diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,21072 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 3507,
+  "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": 1.6055,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.773584905660378e-07,
+      "loss": 1.6064,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.660377358490567e-07,
+      "loss": 1.6006,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.547169811320755e-07,
+      "loss": 1.6094,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.433962264150944e-07,
+      "loss": 1.6123,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1320754716981133e-06,
+      "loss": 1.6064,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3207547169811322e-06,
+      "loss": 1.6055,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.509433962264151e-06,
+      "loss": 1.6025,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6981132075471698e-06,
+      "loss": 1.5986,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8867924528301889e-06,
+      "loss": 1.6035,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.075471698113208e-06,
+      "loss": 1.6006,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2641509433962266e-06,
+      "loss": 1.6084,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4528301886792453e-06,
+      "loss": 1.6045,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6415094339622644e-06,
+      "loss": 1.5918,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 1.6055,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.018867924528302e-06,
+      "loss": 1.5928,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.207547169811321e-06,
+      "loss": 1.585,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3962264150943395e-06,
+      "loss": 1.6016,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5849056603773586e-06,
+      "loss": 1.5977,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7735849056603777e-06,
+      "loss": 1.585,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.962264150943396e-06,
+      "loss": 1.6045,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.150943396226416e-06,
+      "loss": 1.5811,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.339622641509435e-06,
+      "loss": 1.6123,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.528301886792453e-06,
+      "loss": 1.5908,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.716981132075472e-06,
+      "loss": 1.5947,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.905660377358491e-06,
+      "loss": 1.584,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.09433962264151e-06,
+      "loss": 1.6084,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.283018867924529e-06,
+      "loss": 1.5898,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4716981132075475e-06,
+      "loss": 1.5928,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 1.584,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.849056603773585e-06,
+      "loss": 1.5742,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.037735849056604e-06,
+      "loss": 1.5957,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.226415094339623e-06,
+      "loss": 1.6055,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.415094339622642e-06,
+      "loss": 1.582,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.60377358490566e-06,
+      "loss": 1.5918,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.792452830188679e-06,
+      "loss": 1.5498,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.981132075471699e-06,
+      "loss": 1.5742,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.169811320754717e-06,
+      "loss": 1.5908,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.358490566037736e-06,
+      "loss": 1.5781,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5471698113207555e-06,
+      "loss": 1.5869,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.735849056603775e-06,
+      "loss": 1.5801,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.924528301886793e-06,
+      "loss": 1.583,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.113207547169812e-06,
+      "loss": 1.5557,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.301886792452832e-06,
+      "loss": 1.5791,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.49056603773585e-06,
+      "loss": 1.5479,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.67924528301887e-06,
+      "loss": 1.5469,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.867924528301887e-06,
+      "loss": 1.5576,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.056603773584907e-06,
+      "loss": 1.5566,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.245283018867926e-06,
+      "loss": 1.5547,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 1.5508,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.622641509433963e-06,
+      "loss": 1.5664,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.811320754716981e-06,
+      "loss": 1.54,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.5723,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.018867924528302e-05,
+      "loss": 1.541,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0377358490566038e-05,
+      "loss": 1.5439,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0566037735849058e-05,
+      "loss": 1.5156,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0754716981132076e-05,
+      "loss": 1.542,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0943396226415095e-05,
+      "loss": 1.543,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1132075471698115e-05,
+      "loss": 1.5518,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 1.5518,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1509433962264152e-05,
+      "loss": 1.5195,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.169811320754717e-05,
+      "loss": 1.5381,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.188679245283019e-05,
+      "loss": 1.54,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2075471698113209e-05,
+      "loss": 1.543,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2264150943396227e-05,
+      "loss": 1.501,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2452830188679246e-05,
+      "loss": 1.498,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2641509433962264e-05,
+      "loss": 1.5078,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2830188679245283e-05,
+      "loss": 1.5371,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3018867924528303e-05,
+      "loss": 1.5342,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.320754716981132e-05,
+      "loss": 1.541,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.339622641509434e-05,
+      "loss": 1.5283,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3584905660377358e-05,
+      "loss": 1.501,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3773584905660378e-05,
+      "loss": 1.498,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3962264150943397e-05,
+      "loss": 1.5137,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4150943396226415e-05,
+      "loss": 1.5244,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4339622641509435e-05,
+      "loss": 1.501,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4528301886792452e-05,
+      "loss": 1.4814,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4716981132075472e-05,
+      "loss": 1.4717,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4905660377358491e-05,
+      "loss": 1.5215,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5094339622641511e-05,
+      "loss": 1.5303,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5283018867924532e-05,
+      "loss": 1.5371,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.547169811320755e-05,
+      "loss": 1.4238,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5660377358490568e-05,
+      "loss": 1.4746,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5849056603773586e-05,
+      "loss": 1.4639,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6037735849056607e-05,
+      "loss": 1.4863,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6226415094339625e-05,
+      "loss": 1.4512,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6415094339622643e-05,
+      "loss": 1.4883,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6603773584905664e-05,
+      "loss": 1.4707,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.679245283018868e-05,
+      "loss": 1.4697,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 1.4512,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.716981132075472e-05,
+      "loss": 1.4648,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.735849056603774e-05,
+      "loss": 1.418,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7547169811320756e-05,
+      "loss": 1.5,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7735849056603774e-05,
+      "loss": 1.417,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7924528301886795e-05,
+      "loss": 1.4209,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8113207547169813e-05,
+      "loss": 1.3867,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.830188679245283e-05,
+      "loss": 1.3896,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8490566037735852e-05,
+      "loss": 1.3818,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.867924528301887e-05,
+      "loss": 1.4102,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 1.4316,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.905660377358491e-05,
+      "loss": 1.3984,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9245283018867927e-05,
+      "loss": 1.4209,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9433962264150945e-05,
+      "loss": 1.3877,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9622641509433963e-05,
+      "loss": 1.4189,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9811320754716984e-05,
+      "loss": 1.4189,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.3906,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999995733650257e-05,
+      "loss": 1.3447,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999982934604664e-05,
+      "loss": 1.3994,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996160287414e-05,
+      "loss": 1.4258,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993173847689e-05,
+      "loss": 1.3535,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999893341438394e-05,
+      "loss": 1.3545,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999984641179142e-05,
+      "loss": 1.2773,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999790949576007e-05,
+      "loss": 1.4189,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999726954839478e-05,
+      "loss": 1.3027,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999965442763644e-05,
+      "loss": 1.3779,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999573368028785e-05,
+      "loss": 1.2617,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999483776085665e-05,
+      "loss": 1.2705,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999938565188354e-05,
+      "loss": 1.2383,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999278995506124e-05,
+      "loss": 1.3164,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999916380704443e-05,
+      "loss": 1.2695,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999040086596748e-05,
+      "loss": 1.2646,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999890783426864e-05,
+      "loss": 1.291,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998767050172955e-05,
+      "loss": 1.3682,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998617734429815e-05,
+      "loss": 1.2832,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998459887166635e-05,
+      "loss": 1.3867,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998293508518096e-05,
+      "loss": 1.418,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998118598626163e-05,
+      "loss": 1.2188,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997935157640085e-05,
+      "loss": 1.2393,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997743185716386e-05,
+      "loss": 1.2471,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999754268301887e-05,
+      "loss": 1.3359,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997333649718614e-05,
+      "loss": 1.4004,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997116085993986e-05,
+      "loss": 1.3096,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996889992030627e-05,
+      "loss": 1.2178,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996655368021455e-05,
+      "loss": 1.2803,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996412214166667e-05,
+      "loss": 1.3594,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996160530673735e-05,
+      "loss": 1.1963,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995900317757423e-05,
+      "loss": 1.1768,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995631575639752e-05,
+      "loss": 1.165,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995354304550038e-05,
+      "loss": 1.1377,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995068504724863e-05,
+      "loss": 1.0972,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99947741764081e-05,
+      "loss": 1.3047,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999447131985088e-05,
+      "loss": 1.2139,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994159935311633e-05,
+      "loss": 1.3682,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993840023056045e-05,
+      "loss": 1.1216,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993511583357087e-05,
+      "loss": 1.1797,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993174616495013e-05,
+      "loss": 1.0957,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992829122757343e-05,
+      "loss": 1.2637,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992475102438878e-05,
+      "loss": 1.3638,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999211255584169e-05,
+      "loss": 1.1318,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991741483275132e-05,
+      "loss": 1.1816,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999136188505583e-05,
+      "loss": 1.2407,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999097376150768e-05,
+      "loss": 1.1064,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999057711296186e-05,
+      "loss": 1.2588,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990171939756815e-05,
+      "loss": 1.1152,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989758242238268e-05,
+      "loss": 1.1724,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998933602075922e-05,
+      "loss": 1.0874,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998890527567993e-05,
+      "loss": 1.2217,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988466007367944e-05,
+      "loss": 1.1167,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988018216198077e-05,
+      "loss": 1.0405,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998756190255242e-05,
+      "loss": 1.0015,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987097066820324e-05,
+      "loss": 1.0439,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986623709398427e-05,
+      "loss": 1.2437,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986141830690626e-05,
+      "loss": 1.0732,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985651431108095e-05,
+      "loss": 1.0166,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985152511069274e-05,
+      "loss": 1.21,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998464507099988e-05,
+      "loss": 0.9873,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984129111332896e-05,
+      "loss": 1.0283,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983604632508572e-05,
+      "loss": 1.0366,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983071634974436e-05,
+      "loss": 1.0425,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982530119185277e-05,
+      "loss": 1.3691,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981980085603147e-05,
+      "loss": 0.9824,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981421534697384e-05,
+      "loss": 0.9526,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980854466944572e-05,
+      "loss": 0.9507,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980278882828582e-05,
+      "loss": 0.9551,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979694782840536e-05,
+      "loss": 1.1313,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979102167478833e-05,
+      "loss": 1.2202,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978501037249132e-05,
+      "loss": 0.9507,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997789139266436e-05,
+      "loss": 0.9414,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977273234244707e-05,
+      "loss": 0.9761,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976646562517633e-05,
+      "loss": 1.2876,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997601137801785e-05,
+      "loss": 1.1416,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975367681287358e-05,
+      "loss": 0.8979,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974715472875382e-05,
+      "loss": 1.0674,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997405475333845e-05,
+      "loss": 0.9087,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973385523240325e-05,
+      "loss": 1.2319,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972707783152042e-05,
+      "loss": 0.8501,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99720215336519e-05,
+      "loss": 1.0835,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971326775325453e-05,
+      "loss": 1.3188,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970623508765516e-05,
+      "loss": 1.2847,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969911734572166e-05,
+      "loss": 0.8911,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996919145335274e-05,
+      "loss": 1.0166,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968462665721828e-05,
+      "loss": 0.9434,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967725372301287e-05,
+      "loss": 1.2598,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996697957372023e-05,
+      "loss": 0.9668,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966225270615016e-05,
+      "loss": 1.1123,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965462463629274e-05,
+      "loss": 1.106,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964691153413883e-05,
+      "loss": 0.8794,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963911340626982e-05,
+      "loss": 0.8525,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996312302593396e-05,
+      "loss": 0.8574,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962326210007462e-05,
+      "loss": 0.8584,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961520893527385e-05,
+      "loss": 0.8877,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960707077180883e-05,
+      "loss": 0.9624,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995988476166236e-05,
+      "loss": 0.9126,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995905394767348e-05,
+      "loss": 0.9019,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958214635923144e-05,
+      "loss": 0.8535,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995736682712751e-05,
+      "loss": 0.875,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956510522009992e-05,
+      "loss": 0.854,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955645721301252e-05,
+      "loss": 0.8257,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954772425739194e-05,
+      "loss": 1.0684,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953890636068975e-05,
+      "loss": 0.8047,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953000353043e-05,
+      "loss": 0.7715,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952101577420925e-05,
+      "loss": 0.9097,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995119430996964e-05,
+      "loss": 0.8857,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950278551463298e-05,
+      "loss": 0.7871,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994935430268328e-05,
+      "loss": 1.0703,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948421564418227e-05,
+      "loss": 0.8794,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994748033746401e-05,
+      "loss": 0.875,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946530622623753e-05,
+      "loss": 0.8345,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945572420707825e-05,
+      "loss": 0.8315,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994460573253382e-05,
+      "loss": 0.979,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943630558926588e-05,
+      "loss": 0.8359,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942646900718218e-05,
+      "loss": 0.7651,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994165475874803e-05,
+      "loss": 1.3164,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99406541338626e-05,
+      "loss": 1.0464,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993964502691572e-05,
+      "loss": 0.7241,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938627438768433e-05,
