775,
+      "step": 1308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4508011467932317e-05,
+      "loss": 0.2805,
+      "step": 1309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4499824336557908e-05,
+      "loss": 0.3011,
+      "step": 1310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4491633421449846e-05,
+      "loss": 0.306,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4483438729495562e-05,
+      "loss": 0.3735,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4475240267585666e-05,
+      "loss": 0.2656,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4467038042613939e-05,
+      "loss": 0.3076,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4458832061477321e-05,
+      "loss": 0.2714,
+      "step": 1315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4450622331075914e-05,
+      "loss": 0.2878,
+      "step": 1316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.444240885831298e-05,
+      "loss": 0.2852,
+      "step": 1317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.443419165009491e-05,
+      "loss": 0.2926,
+      "step": 1318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4425970713331256e-05,
+      "loss": 0.322,
+      "step": 1319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.441774605493469e-05,
+      "loss": 0.2678,
+      "step": 1320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4409517681821019e-05,
+      "loss": 0.303,
+      "step": 1321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4401285600909175e-05,
+      "loss": 0.2464,
+      "step": 1322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4393049819121208e-05,
+      "loss": 0.2755,
+      "step": 1323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4384810343382277e-05,
+      "loss": 0.281,
+      "step": 1324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4376567180620644e-05,
+      "loss": 0.2811,
+      "step": 1325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4368320337767683e-05,
+      "loss": 0.3394,
+      "step": 1326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4360069821757852e-05,
+      "loss": 0.2697,
+      "step": 1327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.43518156395287e-05,
+      "loss": 0.311,
+      "step": 1328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4343557798020858e-05,
+      "loss": 0.2752,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4335296304178034e-05,
+      "loss": 0.2884,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4327031164947017e-05,
+      "loss": 0.2687,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4318762387277646e-05,
+      "loss": 0.2741,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.431048997812283e-05,
+      "loss": 0.2962,
+      "step": 1333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4302213944438525e-05,
+      "loss": 0.3171,
+      "step": 1334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.429393429318374e-05,
+      "loss": 0.307,
+      "step": 1335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4285651031320517e-05,
+      "loss": 0.2861,
+      "step": 1336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.427736416581395e-05,
+      "loss": 0.257,
+      "step": 1337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4269073703632148e-05,
+      "loss": 0.2902,
+      "step": 1338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.426077965174625e-05,
+      "loss": 0.264,
+      "step": 1339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4252482017130415e-05,
+      "loss": 0.2434,
+      "step": 1340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4244180806761813e-05,
+      "loss": 0.2518,
+      "step": 1341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4235876027620621e-05,
+      "loss": 0.2995,
+      "step": 1342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4227567686690016e-05,
+      "loss": 0.3408,
+      "step": 1343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4219255790956168e-05,
+      "loss": 0.2598,
+      "step": 1344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4210940347408242e-05,
+      "loss": 0.2728,
+      "step": 1345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4202621363038383e-05,
+      "loss": 0.2475,
+      "step": 1346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4194298844841707e-05,
+      "loss": 0.2669,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4185972799816316e-05,
+      "loss": 0.3003,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4177643234963267e-05,
+      "loss": 0.3044,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4169310157286577e-05,
+      "loss": 0.252,
+      "step": 1350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.416097357379322e-05,
+      "loss": 0.2759,
+      "step": 1351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4152633491493115e-05,
+      "loss": 0.3045,
+      "step": 1352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4144289917399128e-05,
+      "loss": 0.299,
+      "step": 1353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4135942858527053e-05,
+      "loss": 0.238,
+      "step": 1354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4127592321895622e-05,
+      "loss": 0.2709,
+      "step": 1355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4119238314526489e-05,
+      "loss": 0.2734,
+      "step": 1356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4110880843444226e-05,
+      "loss": 0.3652,
+      "step": 1357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4102519915676314e-05,
+      "loss": 0.3164,
+      "step": 1358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4094155538253148e-05,
+      "loss": 0.2769,
+      "step": 1359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4085787718208016e-05,
+      "loss": 0.2455,
+      "step": 1360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4077416462577102e-05,
+      "loss": 0.2927,
+      "step": 1361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.406904177839949e-05,
+      "loss": 0.2949,
+      "step": 1362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.406066367271713e-05,
+      "loss": 0.2834,
+      "step": 1363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.405228215257486e-05,
+      "loss": 0.2534,
+      "step": 1364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4043897225020389e-05,
+      "loss": 0.2974,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4035508897104282e-05,
+      "loss": 0.3066,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4027117175879974e-05,
+      "loss": 0.3031,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.401872206840375e-05,
+      "loss": 0.24,
+      "step": 1368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4010323581734737e-05,
+      "loss": 0.2672,
+      "step": 1369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4001921722934909e-05,
+      "loss": 0.2645,
+      "step": 1370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3993516499069078e-05,
+      "loss": 0.2807,
+      "step": 1371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3985107917204879e-05,
+      "loss": 0.3704,
+      "step": 1372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3976695984412771e-05,
+      "loss": 0.2699,
+      "step": 1373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3968280707766038e-05,
+      "loss": 0.315,
+      "step": 1374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3959862094340764e-05,
+      "loss": 0.278,
+      "step": 1375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3951440151215853e-05,
+      "loss": 0.266,
+      "step": 1376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3943014885472998e-05,
+      "loss": 0.2796,
+      "step": 1377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.393458630419669e-05,
+      "loss": 0.2479,
+      "step": 1378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3926154414474202e-05,
+      "loss": 0.2866,
+      "step": 1379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3917719223395603e-05,
+      "loss": 0.2989,
+      "step": 1380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3909280738053722e-05,
+      "loss": 0.3239,
+      "step": 1381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3900838965544168e-05,
+      "loss": 0.268,
+      "step": 1382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3892393912965311e-05,
+      "loss": 0.2742,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3883945587418281e-05,
+      "loss": 0.2743,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3875493996006954e-05,
+      "loss": 0.2914,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.386703914583796e-05,
+      "loss": 0.2818,
+      "step": 1386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3858581044020664e-05,
+      "loss": 0.2493,
+      "step": 1387
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3850119697667168e-05,
+      "loss": 0.2862,
+      "step": 1388
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3841655113892303e-05,
+      "loss": 0.3423,
+      "step": 1389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3833187299813613e-05,
+      "loss": 0.2855,
+      "step": 1390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3824716262551373e-05,
+      "loss": 0.2853,
+      "step": 1391
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3816242009228558e-05,
+      "loss": 0.2707,
+      "step": 1392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.380776454697085e-05,
+      "loss": 0.2895,
+      "step": 1393
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3799283882906626e-05,
+      "loss": 0.2677,
+      "step": 1394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3790800024166963e-05,
+      "loss": 0.3112,
+      "step": 1395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3782312977885624e-05,
+      "loss": 0.3187,
+      "step": 1396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.377382275119904e-05,
+      "loss": 0.2873,
+      "step": 1397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3765329351246328e-05,
+      "loss": 0.304,
+      "step": 1398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.375683278516927e-05,
+      "loss": 0.283,
+      "step": 1399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3748333060112306e-05,
+      "loss": 0.2805,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3739830183222543e-05,
+      "loss": 0.2608,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3731324161649728e-05,
+      "loss": 0.2721,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3722815002546255e-05,
+      "loss": 0.3296,
+      "step": 1403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3714302713067159e-05,
+      "loss": 0.2888,
+      "step": 1404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3705787300370102e-05,
+      "loss": 0.3065,
+      "step": 1405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3697268771615378e-05,
+      "loss": 0.2435,
+      "step": 1406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3688747133965898e-05,
+      "loss": 0.2656,
+      "step": 1407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3680222394587188e-05,
+      "loss": 0.2828,
+      "step": 1408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3671694560647381e-05,
+      "loss": 0.2971,
+      "step": 1409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3663163639317216e-05,
+      "loss": 0.3101,
+      "step": 1410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3654629637770021e-05,
+      "loss": 0.2751,
+      "step": 1411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3646092563181722e-05,
+      "loss": 0.2993,
+      "step": 1412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.363755242273082e-05,
+      "loss": 0.3038,
+      "step": 1413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3629009223598405e-05,
+      "loss": 0.266,
+      "step": 1414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3620462972968128e-05,
+      "loss": 0.2623,
+      "step": 1415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3611913678026215e-05,
+      "loss": 0.2862,
+      "step": 1416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3603361345961443e-05,
+      "loss": 0.2509,
+      "step": 1417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3594805983965152e-05,
+      "loss": 0.2935,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3586247599231223e-05,
+      "loss": 0.2907,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3577686198956081e-05,
+      "loss": 0.2651,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3569121790338689e-05,
+      "loss": 0.2803,
+      "step": 1421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3560554380580536e-05,
+      "loss": 0.2697,
+      "step": 1422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3551983976885633e-05,
+      "loss": 0.2696,
+      "step": 1423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3543410586460512e-05,
+      "loss": 0.2581,
+      "step": 1424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.353483421651422e-05,
+      "loss": 0.3248,
+      "step": 1425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3526254874258303e-05,
+      "loss": 0.273,
+      "step": 1426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3517672566906803e-05,
+      "loss": 0.3194,
+      "step": 1427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.350908730167627e-05,
+      "loss": 0.293,
+      "step": 1428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3500499085785726e-05,
+      "loss": 0.2567,
+      "step": 1429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.349190792645668e-05,
+      "loss": 0.2491,
+      "step": 1430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3483313830913118e-05,
+      "loss": 0.2768,
+      "step": 1431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3474716806381492e-05,
+      "loss": 0.2962,
+      "step": 1432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.346611686009072e-05,
+      "loss": 0.2711,
+      "step": 1433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3457513999272168e-05,
+      "loss": 0.2984,
+      "step": 1434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3448908231159663e-05,
+      "loss": 0.3166,
+      "step": 1435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3440299562989476e-05,
+      "loss": 0.2754,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3431688002000312e-05,
+      "loss": 0.2805,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3423073555433308e-05,
+      "loss": 0.2461,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3414456230532027e-05,
+      "loss": 0.2846,
+      "step": 1439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3405836034542459e-05,
+      "loss": 0.2578,
+      "step": 1440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3397212974713005e-05,
+      "loss": 0.302,
+      "step": 1441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3388587058294466e-05,
+      "loss": 0.2843,
+      "step": 1442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3379958292540058e-05,
+      "loss": 0.2786,
+      "step": 1443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3371326684705386e-05,
+      "loss": 0.295,
+      "step": 1444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3362692242048443e-05,
+      "loss": 0.2719,
+      "step": 1445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.335405497182961e-05,
+      "loss": 0.2626,
+      "step": 1446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3345414881311645e-05,
+      "loss": 0.2761,
+      "step": 1447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3336771977759675e-05,
+      "loss": 0.2599,
+      "step": 1448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3328126268441194e-05,
+      "loss": 0.3205,
+      "step": 1449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.331947776062605e-05,
+      "loss": 0.3101,
+      "step": 1450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3310826461586457e-05,
+      "loss": 0.3024,
+      "step": 1451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3302172378596964e-05,
+      "loss": 0.2607,
+      "step": 1452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3293515518934468e-05,
+      "loss": 0.2529,
+      "step": 1453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3284855889878192e-05,
+      "loss": 0.2774,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.32761934987097e-05,
+      "loss": 0.2695,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3267528352712869e-05,
+      "loss": 0.294,
+      "step": 1456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3258860459173896e-05,
+      "loss": 0.2802,
+      "step": 1457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3250189825381288e-05,
+      "loss": 0.2951,
+      "step": 1458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3241516458625854e-05,
+      "loss": 0.2777,
+      "step": 1459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3232840366200706e-05,
+      "loss": 0.2685,
+      "step": 1460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3224161555401245e-05,
+      "loss": 0.2525,
+      "step": 1461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3215480033525157e-05,
+      "loss": 0.2807,
+      "step": 1462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3206795807872405e-05,
+      "loss": 0.2815,
+      "step": 1463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3198108885745236e-05,
+      "loss": 0.2925,
+      "step": 1464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.318941927444815e-05,
+      "loss": 0.2981,
+      "step": 1465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3180726981287922e-05,
+      "loss": 0.2854,
+      "step": 1466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.317203201357357e-05,
+      "loss": 0.2893,
+      "step": 1467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.316333437861637e-05,
+      "loss": 0.3061,
+      "step": 1468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3154634083729834e-05,
+      "loss": 0.2667,
+      "step": 1469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3145931136229719e-05,
+      "loss": 0.2578,
+      "step": 1470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3137225543434003e-05,
+      "loss": 0.261,
+      "step": 1471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3128517312662895e-05,
+      "loss": 0.2984,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.311980645123882e-05,
+      "loss": 0.3361,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.311109296648641e-05,
+      "loss": 0.2902,
+      "step": 1474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3102376865732514e-05,
+      "loss": 0.2518,
+      "step": 1475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3093658156306173e-05,
+      "loss": 0.2874,
+      "step": 1476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3084936845538623e-05,
+      "loss": 0.2657,
+      "step": 1477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3076212940763287e-05,
+      "loss": 0.2698,
+      "step": 1478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3067486449315769e-05,
+      "loss": 0.2749,
+      "step": 1479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3058757378533845e-05,
+      "loss": 0.3087,
+      "step": 1480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3050025735757474e-05,
+      "loss": 0.3139,
+      "step": 1481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.304129152832876e-05,
+      "loss": 0.2939,
+      "step": 1482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3032554763591967e-05,
+      "loss": 0.2871,
+      "step": 1483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.302381544889352e-05,
+      "loss": 0.2465,
+      "step": 1484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3015073591581977e-05,
+      "loss": 0.2592,
+      "step": 1485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3006329199008043e-05,
+      "loss": 0.2809,
+      "step": 1486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2997582278524545e-05,
+      "loss": 0.3228,
+      "step": 1487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2988832837486445e-05,
+      "loss": 0.2747,
+      "step": 1488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2980080883250814e-05,
+      "loss": 0.3139,
+      "step": 1489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2971326423176849e-05,
+      "loss": 0.2785,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2962569464625841e-05,
+      "loss": 0.2818,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2953810014961194e-05,
+      "loss": 0.286,
+      "step": 1492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2945048081548399e-05,
+      "loss": 0.2534,
+      "step": 1493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2936283671755039e-05,
+      "loss": 0.2654,
+      "step": 1494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2927516792950771e-05,
+      "loss": 0.3161,
+      "step": 1495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2918747452507343e-05,
+      "loss": 0.3141,
+      "step": 1496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2909975657798562e-05,
+      "loss": 0.2948,
+      "step": 1497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.29012014162003e-05,
+      "loss": 0.2367,
+      "step": 1498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2892424735090487e-05,
+      "loss": 0.2544,
+      "step": 1499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.288364562184911e-05,
+      "loss": 0.2875,
+      "step": 1500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2874864083858194e-05,
+      "loss": 0.2761,
+      "step": 1501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2866080128501804e-05,
+      "loss": 0.2763,
+      "step": 1502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2857293763166037e-05,
+      "loss": 0.2726,
+      "step": 1503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2848504995239022e-05,
+      "loss": 0.333,
+      "step": 1504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2839713832110906e-05,
+      "loss": 0.2782,
+      "step": 1505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2830920281173845e-05,
+      "loss": 0.277,
+      "step": 1506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2822124349822003e-05,
+      "loss": 0.2792,
+      "step": 1507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2813326045451557e-05,
+      "loss": 0.2777,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.280452537546067e-05,
+      "loss": 0.2694,
+      "step": 1509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.279572234724949e-05,
+      "loss": 0.3188,
+      "step": 1510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.278691696822016e-05,
+      "loss": 0.2987,
+      "step": 1511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2778109245776791e-05,
+      "loss": 0.2604,
+      "step": 1512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2769299187325465e-05,
+      "loss": 0.3177,
+      "step": 1513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2760486800274227e-05,
+      "loss": 0.2814,
+      "step": 1514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2751672092033095e-05,
+      "loss": 0.2845,
+      "step": 1515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2742855070014015e-05,
+      "loss": 0.2445,
+      "step": 1516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2734035741630894e-05,
+      "loss": 0.2447,
+      "step": 1517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2725214114299576e-05,
+      "loss": 0.3248,
+      "step": 1518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2716390195437835e-05,
+      "loss": 0.305,
+      "step": 1519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2707563992465378e-05,
+      "loss": 0.321,
+      "step": 1520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2698735512803821e-05,
+      "loss": 0.2518,
+      "step": 1521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2689904763876709e-05,
+      "loss": 0.277,
+      "step": 1522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2681071753109479e-05,
+      "loss": 0.284,
+      "step": 1523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2672236487929488e-05,
+      "loss": 0.2826,
+      "step": 1524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2663398975765974e-05,
+      "loss": 0.3027,
+      "step": 1525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2654559224050074e-05,
+      "loss": 0.2779,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2645717240214801e-05,
+      "loss": 0.3132,
+      "step": 1527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.263687303169505e-05,
+      "loss": 0.3126,
+      "step": 1528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2628026605927576e-05,
+      "loss": 0.2588,
+      "step": 1529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.261917797035102e-05,
+      "loss": 0.266,
+      "step": 1530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.261032713240586e-05,
+      "loss": 0.2419,
+      "step": 1531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2601474099534437e-05,
+      "loss": 0.2563,
+      "step": 1532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2592618879180934e-05,
+      "loss": 0.3269,
+      "step": 1533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2583761478791367e-05,
+      "loss": 0.3003,
+      "step": 1534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2574901905813603e-05,
+      "loss": 0.2414,
+      "step": 1535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2566040167697322e-05,
+      "loss": 0.2575,
+      "step": 1536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2557176271894022e-05,
+      "loss": 0.2841,
+      "step": 1537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2548310225857026e-05,
+      "loss": 0.2632,
+      "step": 1538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2539442037041456e-05,
+      "loss": 0.2535,
+      "step": 1539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2530571712904246e-05,
+      "loss": 0.2642,
+      "step": 1540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2521699260904113e-05,
+      "loss": 0.2736,
+      "step": 1541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2512824688501572e-05,
+      "loss": 0.3389,
+      "step": 1542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2503948003158918e-05,
+      "loss": 0.2912,
+      "step": 1543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.249506921234022e-05,
+      "loss": 0.2811,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2486188323511328e-05,
+      "loss": 0.2568,
+      "step": 1545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2477305344139842e-05,
+      "loss": 0.2756,
+      "step": 1546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.246842028169513e-05,
+      "loss": 0.2803,
+      "step": 1547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2459533143648306e-05,
+      "loss": 0.3168,
+      "step": 1548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2450643937472229e-05,
+      "loss": 0.2651,
+      "step": 1549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2441752670641504e-05,
+      "loss": 0.3048,
+      "step": 1550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2432859350632462e-05,
+      "loss": 0.3197,
+      "step": 1551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2423963984923166e-05,
+      "loss": 0.258,
+      "step": 1552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.241506658099339e-05,
+      "loss": 0.2674,
+      "step": 1553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2406167146324628e-05,
+      "loss": 0.2526,
+      "step": 1554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2397265688400085e-05,
+      "loss": 0.2672,
+      "step": 1555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.238836221470466e-05,
+      "loss": 0.2927,
+      "step": 1556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2379456732724955e-05,
+      "loss": 0.3151,
+      "step": 1557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2370549249949254e-05,
+      "loss": 0.2605,
+      "step": 1558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2361639773867519e-05,
+      "loss": 0.2895,
+      "step": 1559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2352728311971402e-05,
+      "loss": 0.2598,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2343814871754219e-05,
+      "loss": 0.2738,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.233489946071094e-05,
+      "loss": 0.2656,
+      "step": 1562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2325982086338208e-05,
+      "loss": 0.2744,
+      "step": 1563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2317062756134302e-05,
+      "loss": 0.2954,
+      "step": 1564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2308141477599155e-05,
+      "loss": 0.2988,
+      "step": 1565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.229921825823434e-05,
+      "loss": 0.2902,
+      "step": 1566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2290293105543052e-05,
+      "loss": 0.2585,
+      "step": 1567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.228136602703012e-05,
+      "loss": 0.2486,
+      "step": 1568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2272437030201989e-05,
+      "loss": 0.255,
+      "step": 1569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.226350612256672e-05,
+      "loss": 0.2681,
+      "step": 1570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2254573311633973e-05,
+      "loss": 0.282,
+      "step": 1571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2245638604915023e-05,
+      "loss": 0.2632,
+      "step": 1572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2236702009922721e-05,
+      "loss": 0.3303,
+      "step": 1573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.222776353417152e-05,
+      "loss": 0.2666,
+      "step": 1574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2218823185177447e-05,
+      "loss": 0.2784,
+      "step": 1575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.220988097045811e-05,
+      "loss": 0.2459,
+      "step": 1576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2200936897532678e-05,
+      "loss": 0.2472,
+      "step": 1577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.219199097392189e-05,
+      "loss": 0.2839,
+      "step": 1578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2183043207148032e-05,
+      "loss": 0.2625,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2174093604734953e-05,
+      "loss": 0.2903,
+      "step": 1580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2165142174208038e-05,
+      "loss": 0.2728,
+      "step": 1581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2156188923094203e-05,
+      "loss": 0.2686,
+      "step": 1582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2147233858921905e-05,
+      "loss": 0.2585,
+      "step": 1583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2138276989221125e-05,
+      "loss": 0.2727,
+      "step": 1584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2129318321523353e-05,
+      "loss": 0.2811,
+      "step": 1585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2120357863361602e-05,
+      "loss": 0.2426,
+      "step": 1586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2111395622270383e-05,
+      "loss": 0.2994,
+      "step": 1587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.210243160578571e-05,
+      "loss": 0.3205,
+      "step": 1588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2093465821445084e-05,
+      "loss": 0.2768,
+      "step": 1589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2084498276787506e-05,
+      "loss": 0.2527,
+      "step": 1590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.207552897935344e-05,
+      "loss": 0.236,
+      "step": 1591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.206655793668484e-05,
+      "loss": 0.2871,
+      "step": 1592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2057585156325111e-05,
+      "loss": 0.2425,
+      "step": 1593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.204861064581913e-05,
+      "loss": 0.2782,
+      "step": 1594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.203963441271323e-05,
+      "loss": 0.2443,
+      "step": 1595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2030656464555186e-05,
+      "loss": 0.2961,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2021676808894223e-05,
+      "loss": 0.3217,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2012695453280987e-05,
+      "loss": 0.2778,
+      "step": 1598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2003712405267573e-05,
+      "loss": 0.2801,
+      "step": 1599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1994727672407487e-05,
+      "loss": 0.2269,
+      "step": 1600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1985741262255653e-05,
+      "loss": 0.2557,
+      "step": 1601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1976753182368406e-05,
+      "loss": 0.3119,
+      "step": 1602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1967763440303485e-05,
+      "loss": 0.2596,
+      "step": 1603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1958772043620028e-05,
+      "loss": 0.2917,
+      "step": 1604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1949778999878566e-05,
+      "loss": 0.2469,
+      "step": 1605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.194078431664101e-05,
+      "loss": 0.2655,
+      "step": 1606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1931788001470655e-05,
+      "loss": 0.2472,
+      "step": 1607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1922790061932159e-05,
+      "loss": 0.2556,
+      "step": 1608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1913790505591556e-05,
+      "loss": 0.2493,
+      "step": 1609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1904789340016238e-05,
+      "loss": 0.2973,
+      "step": 1610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1895786572774945e-05,
+      "loss": 0.3082,
+      "step": 1611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1886782211437765e-05,
+      "loss": 0.2754,
+      "step": 1612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1877776263576132e-05,
+      "loss": 0.2477,
+      "step": 1613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1868768736762808e-05,
+      "loss": 0.2593,
+      "step": 1614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1859759638571887e-05,
+      "loss": 0.2675,
+      "step": 1615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1850748976578776e-05,
+      "loss": 0.2645,
+      "step": 1616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.184173675836021e-05,
+      "loss": 0.3023,
+      "step": 1617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.183272299149422e-05,
+      "loss": 0.2861,
+      "step": 1618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1823707683560146e-05,
+      "loss": 0.2898,
+      "step": 1619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1814690842138628e-05,
+      "loss": 0.2674,
+      "step": 1620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1805672474811582e-05,
+      "loss": 0.2746,
+      "step": 1621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1796652589162224e-05,
+      "loss": 0.2471,
+      "step": 1622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1787631192775031e-05,
+      "loss": 0.2452,
+      "step": 1623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1778608293235755e-05,
+      "loss": 0.2605,
+      "step": 1624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1769583898131423e-05,
+      "loss": 0.2749,
+      "step": 1625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1760558015050302e-05,
+      "loss": 0.3263,
+      "step": 1626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1751530651581926e-05,
+      "loss": 0.2745,
+      "step": 1627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.174250181531706e-05,
+      "loss": 0.2551,
+      "step": 1628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.173347151384772e-05,
+      "loss": 0.2727,
+      "step": 1629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1724439754767145e-05,
+      "loss": 0.2445,
+      "step": 1630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1715406545669803e-05,
+      "loss": 0.2919,
+      "step": 1631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1706371894151382e-05,
+      "loss": 0.2577,
+      "step": 1632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1697335807808779e-05,
+      "loss": 0.308,
+      "step": 1633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1688298294240104e-05,
+      "loss": 0.2851,
+      "step": 1634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.167925936104466e-05,
+      "loss": 0.3064,
+      "step": 1635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.167021901582295e-05,
+      "loss": 0.2592,
+      "step": 1636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.166117726617666e-05,
+      "loss": 0.2512,
+      "step": 1637
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.165213411970866e-05,
+      "loss": 0.2984,
+      "step": 1638
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.164308958402299e-05,
+      "loss": 0.2573,
+      "step": 1639
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1634043666724868e-05,
+      "loss": 0.2894,
+      "step": 1640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.162499637542066e-05,
+      "loss": 0.2652,
+      "step": 1641
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1615947717717895e-05,
+      "loss": 0.2839,
+      "step": 1642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.160689770122525e-05,
+      "loss": 0.2896,
+      "step": 1643
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1597846333552549e-05,
+      "loss": 0.2721,
+      "step": 1644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1588793622310743e-05,
+      "loss": 0.2883,
+      "step": 1645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1579739575111924e-05,
+      "loss": 0.2154,
+      "step": 1646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1570684199569293e-05,
+      "loss": 0.2846,
+      "step": 1647
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1561627503297177e-05,
+      "loss": 0.3151,
+      "step": 1648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1552569493911019e-05,
+      "loss": 0.2851,
+      "step": 1649
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1543510179027355e-05,
+      "loss": 0.2937,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.153444956626382e-05,
+      "loss": 0.2463,
+      "step": 1651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1525387663239153e-05,
+      "loss": 0.2667,
+      "step": 1652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1516324477573163e-05,
+      "loss": 0.2482,
+      "step": 1653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1507260016886742e-05,
+      "loss": 0.297,
+      "step": 1654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1498194288801858e-05,
+      "loss": 0.2663,
+      "step": 1655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.148912730094154e-05,
+      "loss": 0.265,
+      "step": 1656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1480059060929882e-05,
+      "loss": 0.3291,
+      "step": 1657
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.147098957639202e-05,
+      "loss": 0.2926,
+      "step": 1658
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.146191885495415e-05,
+      "loss": 0.2575,
+      "step": 1659
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1452846904243501e-05,
+      "loss": 0.2516,
+      "step": 1660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1443773731888337e-05,
+      "loss": 0.2647,
+      "step": 1661
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1434699345517946e-05,
+      "loss": 0.2718,
+      "step": 1662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1425623752762636e-05,
+      "loss": 0.2408,
+      "step": 1663
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.141654696125374e-05,
+      "loss": 0.2936,
+      "step": 1664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1407468978623591e-05,
+      "loss": 0.2707,
+      "step": 1665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1398389812505522e-05,
+      "loss": 0.288,
+      "step": 1666
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.138930947053386e-05,
+      "loss": 0.2725,
+      "step": 1667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.138022796034393e-05,
+      "loss": 0.2825,
+      "step": 1668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1371145289572032e-05,
+      "loss": 0.2524,
+      "step": 1669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1362061465855445e-05,
+      "loss": 0.2689,
+      "step": 1670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1352976496832412e-05,
+      "loss": 0.302,
+      "step": 1671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.134389039014215e-05,
+      "loss": 0.2983,
+      "step": 1672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1334803153424814e-05,
+      "loss": 0.2631,
+      "step": 1673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1325714794321532e-05,
+      "loss": 0.2661,
+      "step": 1674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.131662532047436e-05,
+      "loss": 0.256,
+      "step": 1675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1307534739526295e-05,
+      "loss": 0.2849,
+      "step": 1676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.129844305912127e-05,
+      "loss": 0.3034,
+      "step": 1677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.128935028690413e-05,
+      "loss": 0.2666,
+      "step": 1678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1280256430520656e-05,
+      "loss": 0.2905,
+      "step": 1679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1271161497617526e-05,
+      "loss": 0.2859,
+      "step": 1680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.126206549584233e-05,
+      "loss": 0.281,
+      "step": 1681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.125296843284355e-05,
+      "loss": 0.2719,
+      "step": 1682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1243870316270571e-05,
+      "loss": 0.2478,
+      "step": 1683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1234771153773658e-05,
+      "loss": 0.2686,
+      "step": 1684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1225670953003958e-05,
+      "loss": 0.2849,
+      "step": 1685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1216569721613482e-05,
+      "loss": 0.2848,
+      "step": 1686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1207467467255118e-05,
+      "loss": 0.2662,
+      "step": 1687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1198364197582608e-05,
+      "loss": 0.2967,
+      "step": 1688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1189259920250554e-05,
+      "loss": 0.2932,
+      "step": 1689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1180154642914403e-05,
+      "loss": 0.2873,
+      "step": 1690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1171048373230438e-05,
+      "loss": 0.2589,
+      "step": 1691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.116194111885578e-05,
+      "loss": 0.2905,
+      "step": 1692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1152832887448374e-05,
+      "loss": 0.237,
+      "step": 1693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1143723686667e-05,
+      "loss": 0.2857,
+      "step": 1694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1134613524171237e-05,
+      "loss": 0.3036,
+      "step": 1695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1125502407621485e-05,
+      "loss": 0.2977,
+      "step": 1696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1116390344678931e-05,
+      "loss": 0.2546,
+      "step": 1697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1107277343005575e-05,
+      "loss": 0.2761,
+      "step": 1698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1098163410264193e-05,
+      "loss": 0.2469,
+      "step": 1699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1089048554118357e-05,
+      "loss": 0.2612,
+      "step": 1700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1079932782232404e-05,
+      "loss": 0.236,
+      "step": 1701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1070816102271442e-05,
+      "loss": 0.2533,
+      "step": 1702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1061698521901347e-05,
+      "loss": 0.3276,
+      "step": 1703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1052580048788754e-05,
+      "loss": 0.294,
+      "step": 1704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1043460690601045e-05,
+      "loss": 0.2667,
+      "step": 1705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1034340455006343e-05,
+      "loss": 0.2217,
+      "step": 1706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1025219349673513e-05,
+      "loss": 0.2498,
+      "step": 1707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.101609738227215e-05,
+      "loss": 0.2649,
+      "step": 1708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1006974560472579e-05,
+      "loss": 0.2608,
+      "step": 1709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0997850891945836e-05,
+      "loss": 0.3034,
+      "step": 1710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0988726384363673e-05,
+      "loss": 0.2696,
+      "step": 1711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0979601045398542e-05,
+      "loss": 0.2955,
+      "step": 1712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0970474882723607e-05,
+      "loss": 0.2579,
+      "step": 1713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.096134790401271e-05,
+      "loss": 0.2547,
+      "step": 1714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0952220116940384e-05,
+      "loss": 0.2548,
+      "step": 1715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0943091529181849e-05,
+      "loss": 0.2664,
+      "step": 1716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0933962148412986e-05,
+      "loss": 0.2626,
+      "step": 1717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0924831982310354e-05,
+      "loss": 0.3006,
+      "step": 1718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0915701038551167e-05,
+      "loss": 0.2871,
+      "step": 1719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0906569324813291e-05,
+      "loss": 0.2519,
+      "step": 1720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0897436848775245e-05,
+      "loss": 0.2587,
+      "step": 1721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0888303618116185e-05,
+      "loss": 0.2681,
+      "step": 1722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0879169640515897e-05,
+      "loss": 0.2632,
+      "step": 1723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0870034923654808e-05,
+      "loss": 0.2782,
+      "step": 1724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0860899475213956e-05,
+      "loss": 0.2717,
+      "step": 1725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0851763302874998e-05,
+      "loss": 0.2791,
+      "step": 1726
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0842626414320194e-05,
+      "loss": 0.298,
+      "step": 1727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.083348881723242e-05,
+      "loss": 0.2562,
+      "step": 1728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0824350519295133e-05,
+      "loss": 0.2622,
+      "step": 1729
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0815211528192386e-05,
+      "loss": 0.2513,
+      "step": 1730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0806071851608814e-05,
+      "loss": 0.2383,
+      "step": 1731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.079693149722963e-05,
+      "loss": 0.2856,
+      "step": 1732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0787790472740616e-05,
+      "loss": 0.2443,
+      "step": 1733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0778648785828113e-05,
+      "loss": 0.278,
+      "step": 1734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.076950644417903e-05,
+      "loss": 0.3118,
+      "step": 1735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0760363455480814e-05,
+      "loss": 0.2614,
+      "step": 1736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0751219827421461e-05,
+      "loss": 0.2574,
+      "step": 1737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0742075567689504e-05,
+      "loss": 0.2589,
+      "step": 1738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0732930683974017e-05,
+      "loss": 0.2334,
+      "step": 1739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0723785183964581e-05,
+      "loss": 0.2572,
+      "step": 1740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.071463907535131e-05,
+      "loss": 0.2917,
+      "step": 1741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0705492365824818e-05,
+      "loss": 0.2999,
+      "step": 1742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0696345063076237e-05,
+      "loss": 0.2494,
+      "step": 1743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0687197174797185e-05,
+      "loss": 0.2742,
+      "step": 1744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0678048708679785e-05,
+      "loss": 0.2663,
+      "step": 1745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0668899672416632e-05,
+      "loss": 0.2659,
+      "step": 1746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0659750073700812e-05,
+      "loss": 0.2722,
+      "step": 1747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0650599920225877e-05,
+      "loss": 0.271,
+      "step": 1748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0641449219685854e-05,
+      "loss": 0.3287,
+      "step": 1749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.063229797977522e-05,
+      "loss": 0.3002,
+      "step": 1750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0623146208188909e-05,
+      "loss": 0.2754,
+      "step": 1751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0613993912622303e-05,
+      "loss": 0.2616,
+      "step": 1752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0604841100771224e-05,
+      "loss": 0.244,
+      "step": 1753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0595687780331932e-05,
+      "loss": 0.2431,
+      "step": 1754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0586533959001109e-05,
+      "loss": 0.2569,
+      "step": 1755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.057737964447586e-05,
+      "loss": 0.3172,
+      "step": 1756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0568224844453702e-05,
+      "loss": 0.2486,
+      "step": 1757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0559069566632569e-05,
+      "loss": 0.2959,
+      "step": 1758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.054991381871079e-05,
+      "loss": 0.2514,
+      "step": 1759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0540757608387086e-05,
+      "loss": 0.2708,
+      "step": 1760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0531600943360576e-05,
+      "loss": 0.244,
+      "step": 1761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.052244383133075e-05,
+      "loss": 0.2304,
+      "step": 1762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0513286279997488e-05,
+      "loss": 0.281,
+      "step": 1763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0504128297061029e-05,
+      "loss": 0.2742,
+      "step": 1764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0494969890221975e-05,
+      "loss": 0.3004,
+      "step": 1765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0485811067181292e-05,
+      "loss": 0.227,
+      "step": 1766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0476651835640288e-05,
+      "loss": 0.2631,
+      "step": 1767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0467492203300615e-05,
+      "loss": 0.2745,
+      "step": 1768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.045833217786427e-05,
+      "loss": 0.29,
+      "step": 1769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0449171767033571e-05,
+      "loss": 0.2469,
+      "step": 1770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0440010978511166e-05,
+      "loss": 0.2546,
+      "step": 1771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0430849820000017e-05,
+      "loss": 0.2846,
+      "step": 1772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0421688299203404e-05,
+      "loss": 0.2722,
+      "step": 1773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.04125264238249e-05,
+      "loss": 0.2711,
+      "step": 1774
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0403364201568387e-05,
+      "loss": 0.2492,
+      "step": 1775
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0394201640138034e-05,
+      "loss": 0.2731,
+      "step": 1776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.038503874723829e-05,
+      "loss": 0.2441,
+      "step": 1777
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0375875530573896e-05,
+      "loss": 0.2843,
+      "step": 1778
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0366711997849852e-05,
+      "loss": 0.2762,
+      "step": 1779
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0357548156771433e-05,
+      "loss": 0.27,
+      "step": 1780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0348384015044168e-05,
+      "loss": 0.3008,
+      "step": 1781
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0339219580373841e-05,
+      "loss": 0.2703,
+      "step": 1782
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0330054860466479e-05,
+      "loss": 0.2705,
+      "step": 1783
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0320889863028356e-05,
+      "loss": 0.2457,
+      "step": 1784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.031172459576597e-05,
+      "loss": 0.2689,
+      "step": 1785
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0302559066386055e-05,
+      "loss": 0.2863,
+      "step": 1786
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0293393282595558e-05,
+      "loss": 0.284,
+      "step": 1787
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0284227252101646e-05,
+      "loss": 0.2927,
+      "step": 1788
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0275060982611688e-05,
+      "loss": 0.3048,
+      "step": 1789
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.026589448183326e-05,
+      "loss": 0.2332,
+      "step": 1790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0256727757474126e-05,
+      "loss": 0.2552,
+      "step": 1791
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0247560817242237e-05,
+      "loss": 0.2848,
+      "step": 1792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0238393668845739e-05,
+      "loss": 0.2759,
+      "step": 1793
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0229226319992938e-05,
+      "loss": 0.2617,
+      "step": 1794
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.022005877839231e-05,
+      "loss": 0.3228,
+      "step": 1795
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0210891051752508e-05,
+      "loss": 0.2946,
+      "step": 1796
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0201723147782322e-05,
+      "loss": 0.2704,
+      "step": 1797
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0192555074190696e-05,
+      "loss": 0.2458,
+      "step": 1798
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0183386838686728e-05,
+      "loss": 0.2678,
+      "step": 1799
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.017421844897964e-05,
+      "loss": 0.252,
+      "step": 1800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0165049912778788e-05,
+      "loss": 0.2423,
+      "step": 1801
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0155881237793646e-05,
+      "loss": 0.3023,
+      "step": 1802
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0146712431733816e-05,
+      "loss": 0.2602,
+      "step": 1803
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0137543502309e-05,
+      "loss": 0.3041,
+      "step": 1804
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0128374457229005e-05,
+      "loss": 0.2675,
+      "step": 1805
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0119205304203741e-05,
+      "loss": 0.2717,
+      "step": 1806
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0110036050943204e-05,
+      "loss": 0.2553,
+      "step": 1807
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0100866705157475e-05,
+      "loss": 0.2273,
+      "step": 1808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0091697274556713e-05,
+      "loss": 0.2851,
+      "step": 1809
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0082527766851151e-05,
+      "loss": 0.2939,
+      "step": 1810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0073358189751082e-05,
+      "loss": 0.3238,
+      "step": 1811
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0064188550966858e-05,
+      "loss": 0.2406,
+      "step": 1812
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0055018858208887e-05,
+      "loss": 0.2649,
+      "step": 1813
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0045849119187622e-05,
+      "loss": 0.2575,
+      "step": 1814
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.003667934161355e-05,
+      "loss": 0.2451,
+      "step": 1815
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0027509533197198e-05,
+      "loss": 0.2711,
+      "step": 1816
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0018339701649105e-05,
+      "loss": 0.2263,
+      "step": 1817
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0009169854679848e-05,
+      "loss": 0.3208,
+      "step": 1818
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.263,
+      "step": 1819
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.990830145320154e-06,
+      "loss": 0.2951,
+      "step": 1820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.981660298350898e-06,
+      "loss": 0.2301,
+      "step": 1821
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.972490466802806e-06,
+      "loss": 0.2353,
+      "step": 1822
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.963320658386451e-06,
+      "loss": 0.2443,
+      "step": 1823
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.954150880812377e-06,
+      "loss": 0.2384,
+      "step": 1824
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.944981141791115e-06,
+      "loss": 0.2879,
+      "step": 1825
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.935811449033145e-06,
+      "loss": 0.2602,
+      "step": 1826
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.926641810248923e-06,
+      "loss": 0.2717,
+      "step": 1827
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.917472233148852e-06,
+      "loss": 0.2653,
+      "step": 1828
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.90830272544329e-06,
+      "loss": 0.2822,
+      "step": 1829
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.899133294842528e-06,
+      "loss": 0.2771,
+      "step": 1830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.8899639490568e-06,
+      "loss": 0.2296,
+      "step": 1831
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.88079469579626e-06,
+      "loss": 0.2501,
+      "step": 1832
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.871625542771e-06,
+      "loss": 0.2943,
+      "step": 1833
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.862456497691004e-06,
+      "loss": 0.3017,
+      "step": 1834
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.85328756826619e-06,
+      "loss": 0.2752,
+      "step": 1835
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.844118762206357e-06,
+      "loss": 0.2393,
+      "step": 1836
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.834950087221215e-06,
+      "loss": 0.2781,
+      "step": 1837
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.825781551020363e-06,
+      "loss": 0.2457,
+      "step": 1838
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.816613161313272e-06,
+      "loss": 0.2509,
+      "step": 1839
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.807444925809306e-06,
+      "loss": 0.2209,
+      "step": 1840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.798276852217683e-06,
+      "loss": 0.303,
+      "step": 1841
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.789108948247497e-06,
+      "loss": 0.3185,
+      "step": 1842
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.77994122160769e-06,
+      "loss": 0.2643,
+      "step": 1843
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.770773680007068e-06,
+      "loss": 0.245,
+      "step": 1844
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.761606331154264e-06,
+      "loss": 0.2492,
+      "step": 1845
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.752439182757765e-06,
+      "loss": 0.249,
+      "step": 1846
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.743272242525877e-06,
+      "loss": 0.2643,
+      "step": 1847
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.734105518166745e-06,
+      "loss": 0.31,
+      "step": 1848
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.724939017388313e-06,
+      "loss": 0.2723,
+      "step": 1849
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.715772747898359e-06,
+      "loss": 0.2532,
+      "step": 1850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.706606717404444e-06,
+      "loss": 0.2485,
+      "step": 1851
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.697440933613947e-06,
+      "loss": 0.2527,
+      "step": 1852
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.688275404234031e-06,
+      "loss": 0.2507,
+      "step": 1853
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.679110136971646e-06,
+      "loss": 0.233,
+      "step": 1854
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.669945139533525e-06,
+      "loss": 0.2601,
+      "step": 1855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.66078041962616e-06,
+      "loss": 0.294,
+      "step": 1856
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.651615984955834e-06,
+      "loss": 0.296,
+      "step": 1857
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.642451843228567e-06,
+      "loss": 0.2478,
+      "step": 1858
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.633288002150151e-06,
+      "loss": 0.2531,
+      "step": 1859
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.624124469426107e-06,
+      "loss": 0.2832,
+      "step": 1860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.614961252761713e-06,
+      "loss": 0.2398,
+      "step": 1861
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.60579835986197e-06,
+      "loss": 0.2481,
+      "step": 1862
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.596635798431618e-06,
+      "loss": 0.2896,
+      "step": 1863
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.587473576175103e-06,
+      "loss": 0.2647,
+      "step": 1864
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.578311700796601e-06,
+      "loss": 0.2904,
+      "step": 1865
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.569150179999985e-06,
+      "loss": 0.2713,
+      "step": 1866
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.559989021488838e-06,
+      "loss": 0.2359,
+      "step": 1867
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.55082823296643e-06,
+      "loss": 0.2318,
+      "step": 1868
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.541667822135732e-06,
+      "loss": 0.2692,
+      "step": 1869
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.532507796699388e-06,
+      "loss": 0.2628,
+      "step": 1870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.523348164359715e-06,
+      "loss": 0.3042,
+      "step": 1871
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.514188932818711e-06,
+      "loss": 0.2245,
+      "step": 1872
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.505030109778025e-06,