+      "loss": 0.7749,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993760137028902e-05,
+      "loss": 1.002,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936566822352998e-05,
+      "loss": 0.7183,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935523795843106e-05,
+      "loss": 0.8188,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993447229164933e-05,
+      "loss": 0.7827,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993341231066889e-05,
+      "loss": 1.019,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932343853806233e-05,
+      "loss": 0.8047,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931266921973042e-05,
+      "loss": 0.8394,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930181516088233e-05,
+      "loss": 1.063,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992908763707795e-05,
+      "loss": 0.7676,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927985285875563e-05,
+      "loss": 0.9609,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926874463421676e-05,
+      "loss": 0.6841,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925755170664123e-05,
+      "loss": 0.7754,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924627408557963e-05,
+      "loss": 0.7954,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992349117806548e-05,
+      "loss": 0.7402,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922346480156184e-05,
+      "loss": 0.8513,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921193315806814e-05,
+      "loss": 0.7959,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920031686001332e-05,
+      "loss": 0.7319,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918861591730915e-05,
+      "loss": 1.1001,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917683033993978e-05,
+      "loss": 0.7656,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991649601379614e-05,
+      "loss": 0.9727,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915300532150257e-05,
+      "loss": 0.8892,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914096590076394e-05,
+      "loss": 1.0107,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991288418860184e-05,
+      "loss": 0.7273,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911663328761097e-05,
+      "loss": 1.2725,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910434011595893e-05,
+      "loss": 0.6853,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909196238155166e-05,
+      "loss": 0.9622,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990795000949507e-05,
+      "loss": 0.666,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906695326678975e-05,
+      "loss": 0.7554,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990543219077746e-05,
+      "loss": 0.6816,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990416060286833e-05,
+      "loss": 0.7478,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902880564036587e-05,
+      "loss": 0.6675,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901592075374447e-05,
+      "loss": 1.2104,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900295137981345e-05,
+      "loss": 0.7375,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898989752963915e-05,
+      "loss": 1.041,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897675921436002e-05,
+      "loss": 0.6289,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989635364451866e-05,
+      "loss": 0.6909,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895022923340152e-05,
+      "loss": 0.7378,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893683759035937e-05,
+      "loss": 0.9016,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989233615274868e-05,
+      "loss": 0.6665,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890980105628266e-05,
+      "loss": 0.9229,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988961561883176e-05,
+      "loss": 0.6594,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888242693523437e-05,
+      "loss": 1.0308,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886861330874777e-05,
+      "loss": 0.6548,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885471532064456e-05,
+      "loss": 1.1724,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988407329827834e-05,
+      "loss": 0.707,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988266663070951e-05,
+      "loss": 1.0103,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881251530558224e-05,
+      "loss": 0.5742,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879827999031952e-05,
+      "loss": 0.655,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878396037345342e-05,
+      "loss": 0.96,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876955646720253e-05,
+      "loss": 0.5879,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875506828385723e-05,
+      "loss": 0.7002,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874049583577983e-05,
+      "loss": 0.937,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987258391354046e-05,
+      "loss": 0.9072,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871109819523765e-05,
+      "loss": 0.5625,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98696273027857e-05,
+      "loss": 0.9258,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868136364591243e-05,
+      "loss": 0.9629,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866637006212582e-05,
+      "loss": 0.7754,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986512922892906e-05,
+      "loss": 0.5441,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 0.8921,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98620884228008e-05,
+      "loss": 0.5676,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860555396550693e-05,
+      "loss": 0.7085,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985901395658498e-05,
+      "loss": 0.5984,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857464104218933e-05,
+      "loss": 1.0095,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855905840774994e-05,
+      "loss": 0.7224,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985433916758278e-05,
+      "loss": 0.6213,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852764085979088e-05,
+      "loss": 0.6313,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851180597307884e-05,
+      "loss": 0.9939,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849588702920318e-05,
+      "loss": 0.6831,
+      "step": 294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98479884041747e-05,
+      "loss": 0.6111,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846379702436518e-05,
+      "loss": 0.9878,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844762599078427e-05,
+      "loss": 0.9211,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843137095480262e-05,
+      "loss": 1.2556,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841503193029005e-05,
+      "loss": 0.9709,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839860893118824e-05,
+      "loss": 0.6733,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983821019715104e-05,
+      "loss": 1.0337,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836551106534138e-05,
+      "loss": 0.6226,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834883622683775e-05,
+      "loss": 0.6045,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833207747022772e-05,
+      "loss": 0.6787,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983152348098109e-05,
+      "loss": 0.6987,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829830825995874e-05,
+      "loss": 0.7954,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828129783511406e-05,
+      "loss": 0.6292,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982642035497914e-05,
+      "loss": 0.8376,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982470254185768e-05,
+      "loss": 0.9644,
+      "step": 309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822976345612784e-05,
+      "loss": 0.6091,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982124176771736e-05,
+      "loss": 0.6021,
+      "step": 311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819498809651472e-05,
+      "loss": 0.6208,
+      "step": 312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981774747290234e-05,
+      "loss": 0.5876,
+      "step": 313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815987758964322e-05,
+      "loss": 0.6047,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981421966933893e-05,
+      "loss": 0.7039,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981244320553482e-05,
+      "loss": 0.8787,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810658369067795e-05,
+      "loss": 0.5354,
+      "step": 317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808865161460807e-05,
+      "loss": 0.99,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980706358424394e-05,
+      "loss": 0.5916,
+      "step": 319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805253638954428e-05,
+      "loss": 0.626,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803435327136647e-05,
+      "loss": 0.6143,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801608650342104e-05,
+      "loss": 1.4397,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799773610129446e-05,
+      "loss": 0.7947,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979793020806446e-05,
+      "loss": 0.6272,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796078445720065e-05,
+      "loss": 1.3381,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794218324676314e-05,
+      "loss": 0.6399,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792349846520395e-05,
+      "loss": 0.5701,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979047301284662e-05,
+      "loss": 0.6172,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978858782525644e-05,
+      "loss": 0.5269,
+      "step": 329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786694285358422e-05,
+      "loss": 0.6667,
+      "step": 330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978479239476827e-05,
+      "loss": 0.5627,
+      "step": 331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978288215510881e-05,
+      "loss": 1.0198,
+      "step": 332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780963568009996e-05,
+      "loss": 0.6331,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779036635108892e-05,
+      "loss": 0.6023,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97771013580497e-05,
+      "loss": 0.9197,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775157738483733e-05,
+      "loss": 0.6292,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773205778069418e-05,
+      "loss": 0.6221,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771245478472308e-05,
+      "loss": 0.5557,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976927684136507e-05,
+      "loss": 0.5537,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767299868427475e-05,
+      "loss": 1.3289,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765314561346424e-05,
+      "loss": 0.5996,
+      "step": 341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763320921815913e-05,
+      "loss": 0.543,
+      "step": 342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761318951537053e-05,
+      "loss": 0.5,
+      "step": 343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759308652218074e-05,
+      "loss": 0.541,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757290025574297e-05,
+      "loss": 0.543,
+      "step": 345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975526307332816e-05,
+      "loss": 0.9287,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97532277972092e-05,
+      "loss": 0.4875,
+      "step": 347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975118419895405e-05,
+      "loss": 0.9487,
+      "step": 348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749132280306456e-05,
+      "loss": 0.5496,
+      "step": 349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974707204301726e-05,
+      "loss": 0.5967,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97450034888444e-05,
+      "loss": 0.7856,
+      "step": 351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974292661955291e-05,
+      "loss": 0.6089,
+      "step": 352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740841436914917e-05,
+      "loss": 0.5789,
+      "step": 353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738747942709652e-05,
+      "loss": 0.6675,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736646138723423e-05,
+      "loss": 0.9719,
+      "step": 355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734536026749643e-05,
+      "loss": 0.6067,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732417608588803e-05,
+      "loss": 0.4873,
+      "step": 357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730290886048487e-05,
+      "loss": 0.5896,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972815586094336e-05,
+      "loss": 0.6013,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726012535095182e-05,
+      "loss": 1.0239,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723860910332783e-05,
+      "loss": 0.5945,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972170098849208e-05,
+      "loss": 0.5154,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971953277141607e-05,
+      "loss": 0.9045,
+      "step": 363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971735626095483e-05,
+      "loss": 0.6208,
+      "step": 364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715171458965505e-05,
+      "loss": 1.0198,
+      "step": 365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712978367312326e-05,
+      "loss": 0.5134,
+      "step": 366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710776987866597e-05,
+      "loss": 0.7432,
+      "step": 367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708567322506676e-05,
+      "loss": 1.3093,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706349373118012e-05,
+      "loss": 0.5208,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704123141593114e-05,
+      "loss": 1.0728,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970188862983156e-05,
+      "loss": 0.3594,
+      "step": 371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699645839739987e-05,
+      "loss": 0.9802,
+      "step": 372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697394773232104e-05,
+      "loss": 0.5378,
+      "step": 373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695135432228678e-05,
+      "loss": 0.7085,
+      "step": 374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692867818657535e-05,
+      "loss": 1.0117,
+      "step": 375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690591934453564e-05,
+      "loss": 0.6653,
+      "step": 376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688307781558705e-05,
+      "loss": 0.5044,
+      "step": 377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968601536192196e-05,
+      "loss": 0.4583,
+      "step": 378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683714677499385e-05,
+      "loss": 1.5541,
+      "step": 379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681405730254078e-05,
+      "loss": 1.0088,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679088522156198e-05,
+      "loss": 0.3608,
+      "step": 381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967676305518295e-05,
+      "loss": 0.5945,
+      "step": 382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967442933131858e-05,
+      "loss": 0.4756,
+      "step": 383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967208735255439e-05,
+      "loss": 0.8237,
+      "step": 384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966973712088872e-05,
+      "loss": 0.9255,
+      "step": 385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667378638326947e-05,
+      "loss": 1.0732,
+      "step": 386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665011906881496e-05,
+      "loss": 0.9521,
+      "step": 387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662636928571827e-05,
+      "loss": 0.8436,
+      "step": 388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966025370542444e-05,
+      "loss": 0.6716,
+      "step": 389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965786223947287e-05,
+      "loss": 0.7627,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655462532757677e-05,
+      "loss": 0.509,
+      "step": 391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965305458732646e-05,
+      "loss": 0.9365,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650638405233852e-05,
+      "loss": 0.9114,
+      "step": 393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96482139885415e-05,
+      "loss": 0.5082,
+      "step": 394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645781339318087e-05,
+      "loss": 0.5493,
+      "step": 395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643340459639327e-05,
+      "loss": 0.4348,
+      "step": 396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640891351587946e-05,
+      "loss": 0.4562,
+      "step": 397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638434017253693e-05,
+      "loss": 0.4783,
+      "step": 398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635968458733338e-05,
+      "loss": 0.3271,
+      "step": 399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633494678130666e-05,
+      "loss": 0.5459,
+      "step": 400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963101267755648e-05,
+      "loss": 0.7529,
+      "step": 401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96285224591286e-05,
+      "loss": 0.8545,
+      "step": 402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962602402497185e-05,
+      "loss": 0.5024,
+      "step": 403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623517377218072e-05,
+      "loss": 0.5947,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621002518006115e-05,
+      "loss": 0.4905,
+      "step": 405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618479449481826e-05,
+      "loss": 1.0601,
+      "step": 406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615948173798073e-05,
+      "loss": 0.5886,
+      "step": 407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613408693114707e-05,
+      "loss": 1.25,
+      "step": 408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96108610095986e-05,
+      "loss": 0.8434,
+      "step": 409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608305125423608e-05,
+      "loss": 0.5173,
+      "step": 410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960574104277059e-05,
+      "loss": 0.3787,
+      "step": 411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603168763827405e-05,
+      "loss": 0.3928,
+      "step": 412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600588290788898e-05,
+      "loss": 0.5625,
+      "step": 413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959799962585691e-05,
+      "loss": 0.4653,
+      "step": 414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959540277124027e-05,
+      "loss": 0.4977,
+      "step": 415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592797729154796e-05,
+      "loss": 0.4635,
+      "step": 416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959018450182329e-05,
+      "loss": 0.9431,
+      "step": 417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958756309147555e-05,
+      "loss": 0.5017,
+      "step": 418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958493350034834e-05,
+      "loss": 0.3733,
+      "step": 419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582295730685406e-05,
+      "loss": 0.668,
+      "step": 420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579649784737484e-05,
+      "loss": 0.8784,
+      "step": 421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957699566476228e-05,
+      "loss": 0.4596,
+      "step": 422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574333373024474e-05,
+      "loss": 0.4961,
+      "step": 423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571662911795718e-05,
+      "loss": 0.8868,
+      "step": 424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568984283354637e-05,
+      "loss": 1.3473,
+      "step": 425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566297489986826e-05,
+      "loss": 0.855,
+      "step": 426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563602533984843e-05,
+      "loss": 0.5729,
+      "step": 427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560899417648214e-05,
+      "loss": 0.6023,
+      "step": 428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558188143283425e-05,
+      "loss": 0.5996,
+      "step": 429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955546871320393e-05,
+      "loss": 0.5717,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552741129730132e-05,
+      "loss": 0.5343,
+      "step": 431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550005395189393e-05,
+      "loss": 0.512,
+      "step": 432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547261511916042e-05,
+      "loss": 0.5133,
+      "step": 433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544509482251344e-05,
+      "loss": 0.3679,
+      "step": 434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541749308543535e-05,
+      "loss": 0.848,
+      "step": 435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538980993147773e-05,
+      "loss": 0.9966,
+      "step": 436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536204538426185e-05,
+      "loss": 1.0093,
+      "step": 437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953341994674784e-05,
+      "loss": 0.4928,
+      "step": 438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530627220488744e-05,
+      "loss": 0.5183,
+      "step": 439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527826362031847e-05,
+      "loss": 1.0,
+      "step": 440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952501737376703e-05,
+      "loss": 0.3981,
+      "step": 441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952220025809113e-05,
+      "loss": 0.9813,
+      "step": 442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519375017407896e-05,
+      "loss": 0.4576,
+      "step": 443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951654165412803e-05,
+      "loss": 0.7488,
+      "step": 444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513700170669152e-05,
+      "loss": 0.4868,
+      "step": 445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510850569455815e-05,
+      "loss": 1.4357,
+      "step": 446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507992852919496e-05,
+      "loss": 0.8027,
+      "step": 447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505127023498603e-05,
+      "loss": 0.7433,
+      "step": 448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950225308363846e-05,
+      "loss": 0.567,
+      "step": 449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949937103579131e-05,
+      "loss": 0.5955,
+      "step": 450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496480882416316e-05,
+      "loss": 0.5687,
+      "step": 451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949358262597957e-05,
+      "loss": 0.8184,
+      "step": 452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490676268954063e-05,
+      "loss": 0.929,
+      "step": 453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487761813819698e-05,
+      "loss": 0.6383,
+      "step": 454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484839263063294e-05,
+      "loss": 0.3152,
+      "step": 455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481908619178576e-05,
+      "loss": 0.5187,
+      "step": 456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478969884666173e-05,
+      "loss": 0.7236,
+      "step": 457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476023062033617e-05,
+      "loss": 0.3152,
+      "step": 458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473068153795353e-05,
+      "loss": 0.9137,
+      "step": 459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470105162472705e-05,
+      "loss": 0.3164,
+      "step": 460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946713409059391e-05,
+      "loss": 1.0916,
+      "step": 461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464154940694086e-05,
+      "loss": 0.7479,
+      "step": 462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461167715315264e-05,
+      "loss": 0.3843,
+      "step": 463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 0.8092,
+      "step": 464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455169048323136e-05,
+      "loss": 0.4742,
+      "step": 465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452157611828312e-05,
+      "loss": 0.2946,
+      "step": 466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449138110091444e-05,
+      "loss": 0.6783,
+      "step": 467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446110545688983e-05,
+      "loss": 0.3647,
+      "step": 468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944307492120426e-05,
+      "loss": 0.8917,
+      "step": 469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440031239227476e-05,
+      "loss": 0.4599,
+      "step": 470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436979502355725e-05,
+      "loss": 0.5378,
+      "step": 471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433919713192952e-05,
+      "loss": 0.6951,
+      "step": 472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430851874349983e-05,
+      "loss": 0.4847,
+      "step": 473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942777598844452e-05,
+      "loss": 0.3315,
+      "step": 474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424692058101123e-05,
+      "loss": 0.9651,
+      "step": 475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942160008595121e-05,
+      "loss": 0.6113,
+      "step": 476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941850007463307e-05,
+      "loss": 0.4823,
+      "step": 477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415392026791857e-05,
+      "loss": 0.2762,
+      "step": 478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412275945079568e-05,
+      "loss": 0.8073,
+      "step": 479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940915183215506e-05,
+      "loss": 0.785,
+      "step": 480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406019690684054e-05,
+      "loss": 0.4011,
+      "step": 481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402879523339103e-05,
+      "loss": 0.3877,
+      "step": 482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939973133279962e-05,
+      "loss": 0.6061,
+      "step": 483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396575121751863e-05,
+      "loss": 0.3854,
+      "step": 484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939341089288893e-05,
+      "loss": 0.6765,
+      "step": 485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390238648910765e-05,
+      "loss": 0.3555,
+      "step": 486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387058392524146e-05,
+      "loss": 0.4027,
+      "step": 487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383870126442694e-05,
+      "loss": 0.6052,
+      "step": 488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380673853386855e-05,
+      "loss": 0.8309,
+      "step": 489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377469576083917e-05,
+      "loss": 0.5615,
+      "step": 490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937425729726799e-05,
+      "loss": 0.4219,
+      "step": 491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371037019680017e-05,
+      "loss": 0.8074,
+      "step": 492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367808746067768e-05,
+      "loss": 0.4386,
+      "step": 493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364572479185824e-05,
+      "loss": 0.8373,
+      "step": 494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93613282217956e-05,
+      "loss": 0.418,
+      "step": 495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935807597666532e-05,
+      "loss": 0.5322,
+      "step": 496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354815746570033e-05,
+      "loss": 0.9189,
+      "step": 497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935154753429159e-05,
+      "loss": 0.9449,
+      "step": 498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348271342618657e-05,
+      "loss": 0.5176,
+      "step": 499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344987174346712e-05,
+      "loss": 0.3682,
+      "step": 500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341695032278038e-05,
+      "loss": 0.3528,
+      "step": 501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933839491922172e-05,
+      "loss": 0.4426,
+      "step": 502
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335086837993648e-05,
+      "loss": 0.4111,
+      "step": 503
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331770791416504e-05,
+      "loss": 0.2427,
+      "step": 504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932844678231977e-05,
+      "loss": 0.3566,
+      "step": 505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932511481353973e-05,
+      "loss": 0.4603,
+      "step": 506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321774887919452e-05,
+      "loss": 0.7996,
+      "step": 507
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318427008308785e-05,
+      "loss": 0.4396,
+      "step": 508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931507117756438e-05,
+      "loss": 0.5026,
+      "step": 509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931170739854967e-05,
+      "loss": 0.3997,
+      "step": 510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930833567413486e-05,
+      "loss": 0.4333,
+      "step": 511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304956007196943e-05,
+      "loss": 0.8384,
+      "step": 512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301568400619693e-05,
+      "loss": 0.3602,
+      "step": 513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929817285729364e-05,
+      "loss": 0.3578,
+      "step": 514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294769380116117e-05,
+      "loss": 0.881,
+      "step": 515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291357971991193e-05,
+      "loss": 0.6452,
+      "step": 516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928793863582973e-05,
+      "loss": 0.5769,
+      "step": 517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284511374549338e-05,
+      "loss": 0.7246,
+      "step": 518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92810761910744e-05,
+      "loss": 0.4211,
+      "step": 519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277633088336053e-05,
+      "loss": 0.8365,
+      "step": 520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274182069272194e-05,
+      "loss": 0.7735,
+      "step": 521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270723136827478e-05,
+      "loss": 0.8163,
+      "step": 522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267256293953298e-05,
+      "loss": 0.866,
+      "step": 523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263781543607817e-05,
+      "loss": 0.7026,
+      "step": 524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260298888755927e-05,
+      "loss": 0.5753,
+      "step": 525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256808332369278e-05,
+      "loss": 0.576,
+      "step": 526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253309877426257e-05,
+      "loss": 0.7861,
+      "step": 527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249803526911988e-05,
+      "loss": 1.2859,
+      "step": 528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246289283818334e-05,
+      "loss": 1.0697,
+      "step": 529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242767151143896e-05,
+      "loss": 0.5551,
+      "step": 530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239237131894e-05,
+      "loss": 0.8229,
+      "step": 531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923569922908071e-05,
+      "loss": 0.8806,
+      "step": 532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923215344572281e-05,
+      "loss": 0.6726,
+      "step": 533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922859978484581e-05,
+      "loss": 0.4244,
+      "step": 534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922503824948194e-05,
+      "loss": 0.9734,
+      "step": 535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221468842670156e-05,
+      "loss": 0.428,
+      "step": 536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217891567456123e-05,
+      "loss": 0.2983,
+      "step": 537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921430642689222e-05,
+      "loss": 0.7286,
+      "step": 538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210713424037546e-05,
+      "loss": 0.4349,
+      "step": 539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207112561957894e-05,
+      "loss": 0.2996,
+      "step": 540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920350384372578e-05,
+      "loss": 0.3784,
+      "step": 541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919988727242041e-05,
+      "loss": 0.439,
+      "step": 542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196262851127695e-05,
+      "loss": 0.296,
+      "step": 543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9192630582940243e-05,
+      "loss": 0.7639,
+      "step": 544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918899047095737e-05,
+      "loss": 0.9631,
+      "step": 545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918534251828506e-05,
+      "loss": 0.3987,
+      "step": 546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918168672803601e-05,
+      "loss": 0.4904,
+      "step": 547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178023103329595e-05,
+      "loss": 1.1387,
+      "step": 548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917435164729187e-05,
+      "loss": 0.8257,
+      "step": 549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917067236305559e-05,
+      "loss": 0.7617,
+      "step": 550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166985253760165e-05,
+      "loss": 1.0757,
+      "step": 551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163290322551704e-05,
+      "loss": 0.9746,
+      "step": 552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159587572582973e-05,
+      "loss": 0.2766,
+      "step": 553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155877007013424e-05,
+      "loss": 0.4918,
+      "step": 554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152158629009168e-05,
+      "loss": 0.4626,
+      "step": 555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148432441742985e-05,
+      "loss": 0.5827,
+      "step": 556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914469844839432e-05,
+      "loss": 0.4862,
+      "step": 557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140956652149275e-05,
+      "loss": 0.2722,
+      "step": 558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137207056200612e-05,
+      "loss": 0.4496,
+      "step": 559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133449663747753e-05,
+      "loss": 0.9,
+      "step": 560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129684477996762e-05,
+      "loss": 0.4193,
+      "step": 561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125911502160365e-05,
+      "loss": 0.759,
+      "step": 562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122130739457926e-05,
+      "loss": 0.3636,
+      "step": 563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118342193115456e-05,
+      "loss": 0.5067,
+      "step": 564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114545866365608e-05,
+      "loss": 0.3976,
+      "step": 565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110741762447673e-05,
+      "loss": 0.5597,
+      "step": 566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910692988460758e-05,
+      "loss": 0.4038,
+      "step": 567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103110236097885e-05,
+      "loss": 0.5552,
+      "step": 568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909928282017779e-05,
+      "loss": 0.5312,
+      "step": 569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9095447640113104e-05,
+      "loss": 0.7185,
+      "step": 570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909160469917627e-05,
+      "loss": 0.239,
+      "step": 571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087754000646362e-05,
+      "loss": 0.2977,
+      "step": 572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908389554780906e-05,
+      "loss": 0.5998,
+      "step": 573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908002934395667e-05,
+      "loss": 0.8334,
+      "step": 574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90761553923881e-05,
+      "loss": 0.772,
+      "step": 575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072273696408886e-05,
+      "loss": 0.421,
+      "step": 576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068384259331156e-05,
+      "loss": 0.5226,
+      "step": 577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064487084473652e-05,
+      "loss": 0.8788,
+      "step": 578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9060582175161713e-05,
+      "loss": 0.3787,
+      "step": 579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056669534727287e-05,
+      "loss": 0.3815,
+      "step": 580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905274916650891e-05,
+      "loss": 0.4909,
+      "step": 581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904882107385171e-05,
+      "loss": 0.5677,
+      "step": 582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044885260107416e-05,
+      "loss": 0.3018,
+      "step": 583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040941728634338e-05,
+      "loss": 0.4001,
+      "step": 584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903699048279737e-05,
+      "loss": 0.3793,
+      "step": 585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033031525967992e-05,
+      "loss": 0.2059,
+      "step": 586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029064861524267e-05,
+      "loss": 0.3337,
+      "step": 587
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902509049285083e-05,
+      "loss": 0.7891,
+      "step": 588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021108423338886e-05,
+      "loss": 0.4841,
+      "step": 589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901711865638622e-05,
+      "loss": 0.7152,
+      "step": 590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013121195397175e-05,
+      "loss": 0.4467,
+      "step": 591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900911604378267e-05,
+      "loss": 0.8743,
+      "step": 592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005103204960174e-05,
+      "loss": 0.2305,
+      "step": 593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900108268235373e-05,
+      "loss": 0.7198,
+      "step": 594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997054479393925e-05,
+      "loss": 0.2317,
+      "step": 595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993018599517897e-05,
+      "loss": 0.241,
+      "step": 596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988975046169352e-05,
+      "loss": 0.4935,
+      "step": 597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898492382279853e-05,
+      "loss": 0.403,
+      "step": 598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980864932862214e-05,
+      "loss": 0.4838,
+      "step": 599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897679837982373e-05,
+      "loss": 0.275,
+      "step": 600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972724167152958e-05,
+      "loss": 0.3377,
+      "step": 601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896864229832629e-05,
+      "loss": 0.2826,
+      "step": 602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964552776826662e-05,
+      "loss": 0.9433,
+      "step": 603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896045560614355e-05,
+      "loss": 0.3991,
+      "step": 604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956350789772937e-05,
+      "loss": 0.2302,
+      "step": 605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895223833121734e-05,
+      "loss": 0.2672,
+      "step": 606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948118233985803e-05,
+      "loss": 0.2455,
+      "step": 607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943990501593873e-05,
+      "loss": 0.6898,
+      "step": 608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939855137563627e-05,
+      "loss": 0.3153,
+      "step": 609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935712145423643e-05,
+      "loss": 2.3079,
+      "step": 610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893156152870901e-05,
+      "loss": 0.571,
+      "step": 611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892740329096133e-05,
+      "loss": 0.189,
+      "step": 612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89232374357287e-05,
+      "loss": 0.5747,
+      "step": 613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8919063966565717e-05,
+      "loss": 0.2634,
+      "step": 614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891488288703348e-05,
+      "loss": 0.9557,
+      "step": 615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891069420069957e-05,
+      "loss": 0.5302,
+      "step": 616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906497911138082e-05,
+      "loss": 0.4465,
+      "step": 617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902294021929572e-05,
+      "loss": 0.4233,
+      "step": 618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898082536661097e-05,
+      "loss": 0.4144,
+      "step": 619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893863458926185e-05,
+      "loss": 0.3876,
+      "step": 620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888963679232486e-05,
+      "loss": 0.9451,