+      "loss": 0.3159,
+      "step": 1873
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.495871702938974e-06,
+      "loss": 0.2428,
+      "step": 1874
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.486713720002512e-06,
+      "loss": 0.2804,
+      "step": 1875
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.47755616866925e-06,
+      "loss": 0.2693,
+      "step": 1876
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.468399056639428e-06,
+      "loss": 0.2365,
+      "step": 1877
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.459242391612919e-06,
+      "loss": 0.2598,
+      "step": 1878
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.450086181289213e-06,
+      "loss": 0.2942,
+      "step": 1879
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.440930433367436e-06,
+      "loss": 0.307,
+      "step": 1880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.4317751555463e-06,
+      "loss": 0.2432,
+      "step": 1881
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.422620355524144e-06,
+      "loss": 0.2464,
+      "step": 1882
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.413466040998894e-06,
+      "loss": 0.2404,
+      "step": 1883
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.404312219668068e-06,
+      "loss": 0.2427,
+      "step": 1884
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.395158899228777e-06,
+      "loss": 0.2658,
+      "step": 1885
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.386006087377699e-06,
+      "loss": 0.2663,
+      "step": 1886
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.376853791811095e-06,
+      "loss": 0.2538,
+      "step": 1887
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.367702020224782e-06,
+      "loss": 0.3241,
+      "step": 1888
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.35855078031415e-06,
+      "loss": 0.2437,
+      "step": 1889
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.349400079774124e-06,
+      "loss": 0.2398,
+      "step": 1890
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.340249926299192e-06,
+      "loss": 0.2069,
+      "step": 1891
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.33110032758337e-06,
+      "loss": 0.2415,
+      "step": 1892
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.321951291320222e-06,
+      "loss": 0.2659,
+      "step": 1893
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.312802825202817e-06,
+      "loss": 0.2815,
+      "step": 1894
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.30365493692377e-06,
+      "loss": 0.2667,
+      "step": 1895
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.294507634175185e-06,
+      "loss": 0.2588,
+      "step": 1896
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.285360924648695e-06,
+      "loss": 0.2723,
+      "step": 1897
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.27621481603542e-06,
+      "loss": 0.2463,
+      "step": 1898
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.267069316025988e-06,
+      "loss": 0.2618,
+      "step": 1899
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.257924432310497e-06,
+      "loss": 0.2484,
+      "step": 1900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.248780172578542e-06,
+      "loss": 0.2416,
+      "step": 1901
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.23963654451919e-06,
+      "loss": 0.2609,
+      "step": 1902
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.23049355582097e-06,
+      "loss": 0.2739,
+      "step": 1903
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.221351214171888e-06,
+      "loss": 0.2709,
+      "step": 1904
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.212209527259386e-06,
+      "loss": 0.2542,
+      "step": 1905
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.203068502770371e-06,
+      "loss": 0.2536,
+      "step": 1906
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.193928148391187e-06,
+      "loss": 0.2487,
+      "step": 1907
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.184788471807619e-06,
+      "loss": 0.2423,
+      "step": 1908
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.17564948070487e-06,
+      "loss": 0.2537,
+      "step": 1909
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.166511182767585e-06,
+      "loss": 0.239,
+      "step": 1910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.157373585679809e-06,
+      "loss": 0.2835,
+      "step": 1911
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.148236697125009e-06,
+      "loss": 0.251,
+      "step": 1912
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.139100524786046e-06,
+      "loss": 0.2491,
+      "step": 1913
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.129965076345197e-06,
+      "loss": 0.2274,
+      "step": 1914
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.120830359484104e-06,
+      "loss": 0.2324,
+      "step": 1915
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.111696381883819e-06,
+      "loss": 0.2663,
+      "step": 1916
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.102563151224757e-06,
+      "loss": 0.2854,
+      "step": 1917
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.093430675186709e-06,
+      "loss": 0.2724,
+      "step": 1918
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.084298961448836e-06,
+      "loss": 0.2623,
+      "step": 1919
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.075168017689646e-06,
+      "loss": 0.2497,
+      "step": 1920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.066037851587015e-06,
+      "loss": 0.2669,
+      "step": 1921
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.056908470818153e-06,
+      "loss": 0.2415,
+      "step": 1922
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.04777988305962e-06,
+      "loss": 0.2459,
+      "step": 1923
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.038652095987294e-06,
+      "loss": 0.2475,
+      "step": 1924
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.029525117276398e-06,
+      "loss": 0.2648,
+      "step": 1925
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.02039895460146e-06,
+      "loss": 0.3374,
+      "step": 1926
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.011273615636332e-06,
+      "loss": 0.2509,
+      "step": 1927
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.002149108054166e-06,
+      "loss": 0.2353,
+      "step": 1928
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.993025439527426e-06,
+      "loss": 0.2596,
+      "step": 1929
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.983902617727851e-06,
+      "loss": 0.2406,
+      "step": 1930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.974780650326492e-06,
+      "loss": 0.2556,
+      "step": 1931
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.96565954499366e-06,
+      "loss": 0.2633,
+      "step": 1932
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.956539309398957e-06,
+      "loss": 0.2831,
+      "step": 1933
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.947419951211248e-06,
+      "loss": 0.2681,
+      "step": 1934
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.938301478098653e-06,
+      "loss": 0.2785,
+      "step": 1935
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.929183897728561e-06,
+      "loss": 0.2573,
+      "step": 1936
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.920067217767598e-06,
+      "loss": 0.2462,
+      "step": 1937
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.910951445881646e-06,
+      "loss": 0.2492,
+      "step": 1938
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.901836589735807e-06,
+      "loss": 0.2675,
+      "step": 1939
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.89272265699443e-06,
+      "loss": 0.303,
+      "step": 1940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.883609655321072e-06,
+      "loss": 0.2839,
+      "step": 1941
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.87449759237852e-06,
+      "loss": 0.2387,
+      "step": 1942
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.865386475828764e-06,
+      "loss": 0.2518,
+      "step": 1943
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.856276313333004e-06,
+      "loss": 0.2668,
+      "step": 1944
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.847167112551628e-06,
+      "loss": 0.2617,
+      "step": 1945
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.838058881144226e-06,
+      "loss": 0.234,
+      "step": 1946
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.828951626769567e-06,
+      "loss": 0.2609,
+      "step": 1947
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.819845357085599e-06,
+      "loss": 0.3018,
+      "step": 1948
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.810740079749448e-06,
+      "loss": 0.2948,
+      "step": 1949
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.801635802417394e-06,
+      "loss": 0.295,
+      "step": 1950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.792532532744886e-06,
+      "loss": 0.2436,
+      "step": 1951
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.78343027838652e-06,
+      "loss": 0.2525,
+      "step": 1952
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.774329046996047e-06,
+      "loss": 0.2534,
+      "step": 1953
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.765228846226342e-06,
+      "loss": 0.2442,
+      "step": 1954
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.756129683729432e-06,
+      "loss": 0.2476,
+      "step": 1955
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.747031567156452e-06,
+      "loss": 0.2519,
+      "step": 1956
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.737934504157676e-06,
+      "loss": 0.3035,
+      "step": 1957
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.728838502382477e-06,
+      "loss": 0.2763,
+      "step": 1958
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.71974356947935e-06,
+      "loss": 0.2555,
+      "step": 1959
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.710649713095873e-06,
+      "loss": 0.2368,
+      "step": 1960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.701556940878736e-06,
+      "loss": 0.2476,
+      "step": 1961
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.692465260473707e-06,
+      "loss": 0.2437,
+      "step": 1962
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.683374679525644e-06,
+      "loss": 0.2775,
+      "step": 1963
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.674285205678471e-06,
+      "loss": 0.2974,
+      "step": 1964
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.665196846575187e-06,
+      "loss": 0.2368,
+      "step": 1965
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.656109609857855e-06,
+      "loss": 0.2429,
+      "step": 1966
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.647023503167588e-06,
+      "loss": 0.2707,
+      "step": 1967
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.637938534144558e-06,
+      "loss": 0.261,
+      "step": 1968
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.628854710427968e-06,
+      "loss": 0.2401,
+      "step": 1969
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.619772039656072e-06,
+      "loss": 0.2364,
+      "step": 1970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.610690529466141e-06,
+      "loss": 0.2545,
+      "step": 1971
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.601610187494485e-06,
+      "loss": 0.3241,
+      "step": 1972
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.592531021376412e-06,
+      "loss": 0.2367,
+      "step": 1973
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.583453038746265e-06,
+      "loss": 0.2524,
+      "step": 1974
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.574376247237367e-06,
+      "loss": 0.2363,
+      "step": 1975
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.565300654482061e-06,
+      "loss": 0.2685,
+      "step": 1976
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.556226268111666e-06,
+      "loss": 0.242,
+      "step": 1977
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.547153095756504e-06,
+      "loss": 0.238,
+      "step": 1978
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.538081145045851e-06,
+      "loss": 0.2922,
+      "step": 1979
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.52901042360798e-06,
+      "loss": 0.2698,
+      "step": 1980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.519940939070122e-06,
+      "loss": 0.2796,
+      "step": 1981
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.51087269905846e-06,
+      "loss": 0.2374,
+      "step": 1982
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.501805711198145e-06,
+      "loss": 0.2543,
+      "step": 1983
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.49273998311326e-06,
+      "loss": 0.2509,
+      "step": 1984
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.48367552242684e-06,
+      "loss": 0.2509,
+      "step": 1985
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.474612336760846e-06,
+      "loss": 0.2894,
+      "step": 1986
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.465550433736181e-06,
+      "loss": 0.2708,
+      "step": 1987
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.456489820972648e-06,
+      "loss": 0.2316,
+      "step": 1988
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.447430506088986e-06,
+      "loss": 0.2542,
+      "step": 1989
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.438372496702826e-06,
+      "loss": 0.2534,
+      "step": 1990
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.429315800430712e-06,
+      "loss": 0.226,
+      "step": 1991
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.420260424888081e-06,
+      "loss": 0.2441,
+      "step": 1992
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.41120637768926e-06,
+      "loss": 0.257,
+      "step": 1993
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.402153666447455e-06,
+      "loss": 0.3019,
+      "step": 1994
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.393102298774753e-06,
+      "loss": 0.2795,
+      "step": 1995
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.384052282282109e-06,
+      "loss": 0.2809,
+      "step": 1996
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.37500362457934e-06,
+      "loss": 0.239,
+      "step": 1997
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.365956333275136e-06,
+      "loss": 0.2216,
+      "step": 1998
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.356910415977008e-06,
+      "loss": 0.2539,
+      "step": 1999
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.347865880291341e-06,
+      "loss": 0.2657,
+      "step": 2000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.33882273382334e-06,
+      "loss": 0.2778,
+      "step": 2001
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.329780984177054e-06,
+      "loss": 0.2435,
+      "step": 2002
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.320740638955342e-06,
+      "loss": 0.2862,
+      "step": 2003
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.311701705759903e-06,
+      "loss": 0.2428,
+      "step": 2004
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.302664192191226e-06,
+      "loss": 0.24,
+      "step": 2005
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.293628105848623e-06,
+      "loss": 0.2271,
+      "step": 2006
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.284593454330198e-06,
+      "loss": 0.2433,
+      "step": 2007
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.27556024523286e-06,
+      "loss": 0.2437,
+      "step": 2008
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.266528486152283e-06,
+      "loss": 0.2424,
+      "step": 2009
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.257498184682942e-06,
+      "loss": 0.2779,
+      "step": 2010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.248469348418076e-06,
+      "loss": 0.2969,
+      "step": 2011
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.239441984949696e-06,
+      "loss": 0.2364,
+      "step": 2012
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.23041610186858e-06,
+      "loss": 0.2367,
+      "step": 2013
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.221391706764246e-06,
+      "loss": 0.2434,
+      "step": 2014
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.212368807224972e-06,
+      "loss": 0.2573,
+      "step": 2015
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.203347410837778e-06,
+      "loss": 0.2464,
+      "step": 2016
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.19432752518842e-06,
+      "loss": 0.2956,
+      "step": 2017
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.185309157861374e-06,
+      "loss": 0.2685,
+      "step": 2018
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.176292316439857e-06,
+      "loss": 0.2722,
+      "step": 2019
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.167277008505782e-06,
+      "loss": 0.2348,
+      "step": 2020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.158263241639794e-06,
+      "loss": 0.2417,
+      "step": 2021
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.149251023421226e-06,
+      "loss": 0.2652,
+      "step": 2022
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.140240361428118e-06,
+      "loss": 0.2365,
+      "step": 2023
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.131231263237194e-06,
+      "loss": 0.2478,
+      "step": 2024
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.122223736423871e-06,
+      "loss": 0.2589,
+      "step": 2025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.113217788562236e-06,
+      "loss": 0.3037,
+      "step": 2026
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.10421342722506e-06,
+      "loss": 0.2784,
+      "step": 2027
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.095210659983765e-06,
+      "loss": 0.2341,
+      "step": 2028
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.086209494408444e-06,
+      "loss": 0.2522,
+      "step": 2029
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.077209938067843e-06,
+      "loss": 0.2293,
+      "step": 2030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.068211998529347e-06,
+      "loss": 0.2707,
+      "step": 2031
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.059215683358992e-06,
+      "loss": 0.2677,
+      "step": 2032
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.050221000121434e-06,
+      "loss": 0.254,
+      "step": 2033
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.041227956379975e-06,
+      "loss": 0.2566,
+      "step": 2034
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.032236559696518e-06,
+      "loss": 0.2427,
+      "step": 2035
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.023246817631597e-06,
+      "loss": 0.2571,
+      "step": 2036
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.014258737744349e-06,
+      "loss": 0.2267,
+      "step": 2037
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.005272327592518e-06,
+      "loss": 0.249,
+      "step": 2038
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.996287594732428e-06,
+      "loss": 0.2123,
+      "step": 2039
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.987304546719015e-06,
+      "loss": 0.2559,
+      "step": 2040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.978323191105782e-06,
+      "loss": 0.2787,
+      "step": 2041
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.969343535444818e-06,
+      "loss": 0.2545,
+      "step": 2042
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.960365587286774e-06,
+      "loss": 0.2245,
+      "step": 2043
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.951389354180872e-06,
+      "loss": 0.2335,
+      "step": 2044
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.942414843674892e-06,
+      "loss": 0.2503,
+      "step": 2045
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.933442063315161e-06,
+      "loss": 0.2442,
+      "step": 2046
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.924471020646561e-06,
+      "loss": 0.2627,
+      "step": 2047
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.915501723212494e-06,
+      "loss": 0.2685,
+      "step": 2048
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.906534178554917e-06,
+      "loss": 0.2721,
+      "step": 2049
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.897568394214292e-06,
+      "loss": 0.2728,
+      "step": 2050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.88860437772962e-06,
+      "loss": 0.2544,
+      "step": 2051
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.8796421366384e-06,
+      "loss": 0.2508,
+      "step": 2052
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.87068167847665e-06,
+      "loss": 0.2342,
+      "step": 2053
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.86172301077888e-06,
+      "loss": 0.2383,
+      "step": 2054
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.852766141078097e-06,
+      "loss": 0.2559,
+      "step": 2055
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.8438110769058e-06,
+      "loss": 0.2978,
+      "step": 2056
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.834857825791969e-06,
+      "loss": 0.3044,
+      "step": 2057
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.825906395265049e-06,
+      "loss": 0.2326,
+      "step": 2058
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.816956792851971e-06,
+      "loss": 0.2292,
+      "step": 2059
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.808009026078113e-06,
+      "loss": 0.2455,
+      "step": 2060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.799063102467322e-06,
+      "loss": 0.2397,
+      "step": 2061
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.790119029541894e-06,
+      "loss": 0.2304,
+      "step": 2062
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.781176814822553e-06,
+      "loss": 0.2902,
+      "step": 2063
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.772236465828485e-06,
+      "loss": 0.2882,
+      "step": 2064
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.76329799007728e-06,
+      "loss": 0.2915,
+      "step": 2065
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.754361395084982e-06,
+      "loss": 0.2316,
+      "step": 2066
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.745426688366027e-06,
+      "loss": 0.2368,
+      "step": 2067
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.736493877433285e-06,
+      "loss": 0.226,
+      "step": 2068
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.727562969798014e-06,
+      "loss": 0.2444,
+      "step": 2069
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.718633972969885e-06,
+      "loss": 0.2822,
+      "step": 2070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.70970689445695e-06,
+      "loss": 0.2518,
+      "step": 2071
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.700781741765667e-06,
+      "loss": 0.2504,
+      "step": 2072
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.691858522400847e-06,
+      "loss": 0.2588,
+      "step": 2073
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.682937243865702e-06,
+      "loss": 0.2571,
+      "step": 2074
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.674017913661795e-06,
+      "loss": 0.2469,
+      "step": 2075
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.66510053928906e-06,
+      "loss": 0.2147,
+      "step": 2076
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.656185128245786e-06,
+      "loss": 0.2388,
+      "step": 2077
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.647271688028598e-06,
+      "loss": 0.2755,
+      "step": 2078
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.638360226132484e-06,
+      "loss": 0.2956,
+      "step": 2079
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.629450750050751e-06,
+      "loss": 0.2573,
+      "step": 2080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.620543267275048e-06,
+      "loss": 0.2293,
+      "step": 2081
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.6116377852953395e-06,
+      "loss": 0.2358,
+      "step": 2082
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.602734311599918e-06,
+      "loss": 0.2485,
+      "step": 2083
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.593832853675375e-06,
+      "loss": 0.2445,
+      "step": 2084
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.584933419006615e-06,
+      "loss": 0.2153,
+      "step": 2085
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.576036015076838e-06,
+      "loss": 0.2722,
+      "step": 2086
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5671406493675415e-06,
+      "loss": 0.2943,
+      "step": 2087
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.558247329358498e-06,
+      "loss": 0.2976,
+      "step": 2088
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.549356062527774e-06,
+      "loss": 0.2508,
+      "step": 2089
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.540466856351698e-06,
+      "loss": 0.23,
+      "step": 2090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5315797183048755e-06,
+      "loss": 0.2515,
+      "step": 2091
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.522694655860161e-06,
+      "loss": 0.2684,
+      "step": 2092
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.513811676488672e-06,
+      "loss": 0.2362,
+      "step": 2093
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.504930787659781e-06,
+      "loss": 0.2835,
+      "step": 2094
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.496051996841085e-06,
+      "loss": 0.2553,
+      "step": 2095
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4871753114984314e-06,
+      "loss": 0.2999,
+      "step": 2096
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.478300739095889e-06,
+      "loss": 0.2282,
+      "step": 2097
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.469428287095758e-06,
+      "loss": 0.2298,
+      "step": 2098
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.460557962958545e-06,
+      "loss": 0.2237,
+      "step": 2099
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.451689774142977e-06,
+      "loss": 0.229,
+      "step": 2100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.442823728105979e-06,
+      "loss": 0.2748,
+      "step": 2101
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.433959832302684e-06,
+      "loss": 0.2834,
+      "step": 2102
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4250980941863984e-06,
+      "loss": 0.2505,
+      "step": 2103
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.416238521208634e-06,
+      "loss": 0.2347,
+      "step": 2104
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.40738112081907e-06,
+      "loss": 0.2385,
+      "step": 2105
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.398525900465567e-06,
+      "loss": 0.2432,
+      "step": 2106
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.389672867594142e-06,
+      "loss": 0.248,
+      "step": 2107
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.38082202964898e-06,
+      "loss": 0.2642,
+      "step": 2108
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3719733940724244e-06,
+      "loss": 0.2648,
+      "step": 2109
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.363126968304954e-06,
+      "loss": 0.2689,
+      "step": 2110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3542827597852e-06,
+      "loss": 0.2869,
+      "step": 2111
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.345440775949926e-06,
+      "loss": 0.2453,
+      "step": 2112
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.336601024234028e-06,
+      "loss": 0.2209,
+      "step": 2113
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.327763512070514e-06,
+      "loss": 0.2414,
+      "step": 2114
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.318928246890524e-06,
+      "loss": 0.2407,
+      "step": 2115
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.310095236123296e-06,
+      "loss": 0.2563,
+      "step": 2116
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.301264487196185e-06,
+      "loss": 0.2607,
+      "step": 2117
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2924360075346266e-06,
+      "loss": 0.2518,
+      "step": 2118
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.283609804562169e-06,
+      "loss": 0.2703,
+      "step": 2119
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2747858857004265e-06,
+      "loss": 0.2504,
+      "step": 2120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.26596425836911e-06,
+      "loss": 0.2451,
+      "step": 2121
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.257144929985989e-06,
+      "loss": 0.2224,
+      "step": 2122
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.248327907966911e-06,
+      "loss": 0.2754,
+      "step": 2123
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.239513199725774e-06,
+      "loss": 0.2619,
+      "step": 2124
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.230700812674538e-06,
+      "loss": 0.2978,
+      "step": 2125
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.221890754223213e-06,
+      "loss": 0.2478,
+      "step": 2126
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.213083031779841e-06,
+      "loss": 0.2638,
+      "step": 2127
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.204277652750513e-06,
+      "loss": 0.2336,
+      "step": 2128
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.195474624539333e-06,
+      "loss": 0.2609,
+      "step": 2129
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.186673954548444e-06,
+      "loss": 0.2542,
+      "step": 2130
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.177875650177998e-06,
+      "loss": 0.2376,
+      "step": 2131
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.169079718826162e-06,
+      "loss": 0.2463,
+      "step": 2132
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.160286167889099e-06,
+      "loss": 0.275,
+      "step": 2133
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.151495004760981e-06,
+      "loss": 0.2739,
+      "step": 2134
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.142706236833965e-06,
+      "loss": 0.2283,
+      "step": 2135
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.133919871498203e-06,
+      "loss": 0.2053,
+      "step": 2136
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.12513591614181e-06,
+      "loss": 0.2446,
+      "step": 2137
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.116354378150895e-06,
+      "loss": 0.258,
+      "step": 2138
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.107575264909516e-06,
+      "loss": 0.2442,
+      "step": 2139
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.098798583799702e-06,
+      "loss": 0.2602,
+      "step": 2140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.09002434220144e-06,
+      "loss": 0.27,
+      "step": 2141
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0812525474926574e-06,
+      "loss": 0.249,
+      "step": 2142
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.072483207049232e-06,
+      "loss": 0.2171,
+      "step": 2143
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.063716328244965e-06,
+      "loss": 0.248,
+      "step": 2144
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.054951918451602e-06,
+      "loss": 0.236,
+      "step": 2145
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.046189985038807e-06,
+      "loss": 0.2333,
+      "step": 2146
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.037430535374162e-06,
+      "loss": 0.2541,
+      "step": 2147
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.028673576823155e-06,
+      "loss": 0.2597,
+      "step": 2148
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0199191167491894e-06,
+      "loss": 0.3106,
+      "step": 2149
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.011167162513558e-06,
+      "loss": 0.22,
+      "step": 2150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0024177214754585e-06,
+      "loss": 0.2537,
+      "step": 2151
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.993670800991961e-06,
+      "loss": 0.2435,
+      "step": 2152
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.984926408418027e-06,
+      "loss": 0.2502,
+      "step": 2153
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.976184551106484e-06,
+      "loss": 0.2371,
+      "step": 2154
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.967445236408037e-06,
+      "loss": 0.251,
+      "step": 2155
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.958708471671245e-06,
+      "loss": 0.2803,
+      "step": 2156
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.949974264242527e-06,
+      "loss": 0.2938,
+      "step": 2157
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.941242621466156e-06,
+      "loss": 0.2782,
+      "step": 2158
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.932513550684234e-06,
+      "loss": 0.2028,
+      "step": 2159
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.923787059236717e-06,
+      "loss": 0.2502,
+      "step": 2160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.915063154461379e-06,
+      "loss": 0.2433,
+      "step": 2161
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.90634184369383e-06,
+      "loss": 0.2505,
+      "step": 2162
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.897623134267488e-06,
+      "loss": 0.2514,
+      "step": 2163
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.888907033513592e-06,
+      "loss": 0.2543,
+      "step": 2164
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.880193548761184e-06,
+      "loss": 0.2663,
+      "step": 2165
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8714826873371095e-06,
+      "loss": 0.2523,
+      "step": 2166
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8627744565659995e-06,
+      "loss": 0.2341,
+      "step": 2167
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.854068863770287e-06,
+      "loss": 0.2238,
+      "step": 2168
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.845365916270169e-06,
+      "loss": 0.2309,
+      "step": 2169
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8366656213836354e-06,
+      "loss": 0.2803,
+      "step": 2170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.827967986426434e-06,
+      "loss": 0.2543,
+      "step": 2171
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8192730187120805e-06,
+      "loss": 0.2712,
+      "step": 2172
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.810580725551853e-06,
+      "loss": 0.2252,
+      "step": 2173
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.801891114254766e-06,
+      "loss": 0.2307,
+      "step": 2174
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.793204192127596e-06,
+      "loss": 0.2578,
+      "step": 2175
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.784519966474845e-06,
+      "loss": 0.2222,
+      "step": 2176
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.775838444598759e-06,
+      "loss": 0.2129,
+      "step": 2177
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7671596337992965e-06,
+      "loss": 0.208,
+      "step": 2178
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.758483541374148e-06,
+      "loss": 0.2686,
+      "step": 2179
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7498101746187135e-06,
+      "loss": 0.2901,
+      "step": 2180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.741139540826109e-06,
+      "loss": 0.2379,
+      "step": 2181
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.732471647287133e-06,
+      "loss": 0.2399,
+      "step": 2182
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.723806501290305e-06,
+      "loss": 0.2315,
+      "step": 2183
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.715144110121811e-06,
+      "loss": 0.2227,
+      "step": 2184
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.706484481065538e-06,
+      "loss": 0.2515,
+      "step": 2185
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.697827621403039e-06,
+      "loss": 0.2584,
+      "step": 2186
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.689173538413549e-06,
+      "loss": 0.2475,
+      "step": 2187
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.680522239373953e-06,
+      "loss": 0.2647,
+      "step": 2188
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6718737315588115e-06,
+      "loss": 0.2459,
+      "step": 2189
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.663228022240329e-06,
+      "loss": 0.2464,
+      "step": 2190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.654585118688356e-06,
+      "loss": 0.2259,
+      "step": 2191
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.645945028170393e-06,
+      "loss": 0.2155,
+      "step": 2192
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6373077579515586e-06,
+      "loss": 0.227,
+      "step": 2193
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.628673315294616e-06,
+      "loss": 0.2812,
+      "step": 2194
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.620041707459942e-06,
+      "loss": 0.2854,
+      "step": 2195
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.611412941705537e-06,
+      "loss": 0.21,
+      "step": 2196
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.602787025286999e-06,
+      "loss": 0.2381,
+      "step": 2197
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.594163965457545e-06,
+      "loss": 0.241,
+      "step": 2198
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.585543769467977e-06,
+      "loss": 0.2512,
+      "step": 2199
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.576926444566698e-06,
+      "loss": 0.249,
+      "step": 2200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.568311997999691e-06,
+      "loss": 0.2472,
+      "step": 2201
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.559700437010528e-06,
+      "loss": 0.2899,
+      "step": 2202
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5510917688403385e-06,
+      "loss": 0.2466,
+      "step": 2203
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.542486000727835e-06,
+      "loss": 0.2529,
+      "step": 2204
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.533883139909285e-06,
+      "loss": 0.2147,
+      "step": 2205
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.525283193618508e-06,
+      "loss": 0.2342,
+      "step": 2206
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.516686169086885e-06,
+      "loss": 0.2415,
+      "step": 2207
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.508092073543322e-06,
+      "loss": 0.2353,
+      "step": 2208
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.499500914214276e-06,
+      "loss": 0.2601,
+      "step": 2209
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.49091269832373e-06,
+      "loss": 0.2139,
+      "step": 2210
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4823274330931995e-06,
+      "loss": 0.2813,
+      "step": 2211
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4737451257417015e-06,
+      "loss": 0.2401,
+      "step": 2212
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.465165783485785e-06,
+      "loss": 0.2452,
+      "step": 2213
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.456589413539491e-06,
+      "loss": 0.2283,
+      "step": 2214
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.448016023114372e-06,
+      "loss": 0.2038,
+      "step": 2215
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.439445619419469e-06,
+      "loss": 0.2417,
+      "step": 2216
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4308782096613165e-06,
+      "loss": 0.258,
+      "step": 2217
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.422313801043922e-06,
+      "loss": 0.2825,
+      "step": 2218
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.41375240076878e-06,
+      "loss": 0.2398,
+      "step": 2219
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.40519401603485e-06,
+      "loss": 0.2324,
+      "step": 2220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3966386540385575e-06,
+      "loss": 0.2308,
+      "step": 2221
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.38808632197379e-06,
+      "loss": 0.2393,
+      "step": 2222
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.379537027031873e-06,
+      "loss": 0.2484,
+      "step": 2223
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.370990776401598e-06,
+      "loss": 0.2105,
+      "step": 2224
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.36244757726918e-06,
+      "loss": 0.2896,
+      "step": 2225
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.353907436818283e-06,
+      "loss": 0.2697,
+      "step": 2226
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.345370362229981e-06,
+      "loss": 0.2356,
+      "step": 2227
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.336836360682789e-06,
+      "loss": 0.2271,
+      "step": 2228
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3283054393526215e-06,
+      "loss": 0.2148,
+      "step": 2229
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.319777605412816e-06,
+      "loss": 0.2442,
+      "step": 2230
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.311252866034106e-06,
+      "loss": 0.2448,
+      "step": 2231
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.302731228384628e-06,
+      "loss": 0.2631,
+      "step": 2232
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.294212699629902e-06,
+      "loss": 0.2369,
+      "step": 2233
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.285697286932847e-06,
+      "loss": 0.258,
+      "step": 2234
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.277184997453748e-06,
+      "loss": 0.2542,
+      "step": 2235
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.268675838350273e-06,
+      "loss": 0.2444,
+      "step": 2236
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.26016981677746e-06,
+      "loss": 0.234,
+      "step": 2237
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.251666939887695e-06,
+      "loss": 0.2103,
+      "step": 2238
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.243167214830733e-06,
+      "loss": 0.2365,
+      "step": 2239
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.234670648753672e-06,
+      "loss": 0.2781,
+      "step": 2240
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.226177248800962e-06,
+      "loss": 0.2408,
+      "step": 2241
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.217687022114377e-06,
+      "loss": 0.2339,
+      "step": 2242
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.209199975833038e-06,
+      "loss": 0.2197,
+      "step": 2243
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.200716117093375e-06,
+      "loss": 0.2484,
+      "step": 2244
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.192235453029155e-06,
+      "loss": 0.2351,
+      "step": 2245
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.183757990771445e-06,
+      "loss": 0.2497,
+      "step": 2246
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.175283737448632e-06,
+      "loss": 0.2391,
+      "step": 2247
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1668127001863895e-06,
+      "loss": 0.274,
+      "step": 2248
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.158344886107703e-06,
+      "loss": 0.2588,
+      "step": 2249
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.149880302332833e-06,
+      "loss": 0.2724,
+      "step": 2250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.141418955979339e-06,
+      "loss": 0.2046,
+      "step": 2251
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.1329608541620426e-06,
+      "loss": 0.2267,
+      "step": 2252
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.124506003993047e-06,
+      "loss": 0.2298,
+      "step": 2253
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.116054412581721e-06,
+      "loss": 0.2227,
+      "step": 2254
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.107606087034689e-06,
+      "loss": 0.2741,
+      "step": 2255
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.099161034455834e-06,
+      "loss": 0.2233,
+      "step": 2256
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.090719261946279e-06,
+      "loss": 0.2346,
+      "step": 2257
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0822807766044e-06,
+      "loss": 0.2465,
+      "step": 2258
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.073845585525798e-06,
+      "loss": 0.2441,
+      "step": 2259
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.065413695803316e-06,
+      "loss": 0.2138,
+      "step": 2260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.056985114527005e-06,
+      "loss": 0.2135,
+      "step": 2261
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.048559848784153e-06,
+      "loss": 0.2529,
+      "step": 2262
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.040137905659238e-06,
+      "loss": 0.2556,
+      "step": 2263
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.031719292233968e-06,
+      "loss": 0.2653,
+      "step": 2264
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.023304015587231e-06,
+      "loss": 0.2186,
+      "step": 2265
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.014892082795127e-06,
+      "loss": 0.2319,
+      "step": 2266
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0064835009309255e-06,
+      "loss": 0.2348,
+      "step": 2267
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.998078277065091e-06,
+      "loss": 0.2418,
+      "step": 2268
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.9896764182652655e-06,
+      "loss": 0.249,
+      "step": 2269
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.981277931596252e-06,
+      "loss": 0.214,
+      "step": 2270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.972882824120028e-06,
+      "loss": 0.2321,
+      "step": 2271
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.96449110289572e-06,
+      "loss": 0.2902,
+      "step": 2272
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.956102774979615e-06,
+      "loss": 0.2147,
+      "step": 2273
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.947717847425141e-06,
+      "loss": 0.227,
+      "step": 2274
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9393363272828745e-06,
+      "loss": 0.2169,
+      "step": 2275
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.930958221600514e-06,
+      "loss": 0.2322,
+      "step": 2276
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.922583537422901e-06,
+      "loss": 0.2343,
+      "step": 2277
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9142122817919885e-06,
+      "loss": 0.267,
+      "step": 2278
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9058444617468566e-06,
+      "loss": 0.2549,
+      "step": 2279
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.897480084323688e-06,
+      "loss": 0.2476,
+      "step": 2280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.889119156555778e-06,
+      "loss": 0.2407,
+      "step": 2281
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.880761685473513e-06,
+      "loss": 0.2356,
+      "step": 2282
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.872407678104381e-06,
+      "loss": 0.2198,
+      "step": 2283
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.864057141472948e-06,
+      "loss": 0.2054,
+      "step": 2284
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.855710082600874e-06,
+      "loss": 0.2367,
+      "step": 2285
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8473665085068876e-06,
+      "loss": 0.2991,
+      "step": 2286
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.839026426206782e-06,
+      "loss": 0.2801,
+      "step": 2287
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.830689842713424e-06,
+      "loss": 0.2373,
+      "step": 2288
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.822356765036735e-06,
+      "loss": 0.2253,
+      "step": 2289
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.814027200183685e-06,
+      "loss": 0.235,
+      "step": 2290
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8057011551582954e-06,
+      "loss": 0.2341,
+      "step": 2291
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7973786369616236e-06,
+      "loss": 0.244,
+      "step": 2292
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.78905965259176e-06,
+      "loss": 0.2549,
+      "step": 2293
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.780744209043838e-06,
+      "loss": 0.2238,
+      "step": 2294
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.772432313309988e-06,
+      "loss": 0.2624,
+      "step": 2295
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7641239723793815e-06,
+      "loss": 0.2433,
+      "step": 2296
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.755819193238189e-06,
+      "loss": 0.2157,
+      "step": 2297
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.747517982869587e-06,
+      "loss": 0.2165,
+      "step": 2298
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.739220348253753e-06,
+      "loss": 0.2496,
+      "step": 2299
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.730926296367852e-06,
+      "loss": 0.2464,
+      "step": 2300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.72263583418605e-06,
+      "loss": 0.2731,
+      "step": 2301
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7143489686794825e-06,
+      "loss": 0.247,
+      "step": 2302
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.706065706816264e-06,
+      "loss": 0.2337,
+      "step": 2303
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.697786055561478e-06,
+      "loss": 0.2377,
+      "step": 2304
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.689510021877172e-06,
+      "loss": 0.2399,
+      "step": 2305
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.681237612722357e-06,
+      "loss": 0.2188,
+      "step": 2306
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.672968835052986e-06,
+      "loss": 0.2112,
+      "step": 2307
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.664703695821965e-06,
+      "loss": 0.2478,
+      "step": 2308
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.656442201979149e-06,
+      "loss": 0.2445,
+      "step": 2309
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.648184360471305e-06,
+      "loss": 0.2666,
+      "step": 2310
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.639930178242152e-06,
+      "loss": 0.2271,
+      "step": 2311
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.63167966223232e-06,
+      "loss": 0.2425,
+      "step": 2312
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.623432819379359e-06,
+      "loss": 0.2305,
+      "step": 2313
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.615189656617727e-06,
+      "loss": 0.2356,
+      "step": 2314
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.606950180878797e-06,
+      "loss": 0.2235,
+      "step": 2315
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.598714399090825e-06,
+      "loss": 0.2302,
+      "step": 2316
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.590482318178983e-06,
+      "loss": 0.2753,
+      "step": 2317
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.582253945065312e-06,
+      "loss": 0.2718,
+      "step": 2318
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.574029286668746e-06,
+      "loss": 0.2494,
+      "step": 2319
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.565808349905091e-06,
+      "loss": 0.2412,
+      "step": 2320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.557591141687024e-06,
+      "loss": 0.1907,
+      "step": 2321
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.549377668924087e-06,
+      "loss": 0.2186,
+      "step": 2322
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.54116793852268e-06,
+      "loss": 0.2265,
+      "step": 2323
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.532961957386067e-06,
+      "loss": 0.2559,
+      "step": 2324
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5247597324143356e-06,
+      "loss": 0.2457,
+      "step": 2325
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.516561270504439e-06,
+      "loss": 0.2319,
+      "step": 2326
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.508366578550156e-06,
+      "loss": 0.2384,
+      "step": 2327
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.500175663442093e-06,
+      "loss": 0.2281,
+      "step": 2328
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.491988532067687e-06,
+      "loss": 0.2201,
+      "step": 2329
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.483805191311187e-06,
+      "loss": 0.2102,
+      "step": 2330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.475625648053653e-06,
+      "loss": 0.2153,
+      "step": 2331
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.467449909172966e-06,
+      "loss": 0.2692,
+      "step": 2332
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.459277981543796e-06,
+      "loss": 0.2704,
+      "step": 2333
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.451109872037608e-06,
+      "loss": 0.2356,
+      "step": 2334
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.442945587522659e-06,
+      "loss": 0.2138,
+      "step": 2335
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.434785134863991e-06,
+      "loss": 0.2321,
+      "step": 2336
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.426628520923425e-06,
+      "loss": 0.2378,
+      "step": 2337
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.418475752559544e-06,
+      "loss": 0.2183,
+      "step": 2338
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.410326836627719e-06,
+      "loss": 0.2175,
+      "step": 2339
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.402181779980053e-06,
+      "loss": 0.2102,
+      "step": 2340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.394040589465431e-06,
+      "loss": 0.2552,
+      "step": 2341
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3859032719294705e-06,
+      "loss": 0.2536,
+      "step": 2342
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.377769834214535e-06,
+      "loss": 0.2382,
+      "step": 2343
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.369640283159729e-06,
+      "loss": 0.2153,
+      "step": 2344
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.361514625600888e-06,
+      "loss": 0.2267,
+      "step": 2345
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.353392868370569e-06,
+      "loss": 0.2518,
+      "step": 2346
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3452750182980625e-06,
+      "loss": 0.2454,
+      "step": 2347
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.337161082209352e-06,
+      "loss": 0.2618,
+      "step": 2348
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3290510669271535e-06,
+      "loss": 0.2438,
+      "step": 2349
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3209449792708675e-06,
+      "loss": 0.2449,
+      "step": 2350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.312842826056604e-06,
+      "loss": 0.2557,
+      "step": 2351
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3047446140971586e-06,
+      "loss": 0.2252,
+      "step": 2352
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.296650350202008e-06,
+      "loss": 0.2317,
+      "step": 2353
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2885600411773304e-06,
+      "loss": 0.2326,
+      "step": 2354
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.280473693825946e-06,
+      "loss": 0.2849,
+      "step": 2355
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2723913149473735e-06,
+      "loss": 0.2403,
+      "step": 2356
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.264312911337778e-06,
+      "loss": 0.2619,
+      "step": 2357
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.256238489789987e-06,
+      "loss": 0.199,
+      "step": 2358
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.248168057093478e-06,
+      "loss": 0.2208,
+      "step": 2359
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.240101620034377e-06,
+      "loss": 0.2561,
+      "step": 2360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2320391853954435e-06,
+      "loss": 0.216,
+      "step": 2361
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.223980759956089e-06,
+      "loss": 0.2281,
+      "step": 2362
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.215926350492324e-06,
+      "loss": 0.2457,
+      "step": 2363
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.207875963776814e-06,
+      "loss": 0.2586,
+      "step": 2364
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.199829606578823e-06,
+      "loss": 0.2518,
+      "step": 2365
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.191787285664229e-06,
+      "loss": 0.2247,
+      "step": 2366
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.183749007795522e-06,
+      "loss": 0.216,
+      "step": 2367
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.175714779731782e-06,
+      "loss": 0.2224,
+      "step": 2368
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.167684608228702e-06,
+      "loss": 0.2349,
+      "step": 2369
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.159658500038538e-06,
+      "loss": 0.2328,
+      "step": 2370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.151636461910155e-06,
+      "loss": 0.2248,
+      "step": 2371
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.143618500588978e-06,
+      "loss": 0.2423,
+      "step": 2372
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1356046228170146e-06,
+      "loss": 0.2268,
+      "step": 2373
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.127594835332831e-06,
+      "loss": 0.2302,
+      "step": 2374
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.119589144871558e-06,
+      "loss": 0.2349,
+      "step": 2375
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.111587558164877e-06,
+      "loss": 0.2253,
+      "step": 2376
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.103590081941035e-06,
+      "loss": 0.2435,
+      "step": 2377
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.095596722924793e-06,
+      "loss": 0.2578,
+      "step": 2378
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.087607487837483e-06,
+      "loss": 0.2583,
+      "step": 2379
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.079622383396938e-06,
+      "loss": 0.2353,
+      "step": 2380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.071641416317545e-06,
+      "loss": 0.1969,
+      "step": 2381
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0636645933101954e-06,
+      "loss": 0.2328,
+      "step": 2382
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0556919210823e-06,
+      "loss": 0.2249,
+      "step": 2383
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.047723406337787e-06,
+      "loss": 0.2324,
+      "step": 2384
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0397590557770695e-06,
+      "loss": 0.2299,
+      "step": 2385
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0317988760970806e-06,
+      "loss": 0.2286,
+      "step": 2386
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.023842873991236e-06,
+      "loss": 0.2483,
+      "step": 2387
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.015891056149438e-06,
+      "loss": 0.2411,
+      "step": 2388
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.007943429258073e-06,
+      "loss": 0.2245,
+      "step": 2389
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.2117,
+      "step": 2390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.992060775054554e-06,
+      "loss": 0.2265,
+      "step": 2391
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.984125761097537e-06,
+      "loss": 0.2163,
+      "step": 2392
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.97619496480119e-06,
+      "loss": 0.2557,
+      "step": 2393
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.968268392834236e-06,
+      "loss": 0.2399,
+      "step": 2394
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.960346051861827e-06,
+      "loss": 0.2381,
+      "step": 2395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.95242794854556e-06,
+      "loss": 0.2532,
+      "step": 2396
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.944514089543482e-06,
+      "loss": 0.2244,
+      "step": 2397
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.936604481510046e-06,
+      "loss": 0.2385,
+      "step": 2398
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.928699131096159e-06,
+      "loss": 0.2254,
+      "step": 2399
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.920798044949119e-06,
+      "loss": 0.221,
+      "step": 2400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.912901229712666e-06,
+      "loss": 0.2727,
+      "step": 2401
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.90500869202693e-06,
+      "loss": 0.2262,
+      "step": 2402
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.897120438528452e-06,
+      "loss": 0.2516,
+      "step": 2403
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.889236475850169e-06,
+      "loss": 0.2046,
+      "step": 2404
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.881356810621408e-06,
+      "loss": 0.235,
+      "step": 2405
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8734814494678815e-06,
+      "loss": 0.2211,
+      "step": 2406
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8656103990116964e-06,
+      "loss": 0.2319,
+      "step": 2407
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.857743665871309e-06,
+      "loss": 0.2319,
+      "step": 2408
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.849881256661574e-06,
+      "loss": 0.2357,
+      "step": 2409
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.842023177993689e-06,
+      "loss": 0.2618,
+      "step": 2410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.83416943647522e-06,
+      "loss": 0.2606,
+      "step": 2411
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.826320038710084e-06,
+      "loss": 0.2215,
+      "step": 2412
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.818474991298542e-06,
+      "loss": 0.2123,
+      "step": 2413
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.81063430083721e-06,
+      "loss": 0.2085,
+      "step": 2414
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8027979739190165e-06,
+      "loss": 0.2384,
+      "step": 2415
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.794966017133248e-06,
+      "loss": 0.2566,
+      "step": 2416
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7871384370654975e-06,
+      "loss": 0.2609,
+      "step": 2417
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.779315240297685e-06,
+      "loss": 0.2078,
+      "step": 2418
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771496433408043e-06,
+      "loss": 0.2238,
+      "step": 2419
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.763682022971113e-06,
+      "loss": 0.2238,
+      "step": 2420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.755872015557737e-06,
+      "loss": 0.2244,
+      "step": 2421
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.748066417735068e-06,
+      "loss": 0.2189,
+      "step": 2422
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.740265236066526e-06,
+      "loss": 0.2068,
+      "step": 2423
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.732468477111844e-06,
+      "loss": 0.246,
+      "step": 2424
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.724676147427019e-06,
+      "loss": 0.2664,
+      "step": 2425
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.716888253564332e-06,
+      "loss": 0.2247,
+      "step": 2426
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.709104802072327e-06,
+      "loss": 0.216,
+      "step": 2427