+      "step": 621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885402540463598e-05,
+      "loss": 0.4747,
+      "step": 622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881160706955364e-05,
+      "loss": 0.5701,
+      "step": 623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887691129541959e-05,
+      "loss": 0.4028,
+      "step": 624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872654309482163e-05,
+      "loss": 0.2614,
+      "step": 625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868389752775447e-05,
+      "loss": 0.6526,
+      "step": 626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886411762893826e-05,
+      "loss": 0.3917,
+      "step": 627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859837941615878e-05,
+      "loss": 1.1537,
+      "step": 628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855550694460026e-05,
+      "loss": 1.2947,
+      "step": 629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851255891128883e-05,
+      "loss": 0.4467,
+      "step": 630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846953535287078e-05,
+      "loss": 1.0808,
+      "step": 631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884264363060568e-05,
+      "loss": 0.344,
+      "step": 632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838326180762205e-05,
+      "loss": 0.2465,
+      "step": 633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88340011894406e-05,
+      "loss": 0.6399,
+      "step": 634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829668660331252e-05,
+      "loss": 0.8007,
+      "step": 635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882532859713097e-05,
+      "loss": 0.263,
+      "step": 636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820981003543013e-05,
+      "loss": 1.0797,
+      "step": 637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816625883277044e-05,
+      "loss": 0.3289,
+      "step": 638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812263240049152e-05,
+      "loss": 0.8176,
+      "step": 639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807893077581863e-05,
+      "loss": 0.2141,
+      "step": 640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880351539960409e-05,
+      "loss": 0.288,
+      "step": 641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799130209851182e-05,
+      "loss": 0.5497,
+      "step": 642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 0.4019,
+      "step": 643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879033730999337e-05,
+      "loss": 0.5692,
+      "step": 644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785929607391184e-05,
+      "loss": 0.8259,
+      "step": 645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878151440801929e-05,
+      "loss": 0.2216,
+      "step": 646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877709171564504e-05,
+      "loss": 0.304,
+      "step": 647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772661534042195e-05,
+      "loss": 0.2951,
+      "step": 648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8768223866990884e-05,
+      "loss": 0.722,
+      "step": 649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763778718277645e-05,
+      "loss": 0.4215,
+      "step": 650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759326091695385e-05,
+      "loss": 0.8502,
+      "step": 651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754865991043402e-05,
+      "loss": 0.2514,
+      "step": 652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750398420127353e-05,
+      "loss": 0.3725,
+      "step": 653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745923382759297e-05,
+      "loss": 1.4349,
+      "step": 654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874144088275764e-05,
+      "loss": 0.9513,
+      "step": 655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736950923947164e-05,
+      "loss": 0.6208,
+      "step": 656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732453510159025e-05,
+      "loss": 0.2908,
+      "step": 657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872794864523072e-05,
+      "loss": 0.8509,
+      "step": 658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723436333006124e-05,
+      "loss": 0.5429,
+      "step": 659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871891657733545e-05,
+      "loss": 0.6998,
+      "step": 660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714389382075273e-05,
+      "loss": 0.7268,
+      "step": 661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870985475108851e-05,
+      "loss": 0.3474,
+      "step": 662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705312688244432e-05,
+      "loss": 0.4873,
+      "step": 663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700763197418638e-05,
+      "loss": 0.3262,
+      "step": 664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696206282493076e-05,
+      "loss": 0.5626,
+      "step": 665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691641947356022e-05,
+      "loss": 0.5864,
+      "step": 666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868707019590209e-05,
+      "loss": 0.6277,
+      "step": 667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868249103203221e-05,
+      "loss": 1.202,
+      "step": 668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867790445965365e-05,
+      "loss": 0.8488,
+      "step": 669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673310482679997e-05,
+      "loss": 0.4392,
+      "step": 670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866870910503115e-05,
+      "loss": 0.2325,
+      "step": 671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664100330633327e-05,
+      "loss": 1.0388,
+      "step": 672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865948416341906e-05,
+      "loss": 0.6413,
+      "step": 673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654860607327177e-05,
+      "loss": 0.9382,
+      "step": 674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650229666302827e-05,
+      "loss": 1.2094,
+      "step": 675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864559134429745e-05,
+      "loss": 1.2137,
+      "step": 676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864094564526879e-05,
+      "loss": 0.9352,
+      "step": 677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863629257318088e-05,
+      "loss": 0.689,
+      "step": 678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631632132004048e-05,
+      "loss": 0.8406,
+      "step": 679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626964325714903e-05,
+      "loss": 0.3087,
+      "step": 680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862228915829635e-05,
+      "loss": 1.0965,
+      "step": 681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617606633737565e-05,
+      "loss": 0.6866,
+      "step": 682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861291675603401e-05,
+      "loss": 0.2468,
+      "step": 683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860821952918741e-05,
+      "loss": 0.8484,
+      "step": 684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860351495720577e-05,
+      "loss": 0.5486,
+      "step": 685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859880304410337e-05,
+      "loss": 0.2621,
+      "step": 686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859408379390073e-05,
+      "loss": 0.2679,
+      "step": 687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589357210624647e-05,
+      "loss": 0.7312,
+      "step": 688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584623298308176e-05,
+      "loss": 0.4461,
+      "step": 689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579882060990627e-05,
+      "loss": 0.4261,
+      "step": 690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575133502717545e-05,
+      "loss": 0.323,
+      "step": 691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570377627540735e-05,
+      "loss": 0.5718,
+      "step": 692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565614439518246e-05,
+      "loss": 0.2139,
+      "step": 693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560843942714363e-05,
+      "loss": 0.2709,
+      "step": 694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85560661411996e-05,
+      "loss": 0.3137,
+      "step": 695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855128103905072e-05,
+      "loss": 0.3999,
+      "step": 696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546488640350704e-05,
+      "loss": 1.0627,
+      "step": 697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541688949188762e-05,
+      "loss": 0.3723,
+      "step": 698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536881969660326e-05,
+      "loss": 0.2489,
+      "step": 699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853206770586705e-05,
+      "loss": 0.6239,
+      "step": 700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527246161916796e-05,
+      "loss": 0.5787,
+      "step": 701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852241734192364e-05,
+      "loss": 0.1808,
+      "step": 702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517581250007878e-05,
+      "loss": 0.7042,
+      "step": 703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512737890295996e-05,
+      "loss": 0.9556,
+      "step": 704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850788726692069e-05,
+      "loss": 0.8184,
+      "step": 705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503029384020847e-05,
+      "loss": 0.4521,
+      "step": 706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498164245741558e-05,
+      "loss": 0.7892,
+      "step": 707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493291856234093e-05,
+      "loss": 0.9554,
+      "step": 708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848841221965592e-05,
+      "loss": 1.0041,
+      "step": 709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483525340170687e-05,
+      "loss": 0.799,
+      "step": 710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478631221948217e-05,
+      "loss": 0.6783,
+      "step": 711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473729869164517e-05,
+      "loss": 0.9951,
+      "step": 712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468821286001768e-05,
+      "loss": 0.5847,
+      "step": 713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846390547664831e-05,
+      "loss": 0.7031,
+      "step": 714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458982445298656e-05,
+      "loss": 0.552,
+      "step": 715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454052196153483e-05,
+      "loss": 0.4031,
+      "step": 716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449114733419626e-05,
+      "loss": 0.1505,
+      "step": 717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844417006131007e-05,
+      "loss": 0.5539,
+      "step": 718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8439218184043953e-05,
+      "loss": 0.3538,
+      "step": 719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434259105846574e-05,
+      "loss": 0.2225,
+      "step": 720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842929283094935e-05,
+      "loss": 0.5009,
+      "step": 721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842431936358987e-05,
+      "loss": 0.8411,
+      "step": 722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841933870801183e-05,
+      "loss": 0.4171,
+      "step": 723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841435086846508e-05,
+      "loss": 0.9374,
+      "step": 724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409355849205597e-05,
+      "loss": 0.3606,
+      "step": 725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404353654495478e-05,
+      "loss": 0.7672,
+      "step": 726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839934428860294e-05,
+      "loss": 0.5339,
+      "step": 727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394327755802334e-05,
+      "loss": 1.2039,
+      "step": 728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838930406037411e-05,
+      "loss": 0.2981,
+      "step": 729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838427320660484e-05,
+      "loss": 0.4807,
+      "step": 730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83792351987872e-05,
+      "loss": 0.3856,
+      "step": 731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374190041219964e-05,
+      "loss": 0.8711,
+      "step": 732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836913773820802e-05,
+      "loss": 0.5428,
+      "step": 733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364078294062347e-05,
+      "loss": 0.8307,
+      "step": 734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835901171310001e-05,
+      "loss": 0.6967,
+      "step": 735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353937999644183e-05,
+      "loss": 0.1484,
+      "step": 736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348857158024102e-05,
+      "loss": 0.4369,
+      "step": 737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343769192575096e-05,
+      "loss": 0.9841,
+      "step": 738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833867410763858e-05,
+      "loss": 0.6188,
+      "step": 739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333571907562034e-05,
+      "loss": 1.0972,
+      "step": 740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832846259669901e-05,
+      "loss": 0.1863,
+      "step": 741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832334617940913e-05,
+      "loss": 0.2886,
+      "step": 742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318222660058082e-05,
+      "loss": 0.4259,
+      "step": 743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313092043017606e-05,
+      "loss": 0.8101,
+      "step": 744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830795433266551e-05,
+      "loss": 0.9282,
+      "step": 745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302809533385644e-05,
+      "loss": 0.1477,
+      "step": 746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297657649567912e-05,
+      "loss": 0.8251,
+      "step": 747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292498685608257e-05,
+      "loss": 0.1666,
+      "step": 748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828733264590867e-05,
+      "loss": 0.9816,
+      "step": 749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282159534877183e-05,
+      "loss": 1.185,
+      "step": 750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276979356927853e-05,
+      "loss": 0.3735,
+      "step": 751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271792116480767e-05,
+      "loss": 0.7466,
+      "step": 752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266597817962042e-05,
+      "loss": 0.8626,
+      "step": 753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826139646580382e-05,
+      "loss": 0.7125,
+      "step": 754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825618806444426e-05,
+      "loss": 0.2592,
+      "step": 755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250972618327528e-05,
+      "loss": 0.1321,
+      "step": 756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245750131903813e-05,
+      "loss": 0.6017,
+      "step": 757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82405206096293e-05,
+      "loss": 0.4521,
+      "step": 758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235284055966192e-05,
+      "loss": 0.491,
+      "step": 759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230040475382672e-05,
+      "loss": 0.527,
+      "step": 760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822478987235294e-05,
+      "loss": 0.2097,
+      "step": 761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821953225135717e-05,
+      "loss": 0.8889,
+      "step": 762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214267616881535e-05,
+      "loss": 0.2892,
+      "step": 763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208995973418192e-05,
+      "loss": 1.4106,
+      "step": 764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820371732546527e-05,
+      "loss": 0.8243,
+      "step": 765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819843167752689e-05,
+      "loss": 0.5102,
+      "step": 766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193139034113124e-05,
+      "loss": 0.3443,
+      "step": 767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187839399740034e-05,
+      "loss": 0.314,
+      "step": 768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182532778929637e-05,
+      "loss": 1.4253,
+      "step": 769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177219176209915e-05,
+      "loss": 1.1566,
+      "step": 770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171898596114804e-05,
+      "loss": 0.8369,
+      "step": 771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8166571043184193e-05,
+      "loss": 0.3735,
+      "step": 772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161236521963928e-05,
+      "loss": 0.3326,
+      "step": 773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815589503700579e-05,
+      "loss": 0.8754,
+      "step": 774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150546592867505e-05,
+      "loss": 0.8222,
+      "step": 775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814519119411275e-05,
+      "loss": 1.0836,
+      "step": 776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139828845311118e-05,
+      "loss": 0.3328,
+      "step": 777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134459551038143e-05,
+      "loss": 0.651,
+      "step": 778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8129083315875282e-05,
+      "loss": 0.6948,
+      "step": 779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123700144409916e-05,
+      "loss": 0.2324,
+      "step": 780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811831004123534e-05,
+      "loss": 0.9345,
+      "step": 781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811291301095077e-05,
+      "loss": 0.7956,
+      "step": 782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107509058161328e-05,
+      "loss": 0.6897,
+      "step": 783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102098187478046e-05,
+      "loss": 0.5327,
+      "step": 784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096680403517857e-05,
+      "loss": 0.2158,
+      "step": 785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091255710903593e-05,
+      "loss": 0.5156,
+      "step": 786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808582411426398e-05,
+      "loss": 0.8839,
+      "step": 787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808038561823364e-05,
+      "loss": 0.4101,
+      "step": 788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074940227453074e-05,
+      "loss": 0.1215,
+      "step": 789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069487946568675e-05,
+      "loss": 0.5002,
+      "step": 790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064028780232702e-05,
+      "loss": 0.7568,
+      "step": 791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805856273310331e-05,
+      "loss": 0.5386,
+      "step": 792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80530898098445e-05,
+      "loss": 0.7807,
+      "step": 793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804761001512616e-05,
+      "loss": 0.9308,
+      "step": 794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8042123353624032e-05,
+      "loss": 0.2229,
+      "step": 795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803662983001972e-05,
+      "loss": 0.4825,
+      "step": 796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031129449000687e-05,
+      "loss": 0.6192,
+      "step": 797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025622215260236e-05,
+      "loss": 0.4946,
+      "step": 798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020108133497528e-05,
+      "loss": 0.1065,
+      "step": 799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801458720841756e-05,
+      "loss": 0.6417,
+      "step": 800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800905944473117e-05,
+      "loss": 0.5505,
+      "step": 801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003524847155042e-05,
+      "loss": 0.3276,
+      "step": 802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997983420411674e-05,
+      "loss": 0.1385,
+      "step": 803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992435169229404e-05,
+      "loss": 0.269,
+      "step": 804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798688009834238e-05,
+      "loss": 0.2385,
+      "step": 805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981318212490584e-05,
+      "loss": 0.6038,
+      "step": 806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79757495164198e-05,
+      "loss": 0.2047,
+      "step": 807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797017401488164e-05,
+      "loss": 0.9045,
+      "step": 808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79645917126335e-05,