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.701325799495817e-06,
+      "loss": 0.2096,
+      "step": 2428
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.693551252375883e-06,
+      "loss": 0.2183,
+      "step": 2429
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.685781167249836e-06,
+      "loss": 0.2217,
+      "step": 2430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.678015550651258e-06,
+      "loss": 0.2725,
+      "step": 2431
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.670254409109965e-06,
+      "loss": 0.2088,
+      "step": 2432
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.662497749152009e-06,
+      "loss": 0.2615,
+      "step": 2433
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6547455772996745e-06,
+      "loss": 0.2386,
+      "step": 2434
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.64699790007147e-06,
+      "loss": 0.2283,
+      "step": 2435
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.639254723982127e-06,
+      "loss": 0.2061,
+      "step": 2436
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.631516055542601e-06,
+      "loss": 0.2015,
+      "step": 2437
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.623781901260034e-06,
+      "loss": 0.2232,
+      "step": 2438
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6160522676377975e-06,
+      "loss": 0.2372,
+      "step": 2439
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.608327161175448e-06,
+      "loss": 0.2636,
+      "step": 2440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.600606588368736e-06,
+      "loss": 0.2163,
+      "step": 2441
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.592890555709603e-06,
+      "loss": 0.2254,
+      "step": 2442
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.585179069686171e-06,
+      "loss": 0.2277,
+      "step": 2443
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.577472136782741e-06,
+      "loss": 0.2211,
+      "step": 2444
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.56976976347978e-06,
+      "loss": 0.2113,
+      "step": 2445
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.562071956253931e-06,
+      "loss": 0.2142,
+      "step": 2446
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.554378721577991e-06,
+      "loss": 0.2508,
+      "step": 2447
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.54669006592091e-06,
+      "loss": 0.2541,
+      "step": 2448
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.53900599574779e-06,
+      "loss": 0.2554,
+      "step": 2449
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5313265175198805e-06,
+      "loss": 0.2312,
+      "step": 2450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5236516376945604e-06,
+      "loss": 0.2275,
+      "step": 2451
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.515981362725361e-06,
+      "loss": 0.2313,
+      "step": 2452
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.508315699061915e-06,
+      "loss": 0.2396,
+      "step": 2453
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.500654653150001e-06,
+      "loss": 0.2507,
+      "step": 2454
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.492998231431502e-06,
+      "loss": 0.2272,
+      "step": 2455
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.485346440344418e-06,
+      "loss": 0.2753,
+      "step": 2456
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.477699286322852e-06,
+      "loss": 0.2421,
+      "step": 2457
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.470056775797008e-06,
+      "loss": 0.2237,
+      "step": 2458
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4624189151931895e-06,
+      "loss": 0.2163,
+      "step": 2459
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.454785710933781e-06,
+      "loss": 0.2238,
+      "step": 2460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4471571694372675e-06,
+      "loss": 0.2249,
+      "step": 2461
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.439533297118198e-06,
+      "loss": 0.239,
+      "step": 2462
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.431914100387202e-06,
+      "loss": 0.2692,
+      "step": 2463
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.424299585650978e-06,
+      "loss": 0.2257,
+      "step": 2464
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.416689759312282e-06,
+      "loss": 0.2075,
+      "step": 2465
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.409084627769932e-06,
+      "loss": 0.2301,
+      "step": 2466
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.401484197418807e-06,
+      "loss": 0.2348,
+      "step": 2467
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.39388847464981e-06,
+      "loss": 0.239,
+      "step": 2468
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3862974658499114e-06,
+      "loss": 0.1997,
+      "step": 2469
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.378711177402102e-06,
+      "loss": 0.2423,
+      "step": 2470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.371129615685406e-06,
+      "loss": 0.2754,
+      "step": 2471
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.363552787074878e-06,
+      "loss": 0.2391,
+      "step": 2472
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.355980697941584e-06,
+      "loss": 0.22,
+      "step": 2473
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.348413354652615e-06,
+      "loss": 0.2027,
+      "step": 2474
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.340850763571064e-06,
+      "loss": 0.2076,
+      "step": 2475
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3332929310560265e-06,
+      "loss": 0.2409,
+      "step": 2476
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.325739863462609e-06,
+      "loss": 0.2287,
+      "step": 2477
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.318191567141898e-06,
+      "loss": 0.222,
+      "step": 2478
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.310648048440975e-06,
+      "loss": 0.2393,
+      "step": 2479
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3031093137029e-06,
+      "loss": 0.2551,
+      "step": 2480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.295575369266711e-06,
+      "loss": 0.2248,
+      "step": 2481
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.288046221467429e-06,
+      "loss": 0.2145,
+      "step": 2482
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.280521876636018e-06,
+      "loss": 0.1995,
+      "step": 2483
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.273002341099429e-06,
+      "loss": 0.2171,
+      "step": 2484
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.265487621180553e-06,
+      "loss": 0.2604,
+      "step": 2485
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.257977723198237e-06,
+      "loss": 0.2513,
+      "step": 2486
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.250472653467273e-06,
+      "loss": 0.2431,
+      "step": 2487
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.242972418298393e-06,
+      "loss": 0.2139,
+      "step": 2488
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.235477023998264e-06,
+      "loss": 0.2388,
+      "step": 2489
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.227986476869479e-06,
+      "loss": 0.2223,
+      "step": 2490
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.220500783210559e-06,
+      "loss": 0.2133,
+      "step": 2491
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213019949315949e-06,
+      "loss": 0.2052,
+      "step": 2492
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.205543981475997e-06,
+      "loss": 0.2208,
+      "step": 2493
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1980728859769656e-06,
+      "loss": 0.2514,
+      "step": 2494
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.190606669101019e-06,
+      "loss": 0.2718,
+      "step": 2495
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.183145337126216e-06,
+      "loss": 0.2294,
+      "step": 2496
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.175688896326519e-06,
+      "loss": 0.1949,
+      "step": 2497
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.168237352971759e-06,
+      "loss": 0.2319,
+      "step": 2498
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1607907133276685e-06,
+      "loss": 0.2236,
+      "step": 2499
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.153348983655844e-06,
+      "loss": 0.2303,
+      "step": 2500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1459121702137575e-06,
+      "loss": 0.2296,
+      "step": 2501
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.138480279254746e-06,
+      "loss": 0.2599,
+      "step": 2502
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.131053317028009e-06,
+      "loss": 0.2364,
+      "step": 2503
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.123631289778601e-06,
+      "loss": 0.2246,
+      "step": 2504
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.116214203747425e-06,
+      "loss": 0.2086,
+      "step": 2505
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.108802065171229e-06,
+      "loss": 0.2026,
+      "step": 2506
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.101394880282612e-06,
+      "loss": 0.2176,
+      "step": 2507
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0939926553099855e-06,
+      "loss": 0.234,
+      "step": 2508
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.08659539647761e-06,
+      "loss": 0.2737,
+      "step": 2509
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.079203110005565e-06,
+      "loss": 0.2328,
+      "step": 2510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.07181580210974e-06,
+      "loss": 0.2104,
+      "step": 2511
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.064433479001857e-06,
+      "loss": 0.232,
+      "step": 2512
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.05705614688942e-06,
+      "loss": 0.2395,
+      "step": 2513
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0496838119757635e-06,
+      "loss": 0.2234,
+      "step": 2514
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0423164804600015e-06,
+      "loss": 0.194,
+      "step": 2515
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.034954158537049e-06,
+      "loss": 0.2639,
+      "step": 2516
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.027596852397606e-06,
+      "loss": 0.236,
+      "step": 2517
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.020244568228155e-06,
+      "loss": 0.2508,
+      "step": 2518
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.012897312210958e-06,
+      "loss": 0.2224,
+      "step": 2519
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0055550905240455e-06,
+      "loss": 0.2329,
+      "step": 2520
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.998217909341213e-06,
+      "loss": 0.2298,
+      "step": 2521
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9908857748320325e-06,
+      "loss": 0.2235,
+      "step": 2522
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.983558693161809e-06,
+      "loss": 0.2267,
+      "step": 2523
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.976236670491617e-06,
+      "loss": 0.2225,
+      "step": 2524
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9689197129782694e-06,
+      "loss": 0.2169,
+      "step": 2525
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.96160782677432e-06,
+      "loss": 0.2366,
+      "step": 2526
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.954301018028069e-06,
+      "loss": 0.2067,
+      "step": 2527
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.946999292883523e-06,
+      "loss": 0.2354,
+      "step": 2528
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.939702657480441e-06,
+      "loss": 0.2034,
+      "step": 2529
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.932411117954285e-06,
+      "loss": 0.2196,
+      "step": 2530
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.925124680436239e-06,
+      "loss": 0.2626,
+      "step": 2531
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9178433510531964e-06,
+      "loss": 0.2379,
+      "step": 2532
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.910567135927752e-06,
+      "loss": 0.2412,
+      "step": 2533
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.903296041178203e-06,
+      "loss": 0.2101,
+      "step": 2534
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.89603007291854e-06,
+      "loss": 0.2349,
+      "step": 2535
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.888769237258441e-06,
+      "loss": 0.2202,
+      "step": 2536
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.88151354030328e-06,
+      "loss": 0.2342,
+      "step": 2537
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8742629881540875e-06,
+      "loss": 0.2106,
+      "step": 2538
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.867017586907591e-06,
+      "loss": 0.2299,
+      "step": 2539
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8597773426561745e-06,
+      "loss": 0.285,
+      "step": 2540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.852542261487884e-06,
+      "loss": 0.23,
+      "step": 2541
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8453123494864375e-06,
+      "loss": 0.2192,
+      "step": 2542
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8380876127311835e-06,
+      "loss": 0.2234,
+      "step": 2543
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.830868057297145e-06,
+      "loss": 0.23,
+      "step": 2544
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.82365368925497e-06,
+      "loss": 0.2306,
+      "step": 2545
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.816444514670951e-06,
+      "loss": 0.2154,
+      "step": 2546
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.809240539607011e-06,
+      "loss": 0.2234,
+      "step": 2547
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.802041770120706e-06,
+      "loss": 0.2375,
+      "step": 2548
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7948482122652073e-06,
+      "loss": 0.2394,
+      "step": 2549
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.787659872089312e-06,
+      "loss": 0.2255,
+      "step": 2550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7804767556374212e-06,
+      "loss": 0.2173,
+      "step": 2551
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7732988689495575e-06,
+      "loss": 0.2033,
+      "step": 2552
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7661262180613246e-06,
+      "loss": 0.2094,
+      "step": 2553
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7589588090039454e-06,
+      "loss": 0.2296,
+      "step": 2554
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7517966478042222e-06,
+      "loss": 0.2662,
+      "step": 2555
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7446397404845446e-06,
+      "loss": 0.2326,
+      "step": 2556
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7374880930628987e-06,
+      "loss": 0.2227,
+      "step": 2557
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.730341711552822e-06,
+      "loss": 0.2247,
+      "step": 2558
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7232006019634493e-06,
+      "loss": 0.221,
+      "step": 2559
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.71606477029947e-06,
+      "loss": 0.1966,
+      "step": 2560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7089342225611356e-06,
+      "loss": 0.2161,
+      "step": 2561
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.701808964744258e-06,
+      "loss": 0.2083,
+      "step": 2562
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6946890028401983e-06,
+      "loss": 0.2507,
+      "step": 2563
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6875743428358656e-06,
+      "loss": 0.242,
+      "step": 2564
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6804649907137125e-06,
+      "loss": 0.2284,
+      "step": 2565
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6733609524517223e-06,
+      "loss": 0.1952,
+      "step": 2566
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6662622340234242e-06,
+      "loss": 0.2302,
+      "step": 2567
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6591688413978522e-06,
+      "loss": 0.2215,
+      "step": 2568
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.652080780539584e-06,
+      "loss": 0.2426,
+      "step": 2569
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6449980574086996e-06,
+      "loss": 0.2171,
+      "step": 2570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.637920677960798e-06,
+      "loss": 0.2078,
+      "step": 2571
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.630848648146981e-06,
+      "loss": 0.2497,
+      "step": 2572
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.623781973913849e-06,
+      "loss": 0.2138,
+      "step": 2573
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6167206612035123e-06,
+      "loss": 0.2271,
+      "step": 2574
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.609664715953559e-06,
+      "loss": 0.2061,
+      "step": 2575
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.602614144097071e-06,
+      "loss": 0.209,
+      "step": 2576
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.595568951562608e-06,
+      "loss": 0.2506,
+      "step": 2577
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5885291442742086e-06,
+      "loss": 0.2444,
+      "step": 2578
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5814947281513846e-06,
+      "loss": 0.2334,
+      "step": 2579
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5744657091091117e-06,
+      "loss": 0.2114,
+      "step": 2580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5674420930578257e-06,
+      "loss": 0.2105,
+      "step": 2581
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5604238859034313e-06,
+      "loss": 0.2096,
+      "step": 2582
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5534110935472634e-06,
+      "loss": 0.2149,
+      "step": 2583
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.546403721886126e-06,
+      "loss": 0.2025,
+      "step": 2584
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5394017768122525e-06,
+      "loss": 0.2252,
+      "step": 2585
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5324052642133145e-06,
+      "loss": 0.2622,
+      "step": 2586
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.525414189972419e-06,
+      "loss": 0.2303,
+      "step": 2587
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5184285599680947e-06,
+      "loss": 0.2312,
+      "step": 2588
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5114483800743014e-06,
+      "loss": 0.188,
+      "step": 2589
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5044736561604096e-06,
+      "loss": 0.2105,
+      "step": 2590
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4975043940912013e-06,
+      "loss": 0.2278,
+      "step": 2591
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4905405997268693e-06,
+      "loss": 0.2159,
+      "step": 2592
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4835822789230066e-06,
+      "loss": 0.2473,
+      "step": 2593
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.476629437530604e-06,
+      "loss": 0.2214,
+      "step": 2594
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4696820813960465e-06,
+      "loss": 0.2618,
+      "step": 2595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4627402163611012e-06,
+      "loss": 0.203,
+      "step": 2596
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4558038482629317e-06,
+      "loss": 0.2169,
+      "step": 2597
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4488729829340586e-06,
+      "loss": 0.2116,
+      "step": 2598
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4419476262023953e-06,
+      "loss": 0.2059,
+      "step": 2599
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4350277838912115e-06,
+      "loss": 0.2464,
+      "step": 2600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.428113461819144e-06,
+      "loss": 0.2559,
+      "step": 2601
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4212046658001864e-06,
+      "loss": 0.2455,
+      "step": 2602
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.414301401643686e-06,
+      "loss": 0.1924,
+      "step": 2603
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.407403675154337e-06,
+      "loss": 0.2209,
+      "step": 2604
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.400511492132186e-06,
+      "loss": 0.2159,
+      "step": 2605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.393624858372606e-06,
+      "loss": 0.2138,
+      "step": 2606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3867437796663106e-06,
+      "loss": 0.1913,
+      "step": 2607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3798682617993407e-06,
+      "loss": 0.2282,
+      "step": 2608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.372998310553062e-06,
+      "loss": 0.2553,
+      "step": 2609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3661339317041573e-06,
+      "loss": 0.2504,
+      "step": 2610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3592751310246252e-06,
+      "loss": 0.231,
+      "step": 2611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.352421914281783e-06,
+      "loss": 0.2288,
+      "step": 2612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.345574287238229e-06,
+      "loss": 0.2174,
+      "step": 2613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3387322556518877e-06,
+      "loss": 0.236,
+      "step": 2614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3318958252759636e-06,
+      "loss": 0.205,
+      "step": 2615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3250650018589547e-06,
+      "loss": 0.228,
+      "step": 2616
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.318239791144644e-06,
+      "loss": 0.2123,
+      "step": 2617
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.311420198872095e-06,
+      "loss": 0.2586,
+      "step": 2618
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3046062307756455e-06,
+      "loss": 0.2261,
+      "step": 2619
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.297797892584912e-06,
+      "loss": 0.2167,
+      "step": 2620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.290995190024766e-06,
+      "loss": 0.2064,
+      "step": 2621
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.284198128815348e-06,
+      "loss": 0.203,
+      "step": 2622
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.277406714672049e-06,
+      "loss": 0.2189,
+      "step": 2623
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2706209533055178e-06,
+      "loss": 0.2545,
+      "step": 2624
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2638408504216434e-06,
+      "loss": 0.2514,
+      "step": 2625
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.25706641172156e-06,
+      "loss": 0.2185,
+      "step": 2626
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2502976429016477e-06,
+      "loss": 0.2099,
+      "step": 2627
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2435345496534976e-06,
+      "loss": 0.2087,
+      "step": 2628
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.236777137663951e-06,
+      "loss": 0.2248,
+      "step": 2629
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2300254126150597e-06,
+      "loss": 0.1978,
+      "step": 2630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2232793801840957e-06,
+      "loss": 0.2168,
+      "step": 2631
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2165390460435443e-06,
+      "loss": 0.2501,
+      "step": 2632
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2098044158611007e-06,
+      "loss": 0.2466,
+      "step": 2633
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2030754952996612e-06,
+      "loss": 0.225,
+      "step": 2634
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1963522900173294e-06,
+      "loss": 0.1991,
+      "step": 2635
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1896348056673864e-06,
+      "loss": 0.2158,
+      "step": 2636
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.182923047898322e-06,
+      "loss": 0.2227,
+      "step": 2637
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.176217022353798e-06,
+      "loss": 0.226,
+      "step": 2638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.169516734672661e-06,
+      "loss": 0.2547,
+      "step": 2639
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.16282219048893e-06,
+      "loss": 0.1959,
+      "step": 2640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1561333954317952e-06,
+      "loss": 0.2603,
+      "step": 2641
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1494503551256227e-06,
+      "loss": 0.2027,
+      "step": 2642
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1427730751899176e-06,
+      "loss": 0.2132,
+      "step": 2643
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.136101561239363e-06,
+      "loss": 0.2121,
+      "step": 2644
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1294358188837838e-06,
+      "loss": 0.2001,
+      "step": 2645
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.122775853728152e-06,
+      "loss": 0.234,
+      "step": 2646
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.116121671372582e-06,
+      "loss": 0.2364,
+      "step": 2647
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.109473277412326e-06,
+      "loss": 0.2662,
+      "step": 2648
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.102830677437768e-06,
+      "loss": 0.1845,
+      "step": 2649
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.09619387703443e-06,
+      "loss": 0.2271,
+      "step": 2650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.089562881782937e-06,
+      "loss": 0.2181,
+      "step": 2651
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.082937697259052e-06,
+      "loss": 0.204,
+      "step": 2652
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.076318329033643e-06,
+      "loss": 0.2204,
+      "step": 2653
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0697047826726867e-06,
+      "loss": 0.2211,
+      "step": 2654
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.063097063737268e-06,
+      "loss": 0.2537,
+      "step": 2655
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0564951777835674e-06,
+      "loss": 0.2548,
+      "step": 2656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.049899130362872e-06,
+      "loss": 0.2224,
+      "step": 2657
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0433089270215375e-06,
+      "loss": 0.1997,
+      "step": 2658
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0367245733010298e-06,
+      "loss": 0.2121,
+      "step": 2659
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.030146074737882e-06,
+      "loss": 0.2149,
+      "step": 2660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.023573436863705e-06,
+      "loss": 0.235,
+      "step": 2661
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0170066652051865e-06,
+      "loss": 0.2425,
+      "step": 2662
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.010445765284077e-06,
+      "loss": 0.1761,
+      "step": 2663
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0038907426171883e-06,
+      "loss": 0.2521,
+      "step": 2664
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9973416027164027e-06,
+      "loss": 0.2084,
+      "step": 2665
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.990798351088635e-06,
+      "loss": 0.2281,
+      "step": 2666
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9842609932358714e-06,
+      "loss": 0.2155,
+      "step": 2667
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9777295346551195e-06,
+      "loss": 0.2116,
+      "step": 2668
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9712039808384475e-06,
+      "loss": 0.2182,
+      "step": 2669
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.964684337272946e-06,
+      "loss": 0.2303,
+      "step": 2670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.958170609440735e-06,
+      "loss": 0.2541,
+      "step": 2671
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.951662802818974e-06,
+      "loss": 0.2149,
+      "step": 2672
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9451609228798217e-06,
+      "loss": 0.2068,
+      "step": 2673
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9386649750904737e-06,
+      "loss": 0.2202,
+      "step": 2674
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9321749649131272e-06,
+      "loss": 0.2182,
+      "step": 2675
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9256908978049868e-06,
+      "loss": 0.2283,
+      "step": 2676
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9192127792182633e-06,
+      "loss": 0.1899,
+      "step": 2677
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9127406146001625e-06,
+      "loss": 0.2562,
+      "step": 2678
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.906274409392883e-06,
+      "loss": 0.2485,
+      "step": 2679
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8998141690336223e-06,
+      "loss": 0.2209,
+      "step": 2680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.893359898954544e-06,
+      "loss": 0.2179,
+      "step": 2681
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8869116045828103e-06,
+      "loss": 0.2119,
+      "step": 2682
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8804692913405474e-06,
+      "loss": 0.2128,
+      "step": 2683
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8740329646448505e-06,
+      "loss": 0.2134,
+      "step": 2684
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8676026299077974e-06,
+      "loss": 0.237,
+      "step": 2685
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.861178292536401e-06,
+      "loss": 0.2197,
+      "step": 2686
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.854759957932659e-06,
+      "loss": 0.2429,
+      "step": 2687
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8483476314934966e-06,
+      "loss": 0.2133,
+      "step": 2688
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.841941318610808e-06,
+      "loss": 0.2264,
+      "step": 2689
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8355410246714176e-06,
+      "loss": 0.2098,
+      "step": 2690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.829146755057094e-06,
+      "loss": 0.2069,
+      "step": 2691
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8227585151445367e-06,
+      "loss": 0.2143,
+      "step": 2692
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.81637631030538e-06,
+      "loss": 0.2393,
+      "step": 2693
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8100001459061764e-06,
+      "loss": 0.2599,
+      "step": 2694
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8036300273084126e-06,
+      "loss": 0.2087,
+      "step": 2695
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7972659598684703e-06,
+      "loss": 0.2206,
+      "step": 2696
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7909079489376645e-06,
+      "loss": 0.2074,
+      "step": 2697
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7845559998622064e-06,
+      "loss": 0.2266,
+      "step": 2698
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.778210117983211e-06,
+      "loss": 0.2071,
+      "step": 2699
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7718703086366926e-06,
+      "loss": 0.1902,
+      "step": 2700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7655365771535557e-06,
+      "loss": 0.2484,
+      "step": 2701
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.759208928859608e-06,
+      "loss": 0.2277,
+      "step": 2702
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.75288736907552e-06,
+      "loss": 0.237,
+      "step": 2703
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7465719031168612e-06,
+      "loss": 0.191,
+      "step": 2704
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7402625362940694e-06,
+      "loss": 0.2156,
+      "step": 2705
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7339592739124543e-06,
+      "loss": 0.212,
+      "step": 2706
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.727662121272192e-06,
+      "loss": 0.2273,
+      "step": 2707
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.721371083668324e-06,
+      "loss": 0.237,
+      "step": 2708
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7150861663907436e-06,
+      "loss": 0.2133,
+      "step": 2709
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7088073747242107e-06,
+      "loss": 0.235,
+      "step": 2710
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.702534713948316e-06,
+      "loss": 0.2025,
+      "step": 2711
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.696268189337512e-06,
+      "loss": 0.2131,
+      "step": 2712
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.690007806161082e-06,
+      "loss": 0.2153,
+      "step": 2713
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6837535696831487e-06,
+      "loss": 0.1893,
+      "step": 2714
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6775054851626634e-06,
+      "loss": 0.2364,
+      "step": 2715
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.671263557853404e-06,
+      "loss": 0.2534,
+      "step": 2716
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6650277930039826e-06,
+      "loss": 0.2514,
+      "step": 2717
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.658798195857808e-06,
+      "loss": 0.2093,
+      "step": 2718
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6525747716531237e-06,
+      "loss": 0.2123,
+      "step": 2719
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6463575256229713e-06,
+      "loss": 0.2138,
+      "step": 2720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.640146462995199e-06,
+      "loss": 0.2289,
+      "step": 2721
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.633941588992458e-06,
+      "loss": 0.2156,
+      "step": 2722
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.627742908832194e-06,
+      "loss": 0.1876,
+      "step": 2723
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6215504277266402e-06,
+      "loss": 0.2704,
+      "step": 2724
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.615364150882833e-06,
+      "loss": 0.2332,
+      "step": 2725
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6091840835025683e-06,
+      "loss": 0.2222,
+      "step": 2726
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.603010230782441e-06,
+      "loss": 0.2159,
+      "step": 2727
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.596842597913811e-06,
+      "loss": 0.1895,
+      "step": 2728
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5906811900828087e-06,
+      "loss": 0.2156,
+      "step": 2729
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.584526012470332e-06,
+      "loss": 0.2018,
+      "step": 2730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5783770702520382e-06,
+      "loss": 0.2349,
+      "step": 2731
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.572234368598342e-06,
+      "loss": 0.2429,
+      "step": 2732
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.566097912674409e-06,
+      "loss": 0.226,
+      "step": 2733
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5599677076401607e-06,
+      "loss": 0.2233,
+      "step": 2734
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.553843758650254e-06,
+      "loss": 0.2179,
+      "step": 2735
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5477260708540876e-06,
+      "loss": 0.2238,
+      "step": 2736
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.541614649395796e-06,
+      "loss": 0.1925,
+      "step": 2737
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.535509499414245e-06,
+      "loss": 0.2237,
+      "step": 2738
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.529410626043023e-06,
+      "loss": 0.2407,
+      "step": 2739
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5233180344104536e-06,
+      "loss": 0.2308,
+      "step": 2740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5172317296395552e-06,
+      "loss": 0.2206,
+      "step": 2741
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5111517168480837e-06,
+      "loss": 0.2017,
+      "step": 2742
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.505078001148492e-06,
+      "loss": 0.2033,
+      "step": 2743
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4990105876479375e-06,
+      "loss": 0.2222,
+      "step": 2744
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4929494814482834e-06,
+      "loss": 0.2065,
+      "step": 2745
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.486894687646085e-06,
+      "loss": 0.2231,
+      "step": 2746
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4808462113325914e-06,
+      "loss": 0.2099,
+      "step": 2747
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.474804057593738e-06,
+      "loss": 0.2371,
+      "step": 2748
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4687682315101515e-06,
+      "loss": 0.2358,
+      "step": 2749
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.462738738157128e-06,
+      "loss": 0.1888,
+      "step": 2750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.456715582604642e-06,
+      "loss": 0.2061,
+      "step": 2751
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.450698769917339e-06,
+      "loss": 0.2454,
+      "step": 2752
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4446883051545334e-06,
+      "loss": 0.205,
+      "step": 2753
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4386841933701955e-06,
+      "loss": 0.2534,
+      "step": 2754
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.432686439612967e-06,
+      "loss": 0.2043,
+      "step": 2755
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.426695048926123e-06,
+      "loss": 0.2706,
+      "step": 2756
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4207100263476068e-06,
+      "loss": 0.1994,
+      "step": 2757
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.414731376909998e-06,
+      "loss": 0.2098,
+      "step": 2758
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4087591056405178e-06,
+      "loss": 0.2174,
+      "step": 2759
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.402793217561026e-06,
+      "loss": 0.186,
+      "step": 2760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.396833717688015e-06,
+      "loss": 0.2195,
+      "step": 2761
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.390880611032603e-06,
+      "loss": 0.214,
+      "step": 2762
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.384933902600536e-06,
+      "loss": 0.2606,
+      "step": 2763
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3789935973921754e-06,
+      "loss": 0.1999,
+      "step": 2764
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3730597004025047e-06,
+      "loss": 0.1893,
+      "step": 2765
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3671322166211154e-06,
+      "loss": 0.213,
+      "step": 2766
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3612111510322057e-06,
+      "loss": 0.2159,
+      "step": 2767
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3552965086145764e-06,
+      "loss": 0.2402,
+      "step": 2768
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.349388294341628e-06,
+      "loss": 0.2217,
+      "step": 2769
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3434865131813635e-06,
+      "loss": 0.2302,
+      "step": 2770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.337591170096357e-06,
+      "loss": 0.2344,
+      "step": 2771
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.331702270043793e-06,
+      "loss": 0.228,
+      "step": 2772
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3258198179754233e-06,
+      "loss": 0.2247,
+      "step": 2773
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.31994381883758e-06,
+      "loss": 0.1882,
+      "step": 2774
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.314074277571171e-06,
+      "loss": 0.224,
+      "step": 2775
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.308211199111674e-06,
+      "loss": 0.2324,
+      "step": 2776
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.302354588389132e-06,
+      "loss": 0.2448,
+      "step": 2777
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2965044503281497e-06,
+      "loss": 0.2063,
+      "step": 2778
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2906607898478862e-06,
+      "loss": 0.1928,
+      "step": 2779
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2848236118620605e-06,
+      "loss": 0.2328,
+      "step": 2780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.278992921278936e-06,
+      "loss": 0.2075,
+      "step": 2781
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.273168723001321e-06,
+      "loss": 0.209,
+      "step": 2782
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.267351021926566e-06,
+      "loss": 0.2092,
+      "step": 2783
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2615398229465548e-06,
+      "loss": 0.2175,
+      "step": 2784
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.255735130947716e-06,
+      "loss": 0.2334,
+      "step": 2785
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2499369508109847e-06,
+      "loss": 0.2285,
+      "step": 2786
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.244145287411843e-06,
+      "loss": 0.2331,
+      "step": 2787
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.23836014562028e-06,
+      "loss": 0.1906,
+      "step": 2788
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2325815303008057e-06,
+      "loss": 0.2001,
+      "step": 2789
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.22680944631244e-06,
+      "loss": 0.2044,
+      "step": 2790
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2210438985087134e-06,
+      "loss": 0.2075,
+      "step": 2791
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.215284891737658e-06,
+      "loss": 0.2406,
+      "step": 2792
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2095324308418066e-06,
+      "loss": 0.2205,
+      "step": 2793
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2037865206581855e-06,
+      "loss": 0.2296,
+      "step": 2794
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.198047166018323e-06,
+      "loss": 0.2363,
+      "step": 2795
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1923143717482175e-06,
+      "loss": 0.2302,
+      "step": 2796
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1865881426683686e-06,
+      "loss": 0.2149,
+      "step": 2797
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1808684835937454e-06,
+      "loss": 0.2007,
+      "step": 2798
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1751553993337916e-06,
+      "loss": 0.2032,
+      "step": 2799
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.169448894692433e-06,
+      "loss": 0.2311,
+      "step": 2800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1637489744680464e-06,
+      "loss": 0.2181,
+      "step": 2801
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.158055643453488e-06,
+      "loss": 0.2335,
+      "step": 2802
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1523689064360644e-06,
+      "loss": 0.2014,
+      "step": 2803
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.146688768197539e-06,
+      "loss": 0.2387,
+      "step": 2804
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.141015233514128e-06,
+      "loss": 0.1945,
+      "step": 2805
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.135348307156494e-06,
+      "loss": 0.2021,
+      "step": 2806
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1296879938897417e-06,
+      "loss": 0.2115,
+      "step": 2807
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1240342984734186e-06,
+      "loss": 0.2354,
+      "step": 2808
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1183872256615e-06,
+      "loss": 0.2545,
+      "step": 2809
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1127467802024082e-06,
+      "loss": 0.2048,
+      "step": 2810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1071129668389712e-06,
+      "loss": 0.204,
+      "step": 2811
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1014857903084585e-06,
+      "loss": 0.2053,
+      "step": 2812
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0958652553425507e-06,
+      "loss": 0.2043,
+      "step": 2813
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.090251366667341e-06,
+      "loss": 0.2152,
+      "step": 2814
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.084644129003346e-06,
+      "loss": 0.214,
+      "step": 2815
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0790435470654723e-06,
+      "loss": 0.2258,
+      "step": 2816
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0734496255630476e-06,
+      "loss": 0.2354,
+      "step": 2817
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0678623691997858e-06,
+      "loss": 0.2305,
+      "step": 2818
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.062281782673804e-06,
+      "loss": 0.2251,
+      "step": 2819
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0567078706776066e-06,
+      "loss": 0.1799,
+      "step": 2820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.051140637898088e-06,
+      "loss": 0.1927,
+      "step": 2821
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0455800890165256e-06,
+      "loss": 0.215,
+      "step": 2822
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.040026228708577e-06,
+      "loss": 0.2413,
+      "step": 2823
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.034479061644271e-06,
+      "loss": 0.2222,
+      "step": 2824
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0289385924880223e-06,
+      "loss": 0.2196,
+      "step": 2825
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0234048258985927e-06,
+      "loss": 0.229,
+      "step": 2826
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.017877766529126e-06,
+      "loss": 0.2047,
+      "step": 2827
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.012357419027118e-06,
+      "loss": 0.2132,
+      "step": 2828
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0068437880344184e-06,
+      "loss": 0.1937,
+      "step": 2829
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0013368781872412e-06,
+      "loss": 0.2089,
+      "step": 2830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.995836694116131e-06,
+      "loss": 0.2246,
+      "step": 2831
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.990343240445992e-06,
+      "loss": 0.2329,
+      "step": 2832
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9848565217960636e-06,
+      "loss": 0.2238,
+      "step": 2833
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9793765427799194e-06,
+      "loss": 0.1852,
+      "step": 2834
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9739033080054706e-06,
+      "loss": 0.2228,
+      "step": 2835
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.968436822074954e-06,
+      "loss": 0.2113,
+      "step": 2836
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.962977089584932e-06,
+      "loss": 0.2044,
+      "step": 2837
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9575241151262916e-06,
+      "loss": 0.2176,
+      "step": 2838
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9520779032842286e-06,
+      "loss": 0.2351,
+      "step": 2839
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946638458638269e-06,
+      "loss": 0.2496,
+      "step": 2840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.941205785762227e-06,
+      "loss": 0.2231,
+      "step": 2841
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.935779889224241e-06,
+      "loss": 0.228,
+      "step": 2842
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9303607735867415e-06,
+      "loss": 0.2175,
+      "step": 2843
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.924948443406457e-06,
+      "loss": 0.1933,
+      "step": 2844
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9195429032344213e-06,
+      "loss": 0.2009,
+      "step": 2845
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9141441576159403e-06,
+      "loss": 0.229,
+      "step": 2846
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9087522110906243e-06,
+      "loss": 0.2391,
+      "step": 2847
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9033670681923566e-06,
+      "loss": 0.2067,
+      "step": 2848
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8979887334493018e-06,
+      "loss": 0.2295,
+      "step": 2849
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8926172113839003e-06,
+      "loss": 0.2107,
+      "step": 2850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8872525065128632e-06,
+      "loss": 0.1992,
+      "step": 2851
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8818946233471713e-06,
+      "loss": 0.2087,
+      "step": 2852
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8765435663920662e-06,
+      "loss": 0.1893,
+      "step": 2853
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8711993401470484e-06,
+      "loss": 0.2417,
+      "step": 2854
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.865861949105887e-06,
+      "loss": 0.2475,
+      "step": 2855
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.860531397756582e-06,
+      "loss": 0.2182,
+      "step": 2856
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8552076905814032e-06,
+      "loss": 0.2233,
+      "step": 2857
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8498908320568531e-06,
+      "loss": 0.1852,
+      "step": 2858
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.844580826653679e-06,
+      "loss": 0.2125,
+      "step": 2859
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.839277678836866e-06,
+      "loss": 0.2116,
+      "step": 2860
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.833981393065628e-06,
+      "loss": 0.193,
+      "step": 2861
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8286919737934216e-06,
+      "loss": 0.2406,
+      "step": 2862
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8234094254679158e-06,
+      "loss": 0.2125,
+      "step": 2863
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8181337525310084e-06,
+      "loss": 0.254,
+      "step": 2864
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8128649594188153e-06,
+      "loss": 0.193,
+      "step": 2865
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.807603050561667e-06,
+      "loss": 0.2022,
+      "step": 2866
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8023480303841058e-06,
+      "loss": 0.1853,
+      "step": 2867
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.797099903304881e-06,
+      "loss": 0.2053,
+      "step": 2868
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7918586737369427e-06,
+      "loss": 0.2425,
+      "step": 2869
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7866243460874534e-06,
+      "loss": 0.2334,
+      "step": 2870
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7813969247577524e-06,
+      "loss": 0.1961,
+      "step": 2871
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.776176414143389e-06,
+      "loss": 0.2161,
+      "step": 2872
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7709628186340932e-06,
+      "loss": 0.2172,
+      "step": 2873
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7657561426137836e-06,
+      "loss": 0.1982,
+      "step": 2874
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.760556390460556e-06,
+      "loss": 0.2009,
+      "step": 2875
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7553635665466873e-06,
+      "loss": 0.1911,
+      "step": 2876
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7501776752386312e-06,
+      "loss": 0.2325,
+      "step": 2877
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7449987208970088e-06,
+      "loss": 0.2224,
+      "step": 2878
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7398267078766084e-06,
+      "loss": 0.2351,
+      "step": 2879
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7346616405263804e-06,
+      "loss": 0.2168,
+      "step": 2880
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7295035231894386e-06,
+      "loss": 0.209,
+      "step": 2881
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7243523602030488e-06,
+      "loss": 0.2101,
+      "step": 2882
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7192081558986306e-06,
+      "loss": 0.2143,
+      "step": 2883
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.714070914601752e-06,
+      "loss": 0.227,
+      "step": 2884
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.708940640632134e-06,
+      "loss": 0.1891,
+      "step": 2885
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7038173383036205e-06,
+      "loss": 0.2387,
+      "step": 2886
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.698701011924212e-06,
+      "loss": 0.2254,
+      "step": 2887
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6935916657960328e-06,
+      "loss": 0.215,
+      "step": 2888
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6884893042153428e-06,
+      "loss": 0.197,
+      "step": 2889
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6833939314725245e-06,
+      "loss": 0.1988,
+      "step": 2890
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6783055518520864e-06,
+      "loss": 0.2173,
+      "step": 2891
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6732241696326544e-06,
+      "loss": 0.2256,
+      "step": 2892
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6681497890869757e-06,
+      "loss": 0.2446,
+      "step": 2893
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6630824144819069e-06,
+      "loss": 0.2296,
+      "step": 2894
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6580220500784117e-06,
+      "loss": 0.1958,
+      "step": 2895
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6529687001315597e-06,
+      "loss": 0.196,
+      "step": 2896
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6479223688905266e-06,
+      "loss": 0.1965,
+      "step": 2897
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6428830605985801e-06,
+      "loss": 0.2237,
+      "step": 2898
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6378507794930866e-06,
+      "loss": 0.2025,
+      "step": 2899
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6328255298055063e-06,
+      "loss": 0.2432,
+      "step": 2900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6278073157613762e-06,
+      "loss": 0.2237,
+      "step": 2901
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6227961415803307e-06,
+      "loss": 0.242,
+      "step": 2902
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6177920114760758e-06,
+      "loss": 0.2027,
+      "step": 2903
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6127949296563982e-06,
+      "loss": 0.1974,
+      "step": 2904
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6078049003231554e-06,
+      "loss": 0.2015,
+      "step": 2905
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6028219276722767e-06,
+      "loss": 0.2187,
+      "step": 2906
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5978460158937548e-06,
+      "loss": 0.2136,
+      "step": 2907
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5928771691716516e-06,
+      "loss": 0.19,
+      "step": 2908
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5879153916840828e-06,
+      "loss": 0.2378,
+      "step": 2909
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5829606876032189e-06,
+      "loss": 0.23,
+      "step": 2910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5780130610952883e-06,
+      "loss": 0.2167,
+      "step": 2911
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5730725163205618e-06,
+      "loss": 0.2065,
+      "step": 2912
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5681390574333588e-06,
+      "loss": 0.1873,
+      "step": 2913
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5632126885820387e-06,
+      "loss": 0.2008,
+      "step": 2914
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5582934139090077e-06,
+      "loss": 0.2371,
+      "step": 2915
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5533812375506886e-06,
+      "loss": 0.2158,
+      "step": 2916
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5484761636375544e-06,
+      "loss": 0.2053,
+      "step": 2917
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5435781962940966e-06,
+      "loss": 0.2008,
+      "step": 2918
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5386873396388302e-06,
+      "loss": 0.2072,
+      "step": 2919
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5338035977842948e-06,
+      "loss": 0.2069,
+      "step": 2920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.528926974837045e-06,
+      "loss": 0.2076,
+      "step": 2921
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.524057474897649e-06,
+      "loss": 0.1877,
+      "step": 2922
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.519195102060693e-06,
+      "loss": 0.2179,
+      "step": 2923
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5143398604147552e-06,
+      "loss": 0.2442,
+      "step": 2924
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5094917540424337e-06,
+      "loss": 0.2342,
+      "step": 2925
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5046507870203153e-06,
+      "loss": 0.2041,
+      "step": 2926
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4998169634189908e-06,
+      "loss": 0.205,
+      "step": 2927
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4949902873030386e-06,
+      "loss": 0.2112,
+      "step": 2928
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.490170762731028e-06,
+      "loss": 0.2144,
+      "step": 2929
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.485358393755524e-06,
+      "loss": 0.2139,
+      "step": 2930
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4805531844230591e-06,
+      "loss": 0.1929,
+      "step": 2931
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4757551387741587e-06,
+      "loss": 0.2164,
+      "step": 2932
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4709642608433184e-06,
+      "loss": 0.2261,
+      "step": 2933
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.466180554659008e-06,
+      "loss": 0.207,
+      "step": 2934
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4614040242436667e-06,
+      "loss": 0.1973,
+      "step": 2935
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4566346736136994e-06,
+      "loss": 0.1925,
+      "step": 2936
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4518725067794726e-06,
+      "loss": 0.1993,
+      "step": 2937
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4471175277453198e-06,
+      "loss": 0.225,
+      "step": 2938
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.442369740509517e-06,
+      "loss": 0.2425,
+      "step": 2939
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.437629149064307e-06,
+      "loss": 0.2196,
+      "step": 2940
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4328957573958713e-06,
+      "loss": 0.2054,
+      "step": 2941
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4281695694843444e-06,
+      "loss": 0.1986,
+      "step": 2942
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4234505893037998e-06,
+      "loss": 0.2048,
+      "step": 2943
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4187388208222474e-06,
+      "loss": 0.2028,
+      "step": 2944
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.414034268001645e-06,
+      "loss": 0.1868,
+      "step": 2945
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4093369347978658e-06,
+      "loss": 0.215,
+      "step": 2946
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.404646825160727e-06,
+      "loss": 0.2588,
+      "step": 2947
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.399963943033964e-06,
+      "loss": 0.2366,
+      "step": 2948
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.395288292355238e-06,
+      "loss": 0.2008,
+      "step": 2949
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3906198770561264e-06,
+      "loss": 0.1954,
+      "step": 2950
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3859587010621266e-06,
+      "loss": 0.2018,
+      "step": 2951
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.381304768292644e-06,
+      "loss": 0.2039,
+      "step": 2952
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3766580826610021e-06,
+      "loss": 0.2033,
+      "step": 2953
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.372018648074417e-06,
+      "loss": 0.2314,
+      "step": 2954
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3673864684340244e-06,
+      "loss": 0.2175,
+      "step": 2955
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3627615476348398e-06,
+      "loss": 0.2201,
+      "step": 2956
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.358143889565795e-06,
+      "loss": 0.2082,
+      "step": 2957
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3535334981097016e-06,
+      "loss": 0.2166,
+      "step": 2958
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3489303771432626e-06,
+      "loss": 0.1785,
+      "step": 2959
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.344334530537078e-06,
+      "loss": 0.202,
+      "step": 2960
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.2152,
+      "step": 2961
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3351646758572302e-06,
+      "loss": 0.2329,
+      "step": 2962
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3305906754941588e-06,
+      "loss": 0.2096,
+      "step": 2963
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.326023964912505e-06,
+      "loss": 0.2091,
+      "step": 2964
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3214645479522438e-06,
+      "loss": 0.2315,
+      "step": 2965
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.316912428447219e-06,
+      "loss": 0.204,
+      "step": 2966
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3123676102251337e-06,
+      "loss": 0.2055,
+      "step": 2967
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.307830097107562e-06,
+      "loss": 0.2036,
+      "step": 2968
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3032998929099217e-06,
+      "loss": 0.2204,
+      "step": 2969
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2987770014414957e-06,
+      "loss": 0.2441,
+      "step": 2970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2942614265054143e-06,
+      "loss": 0.2149,
+      "step": 2971
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2897531718986511e-06,
+      "loss": 0.2038,
+      "step": 2972
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.285252241412036e-06,
+      "loss": 0.205,
+      "step": 2973
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.280758638830224e-06,
+      "loss": 0.2027,
+      "step": 2974
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2762723679317246e-06,
+      "loss": 0.2046,
+      "step": 2975
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2717934324888682e-06,
+      "loss": 0.203,
+      "step": 2976
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2673218362678296e-06,
+      "loss": 0.2139,
+      "step": 2977
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2628575830286039e-06,
+      "loss": 0.1916,
+      "step": 2978
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2584006765250135e-06,
+      "loss": 0.223,
+      "step": 2979
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2539511205047062e-06,
+      "loss": 0.2036,
+      "step": 2980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2495089187091447e-06,
+      "loss": 0.2238,
+      "step": 2981
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.245074074873609e-06,