+      "loss": 0.6609,
+      "step": 809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959002614438595e-05,
+      "loss": 0.409,
+      "step": 810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953406725065942e-05,
+      "loss": 0.7516,
+      "step": 811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794780404929033e-05,
+      "loss": 0.477,
+      "step": 812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942194591892366e-05,
+      "loss": 1.0707,
+      "step": 813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793657835765843e-05,
+      "loss": 0.567,
+      "step": 814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793095535138068e-05,
+      "loss": 0.6011,
+      "step": 815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925325577857062e-05,
+      "loss": 0.258,
+      "step": 816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7919689041891292e-05,
+      "loss": 0.291,
+      "step": 817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914045748292858e-05,
+      "loss": 0.4521,
+      "step": 818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908395701877012e-05,
+      "loss": 0.451,
+      "step": 819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790273890746477e-05,
+      "loss": 0.5448,
+      "step": 820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897075369882903e-05,
+      "loss": 0.6539,
+      "step": 821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 0.4166,
+      "step": 822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788572808454615e-05,
+      "loss": 0.1795,
+      "step": 823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788004434647356e-05,
+      "loss": 0.5482,
+      "step": 824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874353884595935e-05,
+      "loss": 0.4229,
+      "step": 825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868656703768773e-05,
+      "loss": 0.4879,
+      "step": 826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862952808853307e-05,
+      "loss": 0.5322,
+      "step": 827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857242204716497e-05,
+      "loss": 0.9259,
+      "step": 828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851524896231032e-05,
+      "loss": 0.2347,
+      "step": 829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784580088827532e-05,
+      "loss": 1.0086,
+      "step": 830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784007018573348e-05,
+      "loss": 0.6985,
+      "step": 831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834332793495363e-05,
+      "loss": 0.4455,
+      "step": 832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782858871645649e-05,
+      "loss": 0.8049,
+      "step": 833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822837959518133e-05,
+      "loss": 0.1666,
+      "step": 834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817080527587222e-05,
+      "loss": 0.4194,
+      "step": 835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811316425576414e-05,
+      "loss": 0.198,
+      "step": 836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780554565840403e-05,
+      "loss": 1.2263,
+      "step": 837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799768230994105e-05,
+      "loss": 0.6628,
+      "step": 838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793984148276342e-05,
+      "loss": 1.5724,
+      "step": 839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778819341518612e-05,
+      "loss": 0.5891,
+      "step": 840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77823960366645e-05,
+      "loss": 0.442,
+      "step": 841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777659201765821e-05,
+      "loss": 0.2418,
+      "step": 842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770781363119644e-05,
+      "loss": 1.1831,
+      "step": 843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776496407800686e-05,
+      "loss": 0.2484,
+      "step": 844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759140167283576e-05,
+      "loss": 0.6769,
+      "step": 845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775330963591916e-05,
+      "loss": 0.3474,
+      "step": 846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747472488888622e-05,
+      "loss": 0.3354,
+      "step": 847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774162873117263e-05,
+      "loss": 0.7199,
+      "step": 848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735778367757484e-05,
+      "loss": 1.1024,
+      "step": 849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729921403635128e-05,
+      "loss": 0.3478,
+      "step": 850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7724057843803127e-05,
+      "loss": 0.3427,
+      "step": 851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718187693264687e-05,
+      "loss": 0.5044,
+      "step": 852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712310957028626e-05,
+      "loss": 0.4734,
+      "step": 853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7706427640109386e-05,
+      "loss": 0.1838,
+      "step": 854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770053774752703e-05,
+      "loss": 0.2723,
+      "step": 855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769464128430722e-05,
+      "loss": 0.6034,
+      "step": 856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688738255481233e-05,
+      "loss": 0.239,
+      "step": 857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768282866608595e-05,
+      "loss": 0.9707,
+      "step": 858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767691252116384e-05,
+      "loss": 0.7888,
+      "step": 859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7670989825762975e-05,
+      "loss": 0.3989,
+      "step": 860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7665060584937018e-05,
+      "loss": 1.216,
+      "step": 861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76591248037452e-05,
+      "loss": 0.8524,
+      "step": 862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653182487252355e-05,
+      "loss": 0.6805,
+      "step": 863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764723364052888e-05,
+      "loss": 0.6284,
+      "step": 864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641278268650743e-05,
+      "loss": 0.3566,
+      "step": 865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763531637669949e-05,
+      "loss": 0.9513,
+      "step": 866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762934796976222e-05,
+      "loss": 0.8322,
+      "step": 867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623373052931598e-05,
+      "loss": 0.6082,
+      "step": 868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617391631305843e-05,
+      "loss": 0.5133,
+      "step": 869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611403709988716e-05,
+      "loss": 0.3288,
+      "step": 870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760540929408953e-05,
+      "loss": 1.0296,
+      "step": 871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759940838872315e-05,
+      "loss": 0.4813,
+      "step": 872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759340099900996e-05,
+      "loss": 0.2971,
+      "step": 873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587387130075883e-05,
+      "loss": 1.0193,
+      "step": 874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581366787052384e-05,
+      "loss": 0.3594,
+      "step": 875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757533997507643e-05,
+      "loss": 1.395,
+      "step": 876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7569306699290517e-05,
+      "loss": 0.5582,
+      "step": 877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563266964842666e-05,
+      "loss": 0.2601,
+      "step": 878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75572207768864e-05,
+      "loss": 0.7582,
+      "step": 879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551168140580745e-05,
+      "loss": 0.5662,
+      "step": 880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545109061090236e-05,
+      "loss": 0.6697,
+      "step": 881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539043543584905e-05,
+      "loss": 0.3607,
+      "step": 882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753297159324027e-05,
+      "loss": 1.0374,
+      "step": 883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752689321523735e-05,
+      "loss": 0.212,
+      "step": 884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752080841476264e-05,
+      "loss": 0.4259,
+      "step": 885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751471719700812e-05,
+      "loss": 0.2545,
+      "step": 886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508619567171238e-05,
+      "loss": 0.4983,
+      "step": 887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7502515530454924e-05,
+      "loss": 0.3051,
+      "step": 888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496405092067563e-05,
+      "loss": 0.2261,
+      "step": 889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490288257223013e-05,
+      "loss": 0.2671,
+      "step": 890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484165031140582e-05,
+      "loss": 0.3861,
+      "step": 891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7478035419045038e-05,
+      "loss": 0.5333,
+      "step": 892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747189942616659e-05,
+      "loss": 0.6714,
+      "step": 893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465757057740905e-05,
+      "loss": 0.4656,
+      "step": 894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459608319009074e-05,
+      "loss": 0.573,
+      "step": 895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453453215217634e-05,
+      "loss": 0.9587,
+      "step": 896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744729175161855e-05,
+      "loss": 0.2686,
+      "step": 897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441123933469208e-05,
+      "loss": 0.48,
+      "step": 898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743494976603243e-05,
+      "loss": 0.7026,
+      "step": 899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428769254576444e-05,
+      "loss": 1.2492,
+      "step": 900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7422582404374893e-05,
+      "loss": 0.3201,
+      "step": 901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416389220706836e-05,
+      "loss": 0.5021,
+      "step": 902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410189708856725e-05,
+      "loss": 0.3507,
+      "step": 903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403983874114422e-05,
+      "loss": 0.4243,
+      "step": 904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397771721775174e-05,
+      "loss": 0.2614,
+      "step": 905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391553257139626e-05,
+      "loss": 0.7387,
+      "step": 906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385328485513804e-05,
+      "loss": 0.1833,
+      "step": 907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737909741220913e-05,
+      "loss": 0.6495,
+      "step": 908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737286004254238e-05,
+      "loss": 0.521,
+      "step": 909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366616381835715e-05,
+      "loss": 0.589,
+      "step": 910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360366435416668e-05,
+      "loss": 0.8738,
+      "step": 911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7354110208618124e-05,
+      "loss": 0.5304,
+      "step": 912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347847706778344e-05,
+      "loss": 0.5983,
+      "step": 913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7341578935240922e-05,
+      "loss": 0.5008,
+      "step": 914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335303899354818e-05,
+      "loss": 0.4961,
+      "step": 915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732902260447433e-05,
+      "loss": 1.1088,
+      "step": 916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322735055959095e-05,
+      "loss": 0.3895,
+      "step": 917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316441259174092e-05,
+      "loss": 0.5335,
+      "step": 918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310141219489633e-05,
+      "loss": 0.2679,
+      "step": 919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303834942281346e-05,
+      "loss": 0.7464,
+      "step": 920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297522432930193e-05,
+      "loss": 0.6451,
+      "step": 921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729120369682244e-05,
+      "loss": 0.2306,
+      "step": 922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728487873934969e-05,
+      "loss": 0.1499,
+      "step": 923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278547565908823e-05,
+      "loss": 1.268,
+      "step": 924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7272210181902044e-05,
+      "loss": 0.7911,
+      "step": 925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726586659273686e-05,
+      "loss": 0.7405,
+      "step": 926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259516803826054e-05,
+      "loss": 0.4915,
+      "step": 927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7253160820587718e-05,
+      "loss": 0.3562,
+      "step": 928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7246798648445216e-05,
+      "loss": 0.812,
+      "step": 929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240430292827205e-05,
+      "loss": 0.2904,
+      "step": 930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234055759167602e-05,
+      "loss": 0.3208,
+      "step": 931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227675052905613e-05,
+      "loss": 0.3676,
+      "step": 932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72212881794857e-05,
+      "loss": 0.8372,
+      "step": 933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7214895144357592e-05,
+      "loss": 0.1929,
+      "step": 934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208495952976273e-05,
+      "loss": 0.4499,
+      "step": 935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202090610801975e-05,
+      "loss": 0.9155,
+      "step": 936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195679123300192e-05,
+      "loss": 0.7885,
+      "step": 937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189261495941648e-05,
+      "loss": 0.6477,
+      "step": 938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182837734202316e-05,
+      "loss": 0.5288,
+      "step": 939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7176407843563398e-05,
+      "loss": 0.2888,
+      "step": 940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7169971829511326e-05,
+      "loss": 0.2581,
+      "step": 941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7163529697537756e-05,
+      "loss": 0.5386,
+      "step": 942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157081453139564e-05,
+      "loss": 0.3321,
+      "step": 943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150627101818848e-05,
+      "loss": 0.4012,
+      "step": 944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144166649082907e-05,
+      "loss": 0.7474,
+      "step": 945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7137700100444257e-05,
+      "loss": 0.4826,
+      "step": 946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131227461420605e-05,
+      "loss": 0.3807,
+      "step": 947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712474873753486e-05,
+      "loss": 0.4928,
+      "step": 948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118263934315122e-05,
+      "loss": 0.2193,
+      "step": 949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711177305729468e-05,
+      "loss": 0.4449,
+      "step": 950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105276112012008e-05,
+      "loss": 0.5865,
+      "step": 951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709877310401075e-05,
+      "loss": 0.6102,
+      "step": 952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7092264038839724e-05,
+      "loss": 0.247,
+      "step": 953
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7085748922052923e-05,
+      "loss": 0.6717,
+      "step": 954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079227759209503e-05,
+      "loss": 0.0977,
+      "step": 955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7072700555873774e-05,
+      "loss": 0.2927,
+      "step": 956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066167317615203e-05,
+      "loss": 0.6452,
+      "step": 957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059628050008403e-05,
+      "loss": 0.4879,
+      "step": 958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053082758633138e-05,
+      "loss": 1.0826,
+      "step": 959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046531449074305e-05,
+      "loss": 0.7062,
+      "step": 960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039974126921946e-05,
+      "loss": 0.5539,
+      "step": 961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703341079777122e-05,
+      "loss": 0.7427,
+      "step": 962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7026841467222425e-05,
+      "loss": 0.5197,
+      "step": 963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7020266140880967e-05,
+      "loss": 0.512,
+      "step": 964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013684824357376e-05,
+      "loss": 0.7734,
+      "step": 965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007097523267292e-05,
+      "loss": 1.1205,
+      "step": 966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000504243231466e-05,
+      "loss": 0.2984,
+      "step": 967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6993904989875737e-05,
+      "loss": 0.2696,
+      "step": 968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6987299768831057e-05,
+      "loss": 0.503,
+      "step": 969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6980688585733456e-05,
+      "loss": 0.6177,
+      "step": 970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6974071446224066e-05,
+      "loss": 0.1667,
+      "step": 971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967448355949087e-05,
+      "loss": 0.1179,
+      "step": 972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960819320559806e-05,
+      "loss": 0.3397,
+      "step": 973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954184345712575e-05,
+      "loss": 0.3598,
+      "step": 974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6947543437068822e-05,
+      "loss": 1.3164,
+      "step": 975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694089660029504e-05,
+      "loss": 0.2598,
+      "step": 976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934243841062767e-05,
+      "loss": 0.7593,
+      "step": 977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927585165048604e-05,
+      "loss": 0.3707,
+      "step": 978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920920577934202e-05,
+      "loss": 0.3514,
+      "step": 979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691425008540625e-05,
+      "loss": 0.2938,
+      "step": 980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690757369315648e-05,
+      "loss": 0.8852,
+      "step": 981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690089140688166e-05,
+      "loss": 0.3579,
+      "step": 982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689420323228358e-05,
+      "loss": 1.406,
+      "step": 983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6887509175069057e-05,
+      "loss": 0.2971,
+      "step": 984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880809240949934e-05,
+      "loss": 0.8343,
+      "step": 985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687410343564306e-05,
+      "loss": 0.2563,
+      "step": 986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68673917648703e-05,
+      "loss": 0.2604,
+      "step": 987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6860674234358517e-05,
+      "loss": 0.2,
+      "step": 988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853950849839582e-05,
+      "loss": 0.2712,
+      "step": 989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847221617050354e-05,
+      "loss": 0.9919,
+      "step": 990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840486541732685e-05,
+      "loss": 0.8363,
+      "step": 991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833745629633414e-05,
+      "loss": 0.8502,
+      "step": 992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682699888650436e-05,
+      "loss": 0.6588,
+      "step": 993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682024631810231e-05,
+      "loss": 0.3248,
+      "step": 994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681348793018904e-05,
+      "loss": 0.4144,
+      "step": 995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680672372853126e-05,
+      "loss": 0.2269,