+      "loss": 0.1864,
+      "step": 2982
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2406465927272005e-06,
+      "loss": 0.2029,
+      "step": 2983
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2362264759928121e-06,
+      "loss": 0.2144,
+      "step": 2984
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2318137283871633e-06,
+      "loss": 0.2229,
+      "step": 2985
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2274083536207638e-06,
+      "loss": 0.2219,
+      "step": 2986
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2230103553979288e-06,
+      "loss": 0.1742,
+      "step": 2987
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.218619737416772e-06,
+      "loss": 0.2097,
+      "step": 2988
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2142365033691949e-06,
+      "loss": 0.2072,
+      "step": 2989
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2098606569409023e-06,
+      "loss": 0.1975,
+      "step": 2990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2054922018113702e-06,
+      "loss": 0.2105,
+      "step": 2991
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2011311416538773e-06,
+      "loss": 0.2034,
+      "step": 2992
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1967774801354704e-06,
+      "loss": 0.2245,
+      "step": 2993
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.192431220916983e-06,
+      "loss": 0.2202,
+      "step": 2994
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.18809236765302e-06,
+      "loss": 0.2245,
+      "step": 2995
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1837609239919612e-06,
+      "loss": 0.1782,
+      "step": 2996
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1794368935759538e-06,
+      "loss": 0.2088,
+      "step": 2997
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.175120280040918e-06,
+      "loss": 0.1992,
+      "step": 2998
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1708110870165235e-06,
+      "loss": 0.2124,
+      "step": 2999
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1665093181262178e-06,
+      "loss": 0.2297,
+      "step": 3000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1622149769871927e-06,
+      "loss": 0.2105,
+      "step": 3001
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1579280672103998e-06,
+      "loss": 0.2307,
+      "step": 3002
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1536485924005415e-06,
+      "loss": 0.2149,
+      "step": 3003
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1493765561560644e-06,
+      "loss": 0.1992,
+      "step": 3004
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1451119620691709e-06,
+      "loss": 0.1789,
+      "step": 3005
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1408548137257902e-06,
+      "loss": 0.1973,
+      "step": 3006
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1366051147056034e-06,
+      "loss": 0.2275,
+      "step": 3007
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1323628685820232e-06,
+      "loss": 0.2386,
+      "step": 3008
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1281280789221938e-06,
+      "loss": 0.2098,
+      "step": 3009
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1239007492869913e-06,
+      "loss": 0.2023,
+      "step": 3010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1196808832310202e-06,
+      "loss": 0.2067,
+      "step": 3011
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1154684843026042e-06,
+      "loss": 0.199,
+      "step": 3012
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.111263556043798e-06,
+      "loss": 0.2092,
+      "step": 3013
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1070661019903606e-06,
+      "loss": 0.1756,
+      "step": 3014
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.10287612567178e-06,
+      "loss": 0.2112,
+      "step": 3015
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0986936306112473e-06,
+      "loss": 0.2265,
+      "step": 3016
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0945186203256664e-06,
+      "loss": 0.2341,
+      "step": 3017
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0903510983256482e-06,
+      "loss": 0.2137,
+      "step": 3018
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.086191068115503e-06,
+      "loss": 0.1828,
+      "step": 3019
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.082038533193247e-06,
+      "loss": 0.2054,
+      "step": 3020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0778934970505883e-06,
+      "loss": 0.1989,
+      "step": 3021
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0737559631729356e-06,
+      "loss": 0.2152,
+      "step": 3022
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0696259350393845e-06,
+      "loss": 0.2196,
+      "step": 3023
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0655034161227218e-06,
+      "loss": 0.2117,
+      "step": 3024
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.061388409889419e-06,
+      "loss": 0.2461,
+      "step": 3025
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0572809197996291e-06,
+      "loss": 0.1952,
+      "step": 3026
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0531809493071864e-06,
+      "loss": 0.2156,
+      "step": 3027
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0490885018596075e-06,
+      "loss": 0.1791,
+      "step": 3028
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0450035808980697e-06,
+      "loss": 0.2058,
+      "step": 3029
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0409261898574374e-06,
+      "loss": 0.2218,
+      "step": 3030
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0368563321662317e-06,
+      "loss": 0.2105,
+      "step": 3031
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0327940112466461e-06,
+      "loss": 0.2616,
+      "step": 3032
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0287392305145327e-06,
+      "loss": 0.1695,
+      "step": 3033
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0246919933794042e-06,
+      "loss": 0.2264,
+      "step": 3034
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0206523032444316e-06,
+      "loss": 0.2079,
+      "step": 3035
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0166201635064354e-06,
+      "loss": 0.2252,
+      "step": 3036
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0125955775558971e-06,
+      "loss": 0.217,
+      "step": 3037
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0085785487769362e-06,
+      "loss": 0.2083,
+      "step": 3038
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.004569080547323e-06,
+      "loss": 0.2275,
+      "step": 3039
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0005671762384672e-06,
+      "loss": 0.2303,
+      "step": 3040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.965728392154229e-07,
+      "loss": 0.2065,
+      "step": 3041
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.925860728368742e-07,
+      "loss": 0.1822,
+      "step": 3042
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.886068804551507e-07,
+      "loss": 0.1944,
+      "step": 3043
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.846352654161983e-07,
+      "loss": 0.2134,
+      "step": 3044
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.806712310596033e-07,
+      "loss": 0.2165,
+      "step": 3045
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.76714780718574e-07,
+      "loss": 0.2179,
+      "step": 3046
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.7276591771994e-07,
+      "loss": 0.2053,
+      "step": 3047
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.688246453841532e-07,
+      "loss": 0.2266,
+      "step": 3048
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.648909670252804e-07,
+      "loss": 0.2145,
+      "step": 3049
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.609648859510056e-07,
+      "loss": 0.1963,
+      "step": 3050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.570464054626227e-07,
+      "loss": 0.1925,
+      "step": 3051
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.531355288550326e-07,
+      "loss": 0.2031,
+      "step": 3052
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.492322594167491e-07,
+      "loss": 0.214,
+      "step": 3053
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.453366004298848e-07,
+      "loss": 0.2243,
+      "step": 3054
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.414485551701502e-07,
+      "loss": 0.2255,
+      "step": 3055
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.375681269068604e-07,
+      "loss": 0.199,
+      "step": 3056
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.336953189029174e-07,
+      "loss": 0.1963,
+      "step": 3057
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.298301344148275e-07,
+      "loss": 0.219,
+      "step": 3058
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.259725766926708e-07,
+      "loss": 0.1884,
+      "step": 3059
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.221226489801304e-07,
+      "loss": 0.1845,
+      "step": 3060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.182803545144626e-07,
+      "loss": 0.2044,
+      "step": 3061
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.144456965265103e-07,
+      "loss": 0.2096,
+      "step": 3062
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.106186782406923e-07,
+      "loss": 0.2486,
+      "step": 3063
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.067993028750066e-07,
+      "loss": 0.2046,
+      "step": 3064
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.02987573641021e-07,
+      "loss": 0.225,
+      "step": 3065
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.991834937438759e-07,
+      "loss": 0.2073,
+      "step": 3066
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.953870663822794e-07,
+      "loss": 0.207,
+      "step": 3067
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.915982947485069e-07,
+      "loss": 0.1995,
+      "step": 3068
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.878171820283931e-07,
+      "loss": 0.204,
+      "step": 3069
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.840437314013351e-07,
+      "loss": 0.2009,
+      "step": 3070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.802779460402844e-07,
+      "loss": 0.202,
+      "step": 3071
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.765198291117482e-07,
+      "loss": 0.209,
+      "step": 3072
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.727693837757922e-07,
+      "loss": 0.2022,
+      "step": 3073
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.690266131860159e-07,
+      "loss": 0.2016,
+      "step": 3074
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.652915204895818e-07,
+      "loss": 0.2013,
+      "step": 3075
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.615641088271876e-07,
+      "loss": 0.2151,
+      "step": 3076
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.578443813330728e-07,
+      "loss": 0.2241,
+      "step": 3077
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.541323411350189e-07,
+      "loss": 0.2538,
+      "step": 3078
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.504279913543401e-07,
+      "loss": 0.1942,
+      "step": 3079
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.467313351058848e-07,
+      "loss": 0.2205,
+      "step": 3080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.430423754980344e-07,
+      "loss": 0.202,
+      "step": 3081
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.393611156326931e-07,
+      "loss": 0.2098,
+      "step": 3082
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.356875586053015e-07,
+      "loss": 0.2095,
+      "step": 3083
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.320217075048076e-07,
+      "loss": 0.21,
+      "step": 3084
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.283635654136945e-07,
+      "loss": 0.2297,
+      "step": 3085
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.247131354079541e-07,
+      "loss": 0.1949,
+      "step": 3086
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.210704205570941e-07,
+      "loss": 0.2153,
+      "step": 3087
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.174354239241445e-07,
+      "loss": 0.1878,
+      "step": 3088
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.138081485656279e-07,
+      "loss": 0.1956,
+      "step": 3089
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.101885975315915e-07,
+      "loss": 0.2146,
+      "step": 3090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.065767738655772e-07,
+      "loss": 0.2119,
+      "step": 3091
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.029726806046323e-07,
+      "loss": 0.239,
+      "step": 3092
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.993763207793048e-07,
+      "loss": 0.195,
+      "step": 3093
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.957876974136381e-07,
+      "loss": 0.2247,
+      "step": 3094
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.922068135251703e-07,
+      "loss": 0.2043,
+      "step": 3095
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.886336721249343e-07,
+      "loss": 0.2109,
+      "step": 3096
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.850682762174478e-07,
+      "loss": 0.1963,
+      "step": 3097
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.815106288007235e-07,
+      "loss": 0.1984,
+      "step": 3098
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.779607328662486e-07,
+      "loss": 0.2159,
+      "step": 3099
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.74418591399001e-07,
+      "loss": 0.2272,
+      "step": 3100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.708842073774336e-07,
+      "loss": 0.2237,
+      "step": 3101
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.673575837734759e-07,
+      "loss": 0.1978,
+      "step": 3102
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.63838723552538e-07,
+      "loss": 0.1868,
+      "step": 3103
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.603276296734918e-07,
+      "loss": 0.2047,
+      "step": 3104
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.568243050886892e-07,
+      "loss": 0.1933,
+      "step": 3105
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.533287527439425e-07,
+      "loss": 0.1823,
+      "step": 3106
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.498409755785308e-07,
+      "loss": 0.1795,
+      "step": 3107
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.463609765251955e-07,
+      "loss": 0.216,
+      "step": 3108
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.428887585101374e-07,
+      "loss": 0.2401,
+      "step": 3109
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.394243244530152e-07,
+      "loss": 0.1873,
+      "step": 3110
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.3596767726694e-07,
+      "loss": 0.1875,
+      "step": 3111
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.325188198584765e-07,
+      "loss": 0.1847,
+      "step": 3112
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.29077755127644e-07,
+      "loss": 0.195,
+      "step": 3113
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.25644485967899e-07,
+      "loss": 0.2097,
+      "step": 3114
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.222190152661524e-07,
+      "loss": 0.2218,
+      "step": 3115
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.188013459027543e-07,
+      "loss": 0.184,
+      "step": 3116
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.153914807514916e-07,
+      "loss": 0.2021,
+      "step": 3117
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.119894226795976e-07,
+      "loss": 0.2182,
+      "step": 3118
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.085951745477294e-07,
+      "loss": 0.2099,
+      "step": 3119
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.052087392099871e-07,
+      "loss": 0.1743,
+      "step": 3120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.018301195138966e-07,
+      "loss": 0.1946,
+      "step": 3121
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.984593183004118e-07,
+      "loss": 0.2053,
+      "step": 3122
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.950963384039134e-07,
+      "loss": 0.2582,
+      "step": 3123
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.917411826522058e-07,
+      "loss": 0.2316,
+      "step": 3124
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.883938538665125e-07,
+      "loss": 0.2101,
+      "step": 3125
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.850543548614774e-07,
+      "loss": 0.1734,
+      "step": 3126
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.817226884451589e-07,
+      "loss": 0.2063,
+      "step": 3127
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.783988574190347e-07,
+      "loss": 0.2185,
+      "step": 3128
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.750828645779828e-07,
+      "loss": 0.2024,
+      "step": 3129
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.717747127103035e-07,
+      "loss": 0.1862,
+      "step": 3130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.684744045976954e-07,
+      "loss": 0.2198,
+      "step": 3131
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.651819430152618e-07,
+      "loss": 0.2258,
+      "step": 3132
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.618973307315168e-07,
+      "loss": 0.2172,
+      "step": 3133
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.586205705083592e-07,
+      "loss": 0.1823,
+      "step": 3134
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.553516651011005e-07,
+      "loss": 0.1845,
+      "step": 3135
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.520906172584374e-07,
+      "loss": 0.1916,
+      "step": 3136
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.488374297224642e-07,
+      "loss": 0.2002,
+      "step": 3137
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.45592105228664e-07,
+      "loss": 0.2684,
+      "step": 3138
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.423546465059083e-07,
+      "loss": 0.1994,
+      "step": 3139
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.39125056276455e-07,
+      "loss": 0.2186,
+      "step": 3140
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.359033372559454e-07,
+      "loss": 0.2072,
+      "step": 3141
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.32689492153401e-07,
+      "loss": 0.1969,
+      "step": 3142
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.294835236712282e-07,
+      "loss": 0.202,
+      "step": 3143
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.262854345051994e-07,
+      "loss": 0.1795,
+      "step": 3144
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.230952273444724e-07,
+      "loss": 0.2082,
+      "step": 3145
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.199129048715724e-07,
+      "loss": 0.2148,
+      "step": 3146
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.167384697623935e-07,
+      "loss": 0.2336,
+      "step": 3147
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.135719246862016e-07,
+      "loss": 0.1997,
+      "step": 3148
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.104132723056211e-07,
+      "loss": 0.1964,
+      "step": 3149
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.072625152766498e-07,
+      "loss": 0.2067,
+      "step": 3150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.041196562486385e-07,
+      "loss": 0.2145,
+      "step": 3151
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.009846978643008e-07,
+      "loss": 0.2072,
+      "step": 3152
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.978576427597049e-07,
+      "loss": 0.1739,
+      "step": 3153
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.947384935642742e-07,
+      "loss": 0.2054,
+      "step": 3154
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.916272529007838e-07,
+      "loss": 0.2454,
+      "step": 3155
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.885239233853613e-07,
+      "loss": 0.2055,
+      "step": 3156
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.854285076274768e-07,
+      "loss": 0.2051,
+      "step": 3157
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.823410082299541e-07,
+      "loss": 0.197,
+      "step": 3158
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.792614277889508e-07,
+      "loss": 0.212,
+      "step": 3159
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.761897688939744e-07,
+      "loss": 0.1957,
+      "step": 3160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.731260341278666e-07,
+      "loss": 0.2304,
+      "step": 3161
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.700702260668079e-07,
+      "loss": 0.2259,
+      "step": 3162
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.670223472803105e-07,
+      "loss": 0.2096,
+      "step": 3163
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.639824003312222e-07,
+      "loss": 0.2193,
+      "step": 3164
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.609503877757228e-07,
+      "loss": 0.2086,
+      "step": 3165
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.57926312163316e-07,
+      "loss": 0.209,
+      "step": 3166
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.549101760368336e-07,
+      "loss": 0.1867,
+      "step": 3167
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.519019819324322e-07,
+      "loss": 0.1991,
+      "step": 3168
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.489017323795897e-07,
+      "loss": 0.2384,
+      "step": 3169
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.459094299011025e-07,
+      "loss": 0.2139,
+      "step": 3170
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.429250770130856e-07,
+      "loss": 0.221,
+      "step": 3171
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.399486762249695e-07,
+      "loss": 0.1795,
+      "step": 3172
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.369802300395e-07,
+      "loss": 0.194,
+      "step": 3173
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.340197409527292e-07,
+      "loss": 0.2072,
+      "step": 3174
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.310672114540239e-07,
+      "loss": 0.2135,
+      "step": 3175
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.281226440260556e-07,
+      "loss": 0.2139,
+      "step": 3176
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.251860411447984e-07,
+      "loss": 0.2001,
+      "step": 3177
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.222574052795337e-07,
+      "loss": 0.2193,
+      "step": 3178
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.193367388928417e-07,
+      "loss": 0.2222,
+      "step": 3179
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.164240444405977e-07,
+      "loss": 0.1864,
+      "step": 3180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.135193243719827e-07,
+      "loss": 0.1889,
+      "step": 3181
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.10622581129464e-07,
+      "loss": 0.2015,
+      "step": 3182
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.077338171488044e-07,
+      "loss": 0.1915,
+      "step": 3183
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.048530348590586e-07,
+      "loss": 0.2153,
+      "step": 3184
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.019802366825676e-07,
+      "loss": 0.2104,
+      "step": 3185
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.991154250349594e-07,
+      "loss": 0.1979,
+      "step": 3186
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.962586023251448e-07,
+      "loss": 0.2052,
+      "step": 3187
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.934097709553254e-07,
+      "loss": 0.202,
+      "step": 3188
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.905689333209684e-07,
+      "loss": 0.1933,
+      "step": 3189
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.877360918108342e-07,
+      "loss": 0.1894,
+      "step": 3190
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.849112488069508e-07,
+      "loss": 0.2344,
+      "step": 3191
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.820944066846223e-07,
+      "loss": 0.2021,
+      "step": 3192
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.792855678124275e-07,
+      "loss": 0.2384,
+      "step": 3193
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7648473455221054e-07,
+      "loss": 0.2134,
+      "step": 3194
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.736919092590886e-07,
+      "loss": 0.1856,
+      "step": 3195
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.709070942814453e-07,
+      "loss": 0.183,
+      "step": 3196
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.681302919609265e-07,
+      "loss": 0.2035,
+      "step": 3197
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.653615046324411e-07,
+      "loss": 0.2147,
+      "step": 3198
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6260073462415787e-07,
+      "loss": 0.2027,
+      "step": 3199
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5984798425750434e-07,
+      "loss": 0.2112,
+      "step": 3200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5710325584716777e-07,
+      "loss": 0.233,
+      "step": 3201
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5436655170108313e-07,
+      "loss": 0.2016,
+      "step": 3202
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5163787412044857e-07,
+      "loss": 0.202,
+      "step": 3203
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4891722539969983e-07,
+      "loss": 0.1761,
+      "step": 3204
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4620460782653363e-07,
+      "loss": 0.2138,
+      "step": 3205
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.435000236818876e-07,
+      "loss": 0.1874,
+      "step": 3206
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.408034752399426e-07,
+      "loss": 0.2214,
+      "step": 3207
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.381149647681282e-07,
+      "loss": 0.2109,
+      "step": 3208
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.354344945271094e-07,
+      "loss": 0.2063,
+      "step": 3209
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3276206677079437e-07,
+      "loss": 0.2067,
+      "step": 3210
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3009768374633e-07,
+      "loss": 0.1982,
+      "step": 3211
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2744134769409086e-07,
+      "loss": 0.1935,
+      "step": 3212
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2479306084769466e-07,
+      "loss": 0.2002,
+      "step": 3213
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.221528254339846e-07,
+      "loss": 0.1936,
+      "step": 3214
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.19520643673037e-07,
+      "loss": 0.2318,
+      "step": 3215
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1689651777815477e-07,
+      "loss": 0.22,
+      "step": 3216
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1428044995586394e-07,
+      "loss": 0.2229,
+      "step": 3217
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1167244240592376e-07,
+      "loss": 0.1898,
+      "step": 3218
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0907249732130337e-07,
+      "loss": 0.1928,
+      "step": 3219
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.064806168882041e-07,
+      "loss": 0.1957,
+      "step": 3220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0389680328603906e-07,
+      "loss": 0.2003,
+      "step": 3221
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0132105868744053e-07,
+      "loss": 0.2118,
+      "step": 3222
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.98753385258257e-07,
+      "loss": 0.2124,
+      "step": 3223
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.961937851575459e-07,
+      "loss": 0.2157,
+      "step": 3224
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9364226053758e-07,
+      "loss": 0.2069,
+      "step": 3225
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9109881354384426e-07,
+      "loss": 0.1901,
+      "step": 3226
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.885634463150234e-07,
+      "loss": 0.1888,
+      "step": 3227
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.860361609830143e-07,
+      "loss": 0.2019,
+      "step": 3228
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8351695967291714e-07,
+      "loss": 0.2035,
+      "step": 3229
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.810058445030329e-07,
+      "loss": 0.2065,
+      "step": 3230
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.785028175848648e-07,
+      "loss": 0.2207,
+      "step": 3231
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.760078810231138e-07,
+      "loss": 0.2074,
+      "step": 3232
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7352103691568055e-07,
+      "loss": 0.2061,
+      "step": 3233
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7104228735365563e-07,
+      "loss": 0.2209,
+      "step": 3234
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.685716344213297e-07,
+      "loss": 0.2015,
+      "step": 3235
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.66109080196182e-07,
+      "loss": 0.1909,
+      "step": 3236
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.636546267488794e-07,
+      "loss": 0.1902,
+      "step": 3237
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.612082761432822e-07,
+      "loss": 0.2182,
+      "step": 3238
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.58770030436435e-07,
+      "loss": 0.2392,
+      "step": 3239
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.563398916785654e-07,
+      "loss": 0.2106,
+      "step": 3240
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.53917861913089e-07,
+      "loss": 0.1729,
+      "step": 3241
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.515039431765954e-07,
+      "loss": 0.2056,
+      "step": 3242
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4909813749886444e-07,
+      "loss": 0.1992,
+      "step": 3243
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.467004469028412e-07,
+      "loss": 0.199,
+      "step": 3244
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4431087340465965e-07,
+      "loss": 0.2073,
+      "step": 3245
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.419294190136191e-07,
+      "loss": 0.2217,
+      "step": 3246
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3955608573219555e-07,
+      "loss": 0.2303,
+      "step": 3247
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3719087555603935e-07,
+      "loss": 0.2063,
+      "step": 3248
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.348337904739618e-07,
+      "loss": 0.2057,
+      "step": 3249
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3248483246795196e-07,
+      "loss": 0.1777,
+      "step": 3250
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.301440035131587e-07,
+      "loss": 0.193,
+      "step": 3251
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2781130557789754e-07,
+      "loss": 0.2054,
+      "step": 3252
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2548674062364724e-07,
+      "loss": 0.234,
+      "step": 3253
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2317031060504544e-07,
+      "loss": 0.2023,
+      "step": 3254
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.208620174698929e-07,
+      "loss": 0.2129,
+      "step": 3255
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1856186315914806e-07,
+      "loss": 0.2013,
+      "step": 3256
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1626984960692165e-07,
+      "loss": 0.2051,
+      "step": 3257
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.139859787404842e-07,
+      "loss": 0.2067,
+      "step": 3258
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1171025248025624e-07,