+      "step": 996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679995371890068e-05,
+      "loss": 0.9505,
+      "step": 997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793177907073937e-05,
+      "loss": 0.6899,
+      "step": 998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786396298832622e-05,
+      "loss": 0.4291,
+      "step": 999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677960889996329e-05,
+      "loss": 0.2953,
+      "step": 1000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 0.5114,
+      "step": 1001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766016753511415e-05,
+      "loss": 0.4798,
+      "step": 1002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675921201752665e-05,
+      "loss": 0.6403,
+      "step": 1003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675240151410939e-05,
+      "loss": 0.2375,
+      "step": 1004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745585249070834e-05,
+      "loss": 0.529,
+      "step": 1005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6738763228227094e-05,
+      "loss": 0.4758,
+      "step": 1006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731935457399205e-05,
+      "loss": 0.3796,
+      "step": 1007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725101942413085e-05,
+      "loss": 0.1761,
+      "step": 1008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718262689099577e-05,
+      "loss": 0.4259,
+      "step": 1009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711417703294404e-05,
+      "loss": 0.232,
+      "step": 1010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704566990838192e-05,
+      "loss": 0.296,
+      "step": 1011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697710557576448e-05,
+      "loss": 0.7192,
+      "step": 1012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6690848409359555e-05,
+      "loss": 0.36,
+      "step": 1013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668398055204278e-05,
+      "loss": 0.234,
+      "step": 1014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677106991486264e-05,
+      "loss": 0.761,
+      "step": 1015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6670227733555004e-05,
+      "loss": 1.0939,
+      "step": 1016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663342784118865e-05,
+      "loss": 0.4236,
+      "step": 1017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656452149052568e-05,
+      "loss": 0.2222,
+      "step": 1018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649555834235686e-05,
+      "loss": 0.4006,
+      "step": 1019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642653845552643e-05,
+      "loss": 0.3524,
+      "step": 1020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663574618889268e-05,
+      "loss": 0.3761,
+      "step": 1021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628832870149913e-05,
+      "loss": 0.185,
+      "step": 1022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662191389522326e-05,
+      "loss": 0.296,
+      "step": 1023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6614989270016474e-05,
+      "loss": 0.2839,
+      "step": 1024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660805900043813e-05,
+      "loss": 0.269,
+      "step": 1025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6601123092401624e-05,
+      "loss": 0.1902,
+      "step": 1026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659418155182515e-05,
+      "loss": 0.3372,
+      "step": 1027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6587234384631718e-05,
+      "loss": 0.9263,
+      "step": 1028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658028159674914e-05,
+      "loss": 1.3338,
+      "step": 1029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657332319411002e-05,
+      "loss": 0.4584,
+      "step": 1030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566359182651758e-05,
+      "loss": 0.8048,
+      "step": 1031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559389568316525e-05,
+      "loss": 0.4604,
+      "step": 1032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655241435705129e-05,
+      "loss": 0.5562,
+      "step": 1033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6545433554807796e-05,
+      "loss": 0.2027,
+      "step": 1034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653844716754254e-05,
+      "loss": 0.2412,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6531455201216803e-05,
+      "loss": 0.3714,
+      "step": 1036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524457661796626e-05,
+      "loss": 0.5173,
+      "step": 1037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517454555252787e-05,
+      "loss": 0.2397,
+      "step": 1038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6510445887560838e-05,
+      "loss": 0.2787,
+      "step": 1039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503431664701052e-05,
+      "loss": 0.5624,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6496411892658465e-05,
+      "loss": 0.405,
+      "step": 1041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648938657742283e-05,
+      "loss": 0.679,
+      "step": 1042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482355724988646e-05,
+      "loss": 0.5183,
+      "step": 1043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647531934135512e-05,
+      "loss": 0.4921,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646827743252619e-05,
+      "loss": 0.1505,
+      "step": 1045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461230004510508e-05,
+      "loss": 0.3436,
+      "step": 1046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6454177063321425e-05,
+      "loss": 0.2171,
+      "step": 1047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447118614977012e-05,
+      "loss": 0.559,
+      "step": 1048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440054665500024e-05,
+      "loss": 0.5167,
+      "step": 1049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643298522091792e-05,
+      "loss": 0.8623,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642591028726285e-05,
+      "loss": 0.4425,
+      "step": 1051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418829870571632e-05,
+      "loss": 1.509,
+      "step": 1052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641174397688578e-05,
+      "loss": 0.3521,
+      "step": 1053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640465261225147e-05,
+      "loss": 0.3757,
+      "step": 1054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397555782719556e-05,
+      "loss": 0.3119,
+      "step": 1055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639045349434554e-05,
+      "loss": 1.0354,
+      "step": 1056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63833457531896e-05,
+      "loss": 0.1494,
+      "step": 1057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6376232565316557e-05,
+      "loss": 0.355,
+      "step": 1058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369113936795876e-05,
+      "loss": 0.1766,
+      "step": 1059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6361989873701668e-05,
+      "loss": 0.3535,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6354860382112692e-05,
+      "loss": 0.4244,
+      "step": 1061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6347725468112316e-05,
+      "loss": 0.6859,
+      "step": 1062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6340585137788557e-05,
+      "loss": 0.5142,
+      "step": 1063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633343939723404e-05,
+      "loss": 0.9777,
+      "step": 1064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6326288252546008e-05,
+      "loss": 0.331,
+      "step": 1065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6319131709826325e-05,
+      "loss": 0.4213,
+      "step": 1066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311969775181447e-05,
+      "loss": 1.3164,
+      "step": 1067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304802454722447e-05,
+      "loss": 0.1629,
+      "step": 1068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297629754564973e-05,
+      "loss": 0.234,
+      "step": 1069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6290451680829283e-05,
+      "loss": 0.5424,
+      "step": 1070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283268239640203e-05,
+      "loss": 0.1311,
+      "step": 1071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276079437127155e-05,
+      "loss": 0.6968,
+      "step": 1072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6268885279424126e-05,
+      "loss": 0.1645,
+      "step": 1073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261685772669675e-05,
+      "loss": 0.2515,
+      "step": 1074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254480923006924e-05,
+      "loss": 0.9348,
+      "step": 1075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247270736583555e-05,
+      "loss": 0.3203,
+      "step": 1076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6240055219551805e-05,
+      "loss": 0.1939,
+      "step": 1077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232834378068454e-05,
+      "loss": 0.3996,
+      "step": 1078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225608218294832e-05,
+      "loss": 0.506,
+      "step": 1079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62183767463968e-05,
+      "loss": 0.9117,
+      "step": 1080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621113996854476e-05,
+      "loss": 0.1548,
+      "step": 1081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620389789091364e-05,
+      "loss": 0.7232,
+      "step": 1082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619665051968288e-05,
+      "loss": 0.3636,
+      "step": 1083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189397861036448e-05,
+      "loss": 0.549,
+      "step": 1084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182139921162817e-05,
+      "loss": 0.2949,
+      "step": 1085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617487670625497e-05,
+      "loss": 0.4517,
+      "step": 1086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167608222510395e-05,
+      "loss": 0.8119,
+      "step": 1087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616033447613106e-05,
+      "loss": 0.2864,
+      "step": 1088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153055473323447e-05,
+      "loss": 0.5526,
+      "step": 1089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6145771220298502e-05,
+      "loss": 0.364,
+      "step": 1090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613848172327166e-05,
+      "loss": 0.2532,
+      "step": 1091
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131186988462835e-05,
+      "loss": 0.4267,
+      "step": 1092
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6123887022096397e-05,
+      "loss": 0.9294,
+      "step": 1093
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116581830401193e-05,
+      "loss": 0.6848,
+      "step": 1094
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109271419610526e-05,
+      "loss": 0.313,
+      "step": 1095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101955795962142e-05,
+      "loss": 0.5518,
+      "step": 1096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6094634965698248e-05,
+      "loss": 0.8516,
+      "step": 1097
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087308935065488e-05,
+      "loss": 0.3035,
+      "step": 1098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079977710314944e-05,
+      "loss": 0.3815,
+      "step": 1099
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6072641297702128e-05,
+      "loss": 0.3176,
+      "step": 1100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6065299703486986e-05,
+      "loss": 0.5852,
+      "step": 1101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605795293393387e-05,
+      "loss": 0.2636,
+      "step": 1102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050600995311565e-05,
+      "loss": 0.1709,
+      "step": 1103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6043243893893256e-05,
+      "loss": 0.1392,
+      "step": 1104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603588163595654e-05,
+      "loss": 0.7295,
+      "step": 1105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028514227783408e-05,
+      "loss": 0.2931,
+      "step": 1106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602114167566025e-05,
+      "loss": 0.2194,
+      "step": 1107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601376398587784e-05,
+      "loss": 0.5313,
+      "step": 1108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6006381164731338e-05,
+      "loss": 0.2633,
+      "step": 1109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599899321852029e-05,
+      "loss": 0.2593,
+      "step": 1110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5991600153548602e-05,
+      "loss": 0.2791,
+      "step": 1111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5984201976124554e-05,
+      "loss": 0.6129,
+      "step": 1112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976798692560796e-05,
+      "loss": 0.405,
+      "step": 1113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596939030917432e-05,
+      "loss": 0.3343,
+      "step": 1114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5961976832286478e-05,
+      "loss": 0.2746,
+      "step": 1115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5954558268222974e-05,
+      "loss": 1.6494,
+      "step": 1116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947134623313834e-05,
+      "loss": 1.2347,
+      "step": 1117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593970590389344e-05,
+      "loss": 0.5798,
+      "step": 1118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932272116300493e-05,
+      "loss": 0.7597,
+      "step": 1119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924833266878015e-05,
+      "loss": 0.2209,
+      "step": 1120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5917389361973365e-05,
+      "loss": 0.4812,
+      "step": 1121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590994040793819e-05,
+      "loss": 0.9929,
+      "step": 1122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590248641112847e-05,
+      "loss": 0.6285,
+      "step": 1123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895027377904468e-05,
+      "loss": 0.4043,
+      "step": 1124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887563314630753e-05,
+      "loss": 0.1464,
+      "step": 1125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880094227676192e-05,
+      "loss": 0.4032,
+      "step": 1126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587262012341393e-05,
+      "loss": 0.967,
+      "step": 1127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5865141008221394e-05,
+      "loss": 0.4589,
+      "step": 1128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5857656888480287e-05,
+      "loss": 0.373,
+      "step": 1129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585016777057659e-05,
+      "loss": 0.2775,
+      "step": 1130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842673660900536e-05,
+      "loss": 0.3065,
+      "step": 1131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5835174565846624e-05,
+      "loss": 0.3055,
+      "step": 1132
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582767049181361e-05,
+      "loss": 0.3475,
+      "step": 1133
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582016144520449e-05,
+      "loss": 0.2375,
+      "step": 1134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812647432426512e-05,
+      "loss": 0.3172,
+      "step": 1135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805128459891154e-05,
+      "loss": 0.1875,
+      "step": 1136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5797604534014134e-05,
+      "loss": 0.4489,
+      "step": 1137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5790075661215384e-05,
+      "loss": 0.4187,
+      "step": 1138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782541847919075e-05,
+      "loss": 1.0744,
+      "step": 1139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5775003100553577e-05,
+      "loss": 0.0976,
+      "step": 1140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576745942555148e-05,
+      "loss": 0.8746,
+      "step": 1141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759910829349568e-05,
+      "loss": 0.4751,
+      "step": 1142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575235731838884e-05,
+      "loss": 0.4353,
+      "step": 1143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5744798899114476e-05,
+      "loss": 1.0333,
+      "step": 1144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573723557797585e-05,
+      "loss": 0.1906,
+      "step": 1145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572966736142651e-05,
+      "loss": 0.8117,
+      "step": 1146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722094255924198e-05,
+      "loss": 0.7816,
+      "step": 1147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571451626793081e-05,
+      "loss": 0.2725,
+      "step": 1148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5706933403912415e-05,
+      "loss": 0.1278,
+      "step": 1149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569934567033925e-05,
+      "loss": 0.8214,
+      "step": 1150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691753073685692e-05,
+      "loss": 0.1579,
+      "step": 1151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568415562043028e-05,
+      "loss": 0.4085,
+      "step": 1152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5676553317055694e-05,
+      "loss": 0.2239,
+      "step": 1153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668946170048746e-05,
+      "loss": 0.3476,
+      "step": 1154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566133418590039e-05,
+      "loss": 0.7637,
+      "step": 1155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653717371105702e-05,
+      "loss": 0.1809,
+      "step": 1156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564609573216388e-05,
+      "loss": 0.4144,
+      "step": 1157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5638469275578244e-05,
+      "loss": 0.5008,
+      "step": 1158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5630838007856214e-05,
+      "loss": 0.1352,
+      "step": 1159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623201935509322e-05,
+      "loss": 0.3947,
+      "step": 1160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615561065053208e-05,
+      "loss": 0.2785,
+      "step": 1161
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560791540300758e-05,
+      "loss": 0.126,
+      "step": 1162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600264955896273e-05,
+      "loss": 0.3726,
+      "step": 1163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5592609730247167e-05,
+      "loss": 0.147,
+      "step": 1164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558494973259224e-05,
+      "loss": 1.1037,
+      "step": 1165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5577284969467545e-05,
+      "loss": 0.9768,
+      "step": 1166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569615447413186e-05,
+      "loss": 0.3222,
+      "step": 1167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561941172973336e-05,
+      "loss": 0.4005,
+      "step": 1168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555426215269623e-05,
+      "loss": 0.7296,
+      "step": 1169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554657839313413e-05,
+      "loss": 0.1303,
+      "step": 1170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553888990084338e-05,
+      "loss": 0.3538,
+      "step": 1171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553119668238432e-05,
+      "loss": 0.4644,
+      "step": 1172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5523498744321352e-05,
+      "loss": 0.3474,
+      "step": 1173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551579609322289e-05,
+      "loss": 0.1631,
+      "step": 1174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5508088735661378e-05,
+      "loss": 0.2207,
+      "step": 1175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550037667821327e-05,
+      "loss": 0.3619,
+      "step": 1176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5492659927459033e-05,
+      "loss": 0.4734,
+      "step": 1177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5484938489983144e-05,
+      "loss": 1.4234,
+      "step": 1178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547721237237407e-05,
+      "loss": 0.6073,
+      "step": 1179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 0.1874,
+      "step": 1180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461746123130202e-05,
+      "loss": 0.2367,
+      "step": 1181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54540060046923e-05,
+      "loss": 0.3117,
+      "step": 1182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544626123251497e-05,
+      "loss": 0.3457,
+      "step": 1183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5438511813206596e-05,
+      "loss": 0.1941,
+      "step": 1184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430757753379527e-05,
+      "loss": 0.3828,
+      "step": 1185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422999059650064e-05,
+      "loss": 0.7411,
+      "step": 1186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541523573863847e-05,
+      "loss": 0.8093,
+      "step": 1187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407467796968957e-05,
+      "loss": 0.2005,
+      "step": 1188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539969524126967e-05,
+      "loss": 0.1428,
+      "step": 1189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391918078172698e-05,
+      "loss": 0.503,
+      "step": 1190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5384136314314065e-05,
+      "loss": 0.6351,
+      "step": 1191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537634995633371e-05,
+      "loss": 0.6601,
+      "step": 1192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536855901087551e-05,
+      "loss": 0.3239,
+      "step": 1193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536076348458723e-05,
+      "loss": 0.5818,
+      "step": 1194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5352963384120567e-05,
+      "loss": 0.6572,
+      "step": 1195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534515871613111e-05,
+      "loss": 0.197,
+      "step": 1196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5337349487278346e-05,
+      "loss": 0.1078,
+      "step": 1197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532953570422566e-05,
+      "loss": 0.5086,
+      "step": 1198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321717373640313e-05,
+      "loss": 0.1712,
+      "step": 1199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313894502193457e-05,
+      "loss": 0.6406,
+      "step": 1200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530606709656011e-05,
+      "loss": 0.1869,
+      "step": 1201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298235163419162e-05,
+      "loss": 0.55,
+      "step": 1202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5290398709453363e-05,
+      "loss": 0.3021,
+      "step": 1203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282557741349328e-05,
+      "loss": 0.3505,
+      "step": 1204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5274712265797523e-05,
+      "loss": 0.6526,
+      "step": 1205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5266862289492247e-05,
+      "loss": 0.9809,
+      "step": 1206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5259007819131658e-05,
+      "loss": 0.2597,
+      "step": 1207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5251148861417733e-05,
+      "loss": 0.313,
+      "step": 1208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243285423056287e-05,
+      "loss": 0.3188,
+      "step": 1209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5235417510756954e-05,
+      "loss": 0.2755,
+      "step": 1210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522754513123319e-05,
+      "loss": 1.4397,
+      "step": 1211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219668291202258e-05,
+      "loss": 0.5638,
+      "step": 1212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211786997385231e-05,
+      "loss": 0.6449,
+      "step": 1213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203901256506979e-05,
+      "loss": 0.4756,
+      "step": 1214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196011075296164e-05,
+      "loss": 0.7454,
+      "step": 1215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188116460485245e-05,
+      "loss": 0.2675,
+      "step": 1216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518021741881046e-05,
+      "loss": 0.4457,
+      "step": 1217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517231395701182e-05,
+      "loss": 0.878,
+      "step": 1218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164406081833117e-05,
+      "loss": 0.2405,
+      "step": 1219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5156493800021896e-05,
+      "loss": 0.8056,
+      "step": 1220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514857711832948e-05,
+      "loss": 0.1204,
+      "step": 1221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140656043510919e-05,
+      "loss": 0.3887,
+      "step": 1222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5132730582325047e-05,
+      "loss": 0.112,
+      "step": 1223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124800741534407e-05,
+      "loss": 0.2603,
+      "step": 1224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116866527905303e-05,
+      "loss": 0.1993,
+      "step": 1225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108927948207752e-05,
+      "loss": 0.199,
+      "step": 1226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5100985009215519e-05,
+      "loss": 0.3893,
+      "step": 1227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093037717706063e-05,
+      "loss": 0.6984,
+      "step": 1228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085086080460573e-05,
+      "loss": 0.436,
+      "step": 1229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077130104263944e-05,
+      "loss": 0.356,
+      "step": 1230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506916979590477e-05,
+      "loss": 0.131,
+      "step": 1231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061205162175343e-05,
+      "loss": 0.9507,
+      "step": 1232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5053236209871647e-05,
+      "loss": 0.5508,
+      "step": 1233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045262945793342e-05,
+      "loss": 0.759,
+      "step": 1234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037285376743787e-05,
+      "loss": 0.0959,
+      "step": 1235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029303509529991e-05,
+      "loss": 0.1046,
+      "step": 1236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021317350962648e-05,
+      "loss": 0.3742,
+      "step": 1237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013326907856105e-05,
+      "loss": 0.6506,
+      "step": 1238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005332187028367e-05,
+      "loss": 0.2179,
+      "step": 1239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4997333195301088e-05,
+      "loss": 0.5024,
+      "step": 1240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498932993949957e-05,
+      "loss": 0.9421,
+      "step": 1241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981322426452747e-05,
+      "loss": 0.2393,
+      "step": 1242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973310662993195e-05,
+      "loss": 0.2156,
+      "step": 1243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4965294655957103e-05,
+      "loss": 0.2808,
+      "step": 1244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4957274412184295e-05,
+      "loss": 0.291,
+      "step": 1245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949249938518203e-05,
+      "loss": 0.3648,
+      "step": 1246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4941221241805868e-05,
+      "loss": 0.3114,
+      "step": 1247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933188328897933e-05,
+      "loss": 0.4169,
+      "step": 1248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492515120664865e-05,
+      "loss": 0.7632,
+      "step": 1249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917109881915844e-05,
+      "loss": 0.3248,
+      "step": 1250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490906436156094e-05,
+      "loss": 0.4931,
+      "step": 1251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901014652448939e-05,
+      "loss": 0.6283,
+      "step": 1252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4892960761448417e-05,
+      "loss": 1.2161,
+      "step": 1253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884902695431516e-05,
+      "loss": 0.5371,
+      "step": 1254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4876840461273939e-05,
+      "loss": 0.4321,
+      "step": 1255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486877406585495e-05,
+      "loss": 0.2438,
+      "step": 1256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4860703516057364e-05,
+      "loss": 0.9911,
+      "step": 1257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4852628818767536e-05,
+      "loss": 0.749,
+      "step": 1258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844549980875363e-05,
+      "loss": 0.8523,
+      "step": 1259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836467009274276e-05,
+      "loss": 0.1898,
+      "step": 1260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482837991086123e-05,
+      "loss": 0.5134,
+      "step": 1261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4820288692536702e-05,
+      "loss": 0.1109,
+      "step": 1262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4812193361204689e-05,
+      "loss": 0.1968,
+      "step": 1263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804093923772691e-05,
+      "loss": 0.7531,
+      "step": 1264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795990387151719e-05,
+      "loss": 0.3377,
+      "step": 1265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787882758256271e-05,
+      "loss": 0.1149,
+      "step": 1266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779771044004347e-05,
+      "loss": 0.4966,
+      "step": 1267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771655251317426e-05,
+      "loss": 0.1088,
+      "step": 1268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4763535387120475e-05,
+      "loss": 0.1574,
+      "step": 1269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755411458341924e-05,
+      "loss": 0.5213,
+      "step": 1270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747283471913685e-05,
+      "loss": 0.3268,
+      "step": 1271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4739151434771114e-05,
+      "loss": 0.5447,
+      "step": 1272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4731015353853046e-05,
+      "loss": 0.3751,
+      "step": 1273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722875236101746e-05,
+      "loss": 0.5148,
+      "step": 1274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4714731088462935e-05,
+      "loss": 0.2632,
+      "step": 1275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4706582917885767e-05,
+      "loss": 0.9428,
+      "step": 1276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698430731322834e-05,
+      "loss": 0.2515,
+      "step": 1277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469027453573015e-05,
+      "loss": 0.2084,
+      "step": 1278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4682114338067152e-05,
+      "loss": 0.1272,
+      "step": 1279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4673950145296691e-05,
+      "loss": 0.8026,
+      "step": 1280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4665781964385028e-05,
+      "loss": 0.6929,
+      "step": 1281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657609802301828e-05,
+      "loss": 0.5626,
+      "step": 1282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4649433666020147e-05,
+      "loss": 1.071,
+      "step": 1283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464125356251644e-05,
+      "loss": 0.4505,
+      "step": 1284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633069498770544e-05,
+      "loss": 0.7181,
+      "step": 1285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624881481765672e-05,
+      "loss": 0.6254,
+      "step": 1286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4616689518488417e-05,
+      "loss": 0.6665,
+      "step": 1287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608493615928725e-05,
+      "loss": 0.4451,
+      "step": 1288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600293781079923e-05,
+      "loss": 0.517,
+      "step": 1289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4592090020938683e-05,
+      "loss": 0.1519,
+      "step": 1290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583882342505025e-05,
+      "loss": 0.6221,
+      "step": 1291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575670752782314e-05,
+      "loss": 0.2041,
+      "step": 1292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567455258777255e-05,
+      "loss": 1.0065,
+      "step": 1293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559235867499874e-05,
+      "loss": 0.3182,
+      "step": 1294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551012585963542e-05,
+      "loss": 0.9487,
+      "step": 1295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542785421184932e-05,
+      "loss": 1.0565,
+      "step": 1296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534554380184039e-05,
+      "loss": 0.5895,
+      "step": 1297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526319469984158e-05,
+      "loss": 0.3009,
+      "step": 1298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518080697611896e-05,
+      "loss": 0.9938,
+      "step": 1299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509838070097147e-05,
+      "loss": 0.1985,
+      "step": 1300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4501591594473098e-05,
+      "loss": 0.3466,
+      "step": 1301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493341277776218e-05,
+      "loss": 0.412,
+      "step": 1302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485087127046256e-05,
+      "loss": 0.3854,
+      "step": 1303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447682914932623e-05,
+      "loss": 0.3128,
+      "step": 1304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4468567351662423e-05,
+      "loss": 0.7328,
+      "step": 1305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460301741104381e-05,
+      "loss": 0.7401,
+      "step": 1306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44520323247049e-05,
+      "loss": 0.1345,
+      "step": 1307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4443759109520023e-05,
+      "loss": 0.4709,
+      "step": 1308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4435482102609038e-05,
+      "loss": 0.731,
+      "step": 1309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427201311034467e-05,
+      "loss": 0.7736,
+      "step": 1310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418916741862057e-05,
+      "loss": 0.3091,
+      "step": 1311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4410628402160785e-05,
+      "loss": 0.1877,
+      "step": 1312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4402336299002842e-05,
+      "loss": 0.1872,
+      "step": 1313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394040439463628e-05,
+      "loss": 0.403,
+      "step": 1314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4385740830621755e-05,
+      "loss": 0.6437,
+      "step": 1315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377437479559021e-05,
+      "loss": 0.2014,
+      "step": 1316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369130393360437e-05,
+      "loss": 0.4357,
+      "step": 1317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4360819579114185e-05,
+      "loss": 0.5767,
+      "step": 1318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352505043911634e-05,
+      "loss": 0.4168,
+      "step": 1319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344186794847326e-05,
+      "loss": 0.3405,
+      "step": 1320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433586483901897e-05,
+      "loss": 0.8633,
+      "step": 1321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327539183527447e-05,
+      "loss": 0.3709,
+      "step": 1322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319209835476783e-05,
+      "loss": 0.8968,
+      "step": 1323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310876801974165e-05,
+      "loss": 0.2119,
+      "step": 1324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302540090129916e-05,
+      "loss": 0.6487,
+      "step": 1325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294199707057505e-05,
+      "loss": 0.83,
+      "step": 1326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285855659873532e-05,
+      "loss": 1.3095,
+      "step": 1327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277507955697716e-05,
+      "loss": 0.5642,
+      "step": 1328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269156601652903e-05,
+      "loss": 0.5341,
+      "step": 1329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4260801604865057e-05,
+      "loss": 0.2293,
+      "step": 1330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425244297246324e-05,
+      "loss": 0.4419,
+      "step": 1331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424408071157963e-05,
+      "loss": 0.4636,
+      "step": 1332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235714829349483e-05,
+      "loss": 0.2216,
+      "step": 1333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422734533291116e-05,
+      "loss": 0.306,
+      "step": 1334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4218972229406103e-05,
+      "loss": 0.5883,
+      "step": 1335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210595525978826e-05,
+      "loss": 0.6689,
+      "step": 1336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202215229776917e-05,
+      "loss": 0.4744,
+      "step": 1337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4193831347951034e-05,
+      "loss": 0.2484,
+      "step": 1338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4185443887654891e-05,
+      "loss": 0.4099,
+      "step": 1339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4177052856045256e-05,
+      "loss": 0.6044,
+      "step": 1340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4168658260281944e-05,
+      "loss": 0.4376,
+      "step": 1341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160260107527812e-05,
+      "loss": 0.1592,
+      "step": 1342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4151858404948748e-05,
+      "loss": 0.3477,
+      "step": 1343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4143453159713675e-05,
+      "loss": 0.275,
+      "step": 1344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135044378994538e-05,
+      "loss": 0.3604,
+      "step": 1345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126632069966292e-05,
+      "loss": 0.4929,
+      "step": 1346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411821623980691e-05,
+      "loss": 0.2275,
+      "step": 1347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4109796895697368e-05,
+      "loss": 0.2305,
+      "step": 1348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101374044821639e-05,
+      "loss": 0.2232,
+      "step": 1349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092947694366687e-05,
+      "loss": 0.1565,
+      "step": 1350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084517851522466e-05,
+      "loss": 0.4107,
+      "step": 1351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4076084523481905e-05,
+      "loss": 1.1107,
+      "step": 1352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067647717440909e-05,
+      "loss": 0.2885,
+      "step": 1353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4059207440598357e-05,
+      "loss": 1.0043,
+      "step": 1354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050763700156074e-05,
+      "loss": 0.5383,
+      "step": 1355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042316503318858e-05,
+      "loss": 0.4583,
+      "step": 1356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033865857294447e-05,
+      "loss": 0.2462,
+      "step": 1357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402541176929352e-05,
+      "loss": 0.4184,
+      "step": 1358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 0.2311,
+      "step": 1359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400849329621953e-05,
+      "loss": 0.5318,
+      "step": 1360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400002892558249e-05,
+      "loss": 1.028,
+      "step": 1361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991561141840976e-05,
+      "loss": 0.5096,
+      "step": 1362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983089952220289e-05,
+      "loss": 0.3528,
+      "step": 1363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397461536394864e-05,
+      "loss": 0.4019,
+      "step": 1364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966137384257145e-05,
+      "loss": 0.3993,
+      "step": 1365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957656020379806e-05,
+      "loss": 0.4237,
+      "step": 1366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3949171279553515e-05,
+      "loss": 0.5573,
+      "step": 1367
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394068316901805e-05,
+      "loss": 0.3234,
+      "step": 1368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932191696016055e-05,