+      "loss": 0.1875,
+      "step": 3259
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0944267273981033e-07,
+      "loss": 0.1892,
+      "step": 3260
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.071832414258713e-07,
+      "loss": 0.2286,
+      "step": 3261
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0493196043830607e-07,
+      "loss": 0.2265,
+      "step": 3262
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.026888316701382e-07,
+      "loss": 0.2045,
+      "step": 3263
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.004538570075255e-07,
+      "loss": 0.1763,
+      "step": 3264
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9822703832977697e-07,
+      "loss": 0.1877,
+      "step": 3265
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.960083775093403e-07,
+      "loss": 0.2038,
+      "step": 3266
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9379787641180436e-07,
+      "loss": 0.2049,
+      "step": 3267
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.915955368958967e-07,
+      "loss": 0.1965,
+      "step": 3268
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.894013608134827e-07,
+      "loss": 0.1969,
+      "step": 3269
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.872153500095609e-07,
+      "loss": 0.2348,
+      "step": 3270
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.850375063222699e-07,
+      "loss": 0.2084,
+      "step": 3271
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8286783158287144e-07,
+      "loss": 0.2102,
+      "step": 3272
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.807063276157673e-07,
+      "loss": 0.1987,
+      "step": 3273
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7855299623848364e-07,
+      "loss": 0.2042,
+      "step": 3274
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7640783926167756e-07,
+      "loss": 0.1965,
+      "step": 3275
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.742708584891296e-07,
+      "loss": 0.2293,
+      "step": 3276
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7214205571774786e-07,
+      "loss": 0.2157,
+      "step": 3277
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.700214327375639e-07,
+      "loss": 0.1931,
+      "step": 3278
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6790899133172564e-07,
+      "loss": 0.2316,
+      "step": 3279
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6580473327651123e-07,
+      "loss": 0.2084,
+      "step": 3280
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.637086603413097e-07,
+      "loss": 0.2085,
+      "step": 3281
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6162077428863117e-07,
+      "loss": 0.184,
+      "step": 3282
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5954107687409914e-07,
+      "loss": 0.186,
+      "step": 3283
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5746956984645576e-07,
+      "loss": 0.2287,
+      "step": 3284
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5540625494755003e-07,
+      "loss": 0.2166,
+      "step": 3285
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.533511339123507e-07,
+      "loss": 0.235,
+      "step": 3286
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5130420846892877e-07,
+      "loss": 0.1858,
+      "step": 3287
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4926548033846756e-07,
+      "loss": 0.2029,
+      "step": 3288
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.47234951235259e-07,
+      "loss": 0.2112,
+      "step": 3289
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4521262286669623e-07,
+      "loss": 0.2064,
+      "step": 3290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4319849693328235e-07,
+      "loss": 0.2173,
+      "step": 3291
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4119257512861595e-07,
+      "loss": 0.2002,
+      "step": 3292
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3919485913940776e-07,
+      "loss": 0.2299,
+      "step": 3293
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3720535064545635e-07,
+      "loss": 0.2248,
+      "step": 3294
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3522405131966797e-07,
+      "loss": 0.1931,
+      "step": 3295
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3325096282804326e-07,
+      "loss": 0.1896,
+      "step": 3296
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3128608682967734e-07,
+      "loss": 0.1706,
+      "step": 3297
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.293294249767597e-07,
+      "loss": 0.1907,
+      "step": 3298
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2738097891457646e-07,
+      "loss": 0.2442,
+      "step": 3299
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2544075028150037e-07,
+      "loss": 0.222,
+      "step": 3300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2350874070900086e-07,
+      "loss": 0.1788,
+      "step": 3301
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2158495182162732e-07,
+      "loss": 0.1961,
+      "step": 3302
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1966938523702465e-07,
+      "loss": 0.2058,
+      "step": 3303
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1776204256592104e-07,
+      "loss": 0.1977,
+      "step": 3304
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1586292541212917e-07,
+      "loss": 0.1815,
+      "step": 3305
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1397203537254497e-07,
+      "loss": 0.1899,
+      "step": 3306
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1208937403714768e-07,
+      "loss": 0.2002,
+      "step": 3307
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1021494298899547e-07,
+      "loss": 0.2425,
+      "step": 3308
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0834874380422864e-07,
+      "loss": 0.2142,
+      "step": 3309
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0649077805206307e-07,
+      "loss": 0.1983,
+      "step": 3310
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.046410472947935e-07,
+      "loss": 0.1874,
+      "step": 3311
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.027995530877891e-07,
+      "loss": 0.2034,
+      "step": 3312
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0096629697949232e-07,
+      "loss": 0.2015,
+      "step": 3313
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9914128051142013e-07,
+      "loss": 0.231,
+      "step": 3314
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9732450521815938e-07,
+      "loss": 0.1908,
+      "step": 3315
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9551597262737255e-07,
+      "loss": 0.2165,
+      "step": 3316
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9371568425978093e-07,
+      "loss": 0.2312,
+      "step": 3317
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9192364162918363e-07,
+      "loss": 0.1926,
+      "step": 3318
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9013984624244087e-07,
+      "loss": 0.1874,
+      "step": 3319
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8836429959947945e-07,
+      "loss": 0.1855,
+      "step": 3320
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8659700319328955e-07,
+      "loss": 0.2,
+      "step": 3321
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.848379585099258e-07,
+      "loss": 0.2166,
+      "step": 3322
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.830871670285017e-07,
+      "loss": 0.2351,
+      "step": 3323
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8134463022119076e-07,
+      "loss": 0.1884,
+      "step": 3324
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7961034955323086e-07,
+      "loss": 0.2168,
+      "step": 3325
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7788432648291108e-07,
+      "loss": 0.1974,
+      "step": 3326
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7616656246157937e-07,
+      "loss": 0.208,
+      "step": 3327
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7445705893364028e-07,
+      "loss": 0.1963,
+      "step": 3328
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7275581733655066e-07,
+      "loss": 0.2032,
+      "step": 3329
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7106283910081955e-07,
+      "loss": 0.217,
+      "step": 3330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6937812565001155e-07,
+      "loss": 0.2216,
+      "step": 3331
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6770167840073682e-07,
+      "loss": 0.2172,
+      "step": 3332
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6603349876265885e-07,
+      "loss": 0.1895,
+      "step": 3333
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6437358813848557e-07,
+      "loss": 0.1787,
+      "step": 3334
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.62721947923975e-07,
+      "loss": 0.1959,
+      "step": 3335
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6107857950792838e-07,
+      "loss": 0.1979,
+      "step": 3336
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5944348427219368e-07,
+      "loss": 0.2078,
+      "step": 3337
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5781666359166004e-07,
+      "loss": 0.1888,
+      "step": 3338
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5619811883426206e-07,
+      "loss": 0.2305,
+      "step": 3339
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5458785136097e-07,
+      "loss": 0.2018,
+      "step": 3340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5298586252580072e-07,
+      "loss": 0.2054,
+      "step": 3341
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5139215367580562e-07,
+      "loss": 0.181,
+      "step": 3342
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4980672615107383e-07,
+      "loss": 0.1828,
+      "step": 3343
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.482295812847323e-07,
+      "loss": 0.2032,
+      "step": 3344
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4666072040294353e-07,
+      "loss": 0.1909,
+      "step": 3345
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4510014482490454e-07,
+      "loss": 0.22,
+      "step": 3346
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4354785586284225e-07,
+      "loss": 0.1947,
+      "step": 3347
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.420038548220215e-07,
+      "loss": 0.1996,
+      "step": 3348
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4046814300073263e-07,
+      "loss": 0.1934,
+      "step": 3349
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3894072169029937e-07,
+      "loss": 0.1971,
+      "step": 3350
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3742159217507213e-07,
+      "loss": 0.2012,
+      "step": 3351
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3591075573243129e-07,
+      "loss": 0.1757,
+      "step": 3352
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3440821363278067e-07,
+      "loss": 0.2204,
+      "step": 3353
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3291396713955295e-07,
+      "loss": 0.234,
+      "step": 3354
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3142801750920308e-07,
+      "loss": 0.2055,
+      "step": 3355
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.299503659912127e-07,
+      "loss": 0.1902,
+      "step": 3356
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.284810138280812e-07,
+      "loss": 0.1711,
+      "step": 3357
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2701996225533253e-07,
+      "loss": 0.2063,
+      "step": 3358
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.255672125015095e-07,
+      "loss": 0.1849,
+      "step": 3359
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2412276578817607e-07,
+      "loss": 0.2068,
+      "step": 3360
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.22686623329914e-07,
+      "loss": 0.1752,
+      "step": 3361
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2125878633431954e-07,
+      "loss": 0.2204,
+      "step": 3362
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1983925600200786e-07,
+      "loss": 0.2362,
+      "step": 3363
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.184280335266108e-07,
+      "loss": 0.2012,
+      "step": 3364
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.170251200947703e-07,
+      "loss": 0.2032,
+      "step": 3365
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.156305168861438e-07,
+      "loss": 0.1669,
+      "step": 3366
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1424422507340216e-07,
+      "loss": 0.1902,
+      "step": 3367
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1286624582222517e-07,
+      "loss": 0.2286,
+      "step": 3368
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1149658029130483e-07,
+      "loss": 0.1986,
+      "step": 3369
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1013522963233991e-07,
+      "loss": 0.2093,
+      "step": 3370
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0878219499004139e-07,
+      "loss": 0.1804,
+      "step": 3371
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0743747750212363e-07,
+      "loss": 0.1892,
+      "step": 3372
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0610107829930993e-07,
+      "loss": 0.1864,
+      "step": 3373
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0477299850532808e-07,
+      "loss": 0.1911,
+      "step": 3374
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0345323923691031e-07,
+      "loss": 0.1806,
+      "step": 3375
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.021418016037945e-07,
+      "loss": 0.2199,
+      "step": 3376
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0083868670871744e-07,
+      "loss": 0.2204,
+      "step": 3377
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.954389564741928e-08,
+      "loss": 0.2036,
+      "step": 3378
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.825742950864359e-08,
+      "loss": 0.1871,
+      "step": 3379
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.697928937412948e-08,
+      "loss": 0.1878,
+      "step": 3380
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.570947631861727e-08,
+      "loss": 0.2011,
+      "step": 3381
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.444799140984618e-08,
+      "loss": 0.1912,
+      "step": 3382
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.319483570855103e-08,
+      "loss": 0.227,
+      "step": 3383
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.195001026846451e-08,
+      "loss": 0.209,
+      "step": 3384
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.071351613631152e-08,
+      "loss": 0.2151,
+      "step": 3385
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.948535435181594e-08,
+      "loss": 0.1988,
+      "step": 3386
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.826552594769056e-08,
+      "loss": 0.199,
+      "step": 3387
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.70540319496449e-08,
+      "loss": 0.188,
+      "step": 3388
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.585087337637965e-08,
+      "loss": 0.179,
+      "step": 3389
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.465605123958331e-08,
+      "loss": 0.188,
+      "step": 3390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.346956654394e-08,
+      "loss": 0.2003,
+      "step": 3391
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.229142028711723e-08,
+      "loss": 0.244,
+      "step": 3392
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.112161345977698e-08,
+      "loss": 0.2007,
+      "step": 3393
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.996014704556465e-08,
+      "loss": 0.1925,
+      "step": 3394
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.880702202111346e-08,
+      "loss": 0.2035,
+      "step": 3395
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.766223935604223e-08,
+      "loss": 0.1833,
+      "step": 3396
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.652580001295761e-08,
+      "loss": 0.2106,
+      "step": 3397
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.539770494744858e-08,
+      "loss": 0.1772,
+      "step": 3398
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.427795510808745e-08,
+      "loss": 0.2246,
+      "step": 3399
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.316655143642881e-08,
+      "loss": 0.217,
+      "step": 3400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.206349486701291e-08,
+      "loss": 0.2202,
+      "step": 3401
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.096878632735671e-08,
+      "loss": 0.1964,
+      "step": 3402
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.988242673796053e-08,
+      "loss": 0.1853,
+      "step": 3403
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.880441701230256e-08,
+      "loss": 0.2185,
+      "step": 3404
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.773475805684105e-08,
+      "loss": 0.1911,
+      "step": 3405
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.667345077101428e-08,
+      "loss": 0.2169,
+      "step": 3406
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.562049604723175e-08,
+      "loss": 0.1962,
+      "step": 3407
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.457589477088632e-08,
+      "loss": 0.2107,
+      "step": 3408
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.353964782034428e-08,
+      "loss": 0.2126,
+      "step": 3409
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.251175606694526e-08,
+      "loss": 0.2087,
+      "step": 3410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.14922203750068e-08,
+      "loss": 0.2113,
+      "step": 3411
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.048104160181645e-08,
+      "loss": 0.1699,
+      "step": 3412
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.947822059763741e-08,
+      "loss": 0.2076,
+      "step": 3413
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.848375820570407e-08,
+      "loss": 0.2236,
+      "step": 3414
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.749765526222306e-08,
+      "loss": 0.2119,
+      "step": 3415
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.651991259637113e-08,
+      "loss": 0.2153,
+      "step": 3416
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.555053103029395e-08,
+      "loss": 0.1924,
+      "step": 3417
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.458951137911061e-08,
+      "loss": 0.1989,
+      "step": 3418
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3636854450905826e-08,
+      "loss": 0.1831,
+      "step": 3419
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.269256104673215e-08,
+      "loss": 0.2247,
+      "step": 3420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.1756631960611095e-08,
+      "loss": 0.1921,
+      "step": 3421
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.082906797953091e-08,
+      "loss": 0.1993,
+      "step": 3422
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9909869883444375e-08,
+      "loss": 0.2365,
+      "step": 3423
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8999038445270986e-08,
+      "loss": 0.2198,
+      "step": 3424
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.809657443089477e-08,
+      "loss": 0.1866,
+      "step": 3425
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.720247859916538e-08,
+      "loss": 0.1935,
+      "step": 3426
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6316751701892535e-08,
+      "loss": 0.1906,
+      "step": 3427
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5439394483850486e-08,
+      "loss": 0.2017,
+      "step": 3428
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.457040768277798e-08,
+      "loss": 0.177,
+      "step": 3429
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.370979202937275e-08,
+      "loss": 0.2214,
+      "step": 3430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.285754824729371e-08,
+      "loss": 0.2041,
+      "step": 3431
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2013677053160954e-08,
+      "loss": 0.2149,
+      "step": 3432
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.1178179156553534e-08,
+      "loss": 0.1969,
+      "step": 3433
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.035105526001171e-08,
+      "loss": 0.199,
+      "step": 3434
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9532306059032464e-08,
+      "loss": 0.1882,
+      "step": 3435
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.872193224207066e-08,
+      "loss": 0.1881,
+      "step": 3436
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.791993449053899e-08,
+      "loss": 0.2239,
+      "step": 3437
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.712631347880913e-08,
+      "loss": 0.2241,
+      "step": 3438
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.634106987420616e-08,
+      "loss": 0.1948,
+      "step": 3439
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.556420433701191e-08,
+      "loss": 0.1937,
+      "step": 3440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.479571752046274e-08,
+      "loss": 0.1942,
+      "step": 3441
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.403561007075173e-08,
+      "loss": 0.2042,
+      "step": 3442
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.328388262702431e-08,
+      "loss": 0.2343,
+      "step": 3443
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.254053582138039e-08,
+      "loss": 0.1988,
+      "step": 3444
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.18055702788711e-08,
+      "loss": 0.206,
+      "step": 3445
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.107898661750319e-08,
+      "loss": 0.211,
+      "step": 3446
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.036078544823129e-08,
+      "loss": 0.2118,
+      "step": 3447
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9650967374966753e-08,
+      "loss": 0.2004,
+      "step": 3448
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.89495329945666e-08,
+      "loss": 0.1824,
+      "step": 3449
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.825648289684124e-08,
+      "loss": 0.1995,
+      "step": 3450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.757181766455119e-08,
+      "loss": 0.1997,
+      "step": 3451
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.689553787340593e-08,
+      "loss": 0.2141,
+      "step": 3452
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6227644092062797e-08,
+      "loss": 0.2019,
+      "step": 3453
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5568136882129223e-08,
+      "loss": 0.2224,
+      "step": 3454
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.49170167981605e-08,
+      "loss": 0.2176,
+      "step": 3455
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4274284387658665e-08,
+      "loss": 0.2108,
+      "step": 3456
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.363994019107363e-08,
+      "loss": 0.1956,
+      "step": 3457
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3013984741800944e-08,
+      "loss": 0.2056,
+      "step": 3458
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2396418566184018e-08,
+      "loss": 0.1753,
+      "step": 3459
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1787242183509694e-08,
+      "loss": 0.2077,
+      "step": 3460
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1186456106012663e-08,
+      "loss": 0.2186,
+      "step": 3461
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0594060838871055e-08,
+      "loss": 0.2193,
+      "step": 3462
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0010056880208627e-08,
+      "loss": 0.187,
+      "step": 3463
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.943444472109146e-08,
+      "loss": 0.2036,
+      "step": 3464
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8867224845531275e-08,
+      "loss": 0.188,
+      "step": 3465
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8308397730479875e-08,
+      "loss": 0.2007,
+      "step": 3466
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7757963845836944e-08,
+      "loss": 0.1796,
+      "step": 3467
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7215923654438916e-08,
+      "loss": 0.1851,
+      "step": 3468
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6682277612070086e-08,
+      "loss": 0.2423,
+      "step": 3469
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.615702616745041e-08,
+      "loss": 0.2178,
+      "step": 3470
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.564016976224658e-08,
+      "loss": 0.1952,
+      "step": 3471
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5131708831062074e-08,
+      "loss": 0.1646,
+      "step": 3472
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4631643801442663e-08,
+      "loss": 0.1914,
+      "step": 3473
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4139975093875325e-08,
+      "loss": 0.2017,
+      "step": 3474
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3656703121784909e-08,
+      "loss": 0.1932,
+      "step": 3475
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.318182829153858e-08,
+      "loss": 0.222,
+      "step": 3476
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2715351002438037e-08,
+      "loss": 0.2001,
+      "step": 3477
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2257271646728408e-08,
+      "loss": 0.2197,
+      "step": 3478
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1807590609591579e-08,
+      "loss": 0.1862,
+      "step": 3479
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1366308269146198e-08,
+      "loss": 0.1949,
+      "step": 3480
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0933424996451002e-08,
+      "loss": 0.1943,
+      "step": 3481
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0508941155500386e-08,
+      "loss": 0.2012,
+      "step": 3482
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0092857103227716e-08,
+      "loss": 0.1965,
+      "step": 3483
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.685173189503127e-09,
+      "loss": 0.2239,
+      "step": 3484
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.285889757131294e-09,
+      "loss": 0.2078,
+      "step": 3485
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.895007141855872e-09,
+      "loss": 0.1835,
+      "step": 3486
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.512525672353944e-09,
+      "loss": 0.1913,
+      "step": 3487
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.13844567024158e-09,
+      "loss": 0.2036,
+      "step": 3488
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.772767450067164e-09,
+      "loss": 0.2014,
+      "step": 3489
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.415491319316959e-09,
+      "loss": 0.2071,
+      "step": 3490
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.066617578410651e-09,
+      "loss": 0.2051,
+      "step": 3491
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.72614652070247e-09,
+      "loss": 0.2165,
+      "step": 3492
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.394078432482298e-09,
+      "loss": 0.2245,
+      "step": 3493
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.070413592974556e-09,
+      "loss": 0.1947,
+      "step": 3494
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.755152274334874e-09,
+      "loss": 0.1907,
+      "step": 3495
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.448294741656757e-09,
+      "loss": 0.1838,
+      "step": 3496
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.149841252964915e-09,
+      "loss": 0.1788,
+      "step": 3497
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.8597920592152735e-09,
+      "loss": 0.2132,
+      "step": 3498
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.578147404302735e-09,
+      "loss": 0.1819,
+      "step": 3499
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3049075250500835e-09,
+      "loss": 0.2069,
+      "step": 3500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.040072651213534e-09,
+      "loss": 0.2229,
+      "step": 3501
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.783643005484949e-09,
+      "loss": 0.1903,
+      "step": 3502
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.5356188034851856e-09,
+      "loss": 0.1952,
+      "step": 3503
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.2960002537685276e-09,
+      "loss": 0.1949,
+      "step": 3504
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.06478755782047e-09,
+      "loss": 0.181,
+      "step": 3505
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.8419809100610487e-09,
+      "loss": 0.201,
+      "step": 3506
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6275804978392883e-09,
+      "loss": 0.2137,
+      "step": 3507
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4215865014354245e-09,
+      "loss": 0.2198,
+      "step": 3508
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2239990940620127e-09,
+      "loss": 0.1846,
+      "step": 3509
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0348184418650386e-09,
+      "loss": 0.2059,
+      "step": 3510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.854044703917257e-09,
+      "loss": 0.1974,
+      "step": 3511
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.681678032225964e-09,
+      "loss": 0.2047,
+      "step": 3512
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5177185717263343e-09,
+      "loss": 0.204,
+      "step": 3513
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3621664602880834e-09,
+      "loss": 0.1973,
+      "step": 3514
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2150218287065862e-09,
+      "loss": 0.2447,
+      "step": 3515
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0762848007106474e-09,
+      "loss": 0.2234,
+      "step": 3516
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.459554929613923e-10,
+      "loss": 0.2047,
+      "step": 3517
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.240340150444948e-10,
+      "loss": 0.2014,
+      "step": 3518
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.105204694823897e-10,
+      "loss": 0.1848,
+      "step": 3519
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.054149517220609e-10,
+      "loss": 0.1867,
+      "step": 3520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.087175501439223e-10,
+      "loss": 0.194,
+      "step": 3521
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.204283460562675e-10,
+      "loss": 0.2332,
+      "step": 3522
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.405474136985998e-10,
+      "loss": 0.1735,
+      "step": 3523
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6907482024052246e-10,
+      "loss": 0.2194,
+      "step": 3524
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0601062578062824e-10,
+      "loss": 0.1867,
+      "step": 3525
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5135488334538927e-10,
+      "loss": 0.2032,
+      "step": 3526
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.051076388947081e-10,
+      "loss": 0.1857,
+      "step": 3527
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.726893131525636e-11,
+      "loss": 0.1722,
+      "step": 3528
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.783879242491573e-11,
+      "loss": 0.2123,
+      "step": 3529
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.68172469694472e-11,
+      "loss": 0.205,
+      "step": 3530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2043126258217714e-12,
+      "loss": 0.2258,
+      "step": 3531
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.1684,
+      "step": 3532
+    },
+    {
+      "epoch": 2.0,
+      "step": 3532,
+      "total_flos": 0.0,
+      "train_loss": 0.05549156494092698,
+      "train_runtime": 5189.8356,
+      "train_samples_per_second": 709.847,
+      "train_steps_per_second": 0.681
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3532,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 100,
+  "total_flos": 0.0,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}