+      "loss": 0.5021,
+      "step": 1369
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923696867793055e-05,
+      "loss": 0.1323,
+      "step": 1370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915198691597427e-05,
+      "loss": 0.5261,
+      "step": 1371
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390669717468041e-05,
+      "loss": 0.8282,
+      "step": 1372
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898192324296096e-05,
+      "loss": 1.4843,
+      "step": 1373
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3889684147701417e-05,
+      "loss": 0.548,
+      "step": 1374
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388117265215614e-05,
+      "loss": 0.4883,
+      "step": 1375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872657844922879e-05,
+      "loss": 0.4903,
+      "step": 1376
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3864139733267047e-05,
+      "loss": 0.1483,
+      "step": 1377
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855618324456912e-05,
+      "loss": 0.386,
+      "step": 1378
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3847093625763517e-05,
+      "loss": 0.8692,
+      "step": 1379
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838565644460745e-05,
+      "loss": 0.5488,
+      "step": 1380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383003438782526e-05,
+      "loss": 0.2931,
+      "step": 1381
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382149986313653e-05,
+      "loss": 0.1651,
+      "step": 1382
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3812962077676801e-05,
+      "loss": 0.756,
+      "step": 1383
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3804421038731122e-05,
+      "loss": 0.1457,
+      "step": 1384
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795876753587292e-05,
+      "loss": 0.5986,
+      "step": 1385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37873292295359e-05,
+      "loss": 0.285,
+      "step": 1386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377877847387029e-05,
+      "loss": 0.7218,
+      "step": 1387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770224493886565e-05,
+      "loss": 0.1031,
+      "step": 1388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761667296883576e-05,
+      "loss": 0.16,
+      "step": 1389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3753106890162927e-05,
+      "loss": 0.437,
+      "step": 1390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374454328102895e-05,
+      "loss": 0.2741,
+      "step": 1391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3735976476788722e-05,
+      "loss": 0.3548,
+      "step": 1392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727406484752033e-05,
+      "loss": 0.2629,
+      "step": 1393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3718833312231405e-05,
+      "loss": 0.4332,
+      "step": 1394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3710256966542065e-05,
+      "loss": 0.2716,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701677455001954e-05,
+      "loss": 0.3951,
+      "step": 1396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693094784931708e-05,
+      "loss": 0.2362,
+      "step": 1397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684508963654667e-05,
+      "loss": 0.5,
+      "step": 1398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675919998496846e-05,
+      "loss": 0.4882,
+      "step": 1399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667327896786959e-05,
+      "loss": 0.2829,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3658732665856382e-05,
+      "loss": 0.0615,
+      "step": 1401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650134313039169e-05,
+      "loss": 0.2346,
+      "step": 1402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364153284567204e-05,
+      "loss": 0.9363,
+      "step": 1403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632928271094366e-05,
+      "loss": 0.1609,
+      "step": 1404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624320596648166e-05,
+      "loss": 0.2975,
+      "step": 1405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3615709829678122e-05,
+      "loss": 0.4892,
+      "step": 1406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360709597753153e-05,
+      "loss": 0.6872,
+      "step": 1407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598479047558341e-05,
+      "loss": 0.3281,
+      "step": 1408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589859047111118e-05,
+      "loss": 0.1236,
+      "step": 1409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358123598354505e-05,
+      "loss": 0.1059,
+      "step": 1410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572609864217934e-05,
+      "loss": 0.4426,
+      "step": 1411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563980696490184e-05,
+      "loss": 0.1024,
+      "step": 1412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555348487724805e-05,
+      "loss": 0.6201,
+      "step": 1413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546713245287407e-05,
+      "loss": 0.3084,
+      "step": 1414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3538074976546174e-05,
+      "loss": 0.5054,
+      "step": 1415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529433688871887e-05,
+      "loss": 0.5334,
+      "step": 1416
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520789389637898e-05,
+      "loss": 0.1875,
+      "step": 1417
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512142086220128e-05,
+      "loss": 0.3155,
+      "step": 1418
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3503491785997053e-05,
+      "loss": 0.5079,
+      "step": 1419
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3494838496349729e-05,
+      "loss": 0.0692,
+      "step": 1420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3486182224661732e-05,
+      "loss": 0.8875,
+      "step": 1421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477522978319208e-05,
+      "loss": 0.1765,
+      "step": 1422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468860764710835e-05,
+      "loss": 0.9716,
+      "step": 1423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3460195591227806e-05,
+      "loss": 0.5538,
+      "step": 1424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3451527465263867e-05,
+      "loss": 0.6105,
+      "step": 1425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442856394215262e-05,
+      "loss": 0.6027,
+      "step": 1426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434182385480756e-05,
+      "loss": 1.3598,
+      "step": 1427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3425505446461625e-05,
+      "loss": 0.1341,
+      "step": 1428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416825584561632e-05,
+      "loss": 0.5252,
+      "step": 1429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408142807187048e-05,
+      "loss": 0.2314,
+      "step": 1430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3399457121746626e-05,
+      "loss": 0.2478,
+      "step": 1431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3390768535651598e-05,
+      "loss": 0.5188,
+      "step": 1432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382077056315672e-05,
+      "loss": 0.3128,
+      "step": 1433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373382691155035e-05,
+      "loss": 0.9036,
+      "step": 1434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3364685447588315e-05,
+      "loss": 0.3115,
+      "step": 1435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335598533303662e-05,
+      "loss": 0.6824,
+      "step": 1436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347282354923486e-05,
+      "loss": 0.3012,
+      "step": 1437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333857652067491e-05,
+      "loss": 0.8136,
+      "step": 1438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332986783771932e-05,
+      "loss": 0.609,
+      "step": 1439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321156313487565e-05,
+      "loss": 0.3369,
+      "step": 1440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331244195541293e-05,
+      "loss": 0.2651,
+      "step": 1441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3303724770931123e-05,
+      "loss": 0.6238,
+      "step": 1442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3295004767480246e-05,
+      "loss": 0.3016,
+      "step": 1443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328628195250082e-05,
+      "loss": 0.4194,
+      "step": 1444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3277556333435757e-05,
+      "loss": 0.7472,
+      "step": 1445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268827917730374e-05,
+      "loss": 0.3536,
+      "step": 1446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3260096712832355e-05,
+      "loss": 0.9572,
+      "step": 1447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251362726191784e-05,
+      "loss": 0.5834,
+      "step": 1448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242625965261102e-05,
+      "loss": 0.6501,
+      "step": 1449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233886437495132e-05,
+      "loss": 1.308,
+      "step": 1450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225144150351042e-05,
+      "loss": 0.5441,
+      "step": 1451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3216399111288372e-05,
+      "loss": 0.5711,
+      "step": 1452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207651327768994e-05,
+      "loss": 0.8036,
+      "step": 1453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198900807257129e-05,
+      "loss": 0.6246,
+      "step": 1454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319014755721934e-05,
+      "loss": 0.7711,
+      "step": 1455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3181391585124503e-05,
+      "loss": 0.4695,
+      "step": 1456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3172632898443833e-05,
+      "loss": 0.2968,
+      "step": 1457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3163871504650851e-05,
+      "loss": 0.7136,
+      "step": 1458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315510741122139e-05,
+      "loss": 0.8905,
+      "step": 1459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146340625633594e-05,
+      "loss": 0.8559,
+      "step": 1460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313757115536789e-05,
+      "loss": 0.2384,
+      "step": 1461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3128799007907004e-05,
+      "loss": 0.5076,
+      "step": 1462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120024190735952e-05,
+      "loss": 0.6154,
+      "step": 1463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111246711342016e-05,
+      "loss": 0.6536,
+      "step": 1464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102466577214756e-05,
+      "loss": 0.3718,
+      "step": 1465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093683795845999e-05,
+      "loss": 0.3883,
+      "step": 1466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084898374729826e-05,
+      "loss": 0.8512,
+      "step": 1467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076110321362576e-05,
+      "loss": 0.4935,
+      "step": 1468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067319643242829e-05,
+      "loss": 0.402,
+      "step": 1469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058526347871407e-05,
+      "loss": 0.7046,
+      "step": 1470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3049730442751362e-05,
+      "loss": 0.73,
+      "step": 1471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304093193538798e-05,
+      "loss": 0.4127,
+      "step": 1472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303213083328876e-05,
+      "loss": 0.6732,
+      "step": 1473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023327143963415e-05,
+      "loss": 0.1006,
+      "step": 1474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014520874923877e-05,
+      "loss": 0.3701,
+      "step": 1475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005712033684263e-05,
+      "loss": 0.4334,
+      "step": 1476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996900627760897e-05,
+      "loss": 0.3931,
+      "step": 1477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988086664672285e-05,
+      "loss": 0.7097,
+      "step": 1478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2979270151939116e-05,
+      "loss": 0.7571,
+      "step": 1479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970451097084258e-05,
+      "loss": 0.4282,
+      "step": 1480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2961629507632743e-05,
+      "loss": 0.1421,
+      "step": 1481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952805391111767e-05,
+      "loss": 0.1284,
+      "step": 1482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943978755050688e-05,
+      "loss": 0.6884,
+      "step": 1483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935149606981008e-05,
+      "loss": 0.4077,
+      "step": 1484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292631795443637e-05,
+      "loss": 0.2639,
+      "step": 1485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917483804952562e-05,
+      "loss": 0.1815,
+      "step": 1486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908647166067496e-05,
+      "loss": 0.1279,
+      "step": 1487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2899808045321208e-05,
+      "loss": 0.4341,
+      "step": 1488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2890966450255862e-05,
+      "loss": 0.4138,
+      "step": 1489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882122388415716e-05,
+      "loss": 0.2477,
+      "step": 1490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.287327586734715e-05,
+      "loss": 0.7512,
+      "step": 1491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864426894598629e-05,
+      "loss": 0.6733,
+      "step": 1492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285557547772072e-05,
+      "loss": 0.6371,
+      "step": 1493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846721624266068e-05,
+      "loss": 0.8355,
+      "step": 1494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837865341789399e-05,
+      "loss": 0.5717,
+      "step": 1495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829006637847514e-05,
+      "loss": 0.4225,
+      "step": 1496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2820145519999285e-05,
+      "loss": 0.4878,
+      "step": 1497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811281995805626e-05,
+      "loss": 0.4217,
+      "step": 1498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802416072829524e-05,
+      "loss": 0.9069,
+      "step": 1499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793547758636002e-05,
+      "loss": 1.0626,
+      "step": 1500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278467706079213e-05,
+      "loss": 0.204,
+      "step": 1501
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2775803986867001e-05,
+      "loss": 0.4712,
+      "step": 1502
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2766928544431748e-05,
+      "loss": 0.4116,
+      "step": 1503
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275805074105951e-05,
+      "loss": 0.9017,
+      "step": 1504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2749170584325465e-05,
+      "loss": 0.1415,
+      "step": 1505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274028808180677e-05,
+      "loss": 0.625,
+      "step": 1506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2731403241082609e-05,
+      "loss": 0.1882,
+      "step": 1507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2722516069734142e-05,
+      "loss": 0.5748,
+      "step": 1508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2713626575344525e-05,
+      "loss": 1.2608,
+      "step": 1509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704734765498896e-05,
+      "loss": 1.0414,
+      "step": 1510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2695840647784378e-05,
+      "loss": 0.2487,
+      "step": 1511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2686944229790044e-05,
+      "loss": 0.3326,
+      "step": 1512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678045519106948e-05,
+      "loss": 0.2756,
+      "step": 1513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669144523328082e-05,
+      "loss": 1.2162,
+      "step": 1514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660241250048409e-05,
+      "loss": 0.9171,
+      "step": 1515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265133570686482e-05,
+      "loss": 0.5528,
+      "step": 1516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642427901376147e-05,
+      "loss": 0.5176,
+      "step": 1517
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2633517841183151e-05,
+      "loss": 0.3011,
+      "step": 1518
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624605533888526e-05,
+      "loss": 0.6916,
+      "step": 1519
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2615690987096866e-05,
+      "loss": 0.8515,
+      "step": 1520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606774208414694e-05,
+      "loss": 0.2761,
+      "step": 1521
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597855205450427e-05,
+      "loss": 0.0578,
+      "step": 1522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2588933985814377e-05,
+      "loss": 0.432,
+      "step": 1523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258001055711876e-05,
+      "loss": 0.2192,
+      "step": 1524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571084926977669e-05,
+      "loss": 0.6918,
+      "step": 1525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2562157103007069e-05,
+      "loss": 0.1687,
+      "step": 1526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553227092824812e-05,
+      "loss": 1.3548,
+      "step": 1527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.25442949040506e-05,
+      "loss": 0.914,
+      "step": 1528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2535360544306007e-05,
+      "loss": 0.2182,
+      "step": 1529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2526424021214452e-05,
+      "loss": 0.636,
+      "step": 1530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2517485342401201e-05,
+      "loss": 0.637,
+      "step": 1531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2508544515493356e-05,
+      "loss": 0.568,
+      "step": 1532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499601548119868e-05,
+      "loss": 0.5932,
+      "step": 1533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2490656447911489e-05,
+      "loss": 0.4308,
+      "step": 1534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481709222500813e-05,
+      "loss": 0.1591,
+      "step": 1535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2472759879522234e-05,
+      "loss": 0.2038,
+      "step": 1536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463808426611958e-05,
+      "loss": 0.2833,
+      "step": 1537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 0.143,
+      "step": 1538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445899221550137e-05,
+      "loss": 0.2639,
+      "step": 1539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2436941484679974e-05,
+      "loss": 0.2873,
+      "step": 1540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242798166844088e-05,
+      "loss": 0.2276,
+      "step": 1541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419019780477985e-05,
+      "loss": 0.5802,
+      "step": 1542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241005582843821e-05,
+      "loss": 0.376,
+      "step": 1543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240108981997022e-05,
+      "loss": 0.4817,
+      "step": 1544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392121762724443e-05,
+      "loss": 0.3023,
+      "step": 1545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383151664353048e-05,
+      "loss": 0.4989,
+      "step": 1546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374179532509958e-05,
+      "loss": 0.7939,
+      "step": 1547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365205374850814e-05,
+      "loss": 0.6385,
+      "step": 1548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356229199033008e-05,
+      "loss": 0.4047,
+      "step": 1549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347251012715629e-05,
+      "loss": 0.7271,
+      "step": 1550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2338270823559497e-05,
+      "loss": 0.2322,
+      "step": 1551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329288639227142e-05,
+      "loss": 0.5527,
+      "step": 1552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320304467382786e-05,
+      "loss": 0.3253,
+      "step": 1553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311318315692355e-05,
+      "loss": 0.4253,
+      "step": 1554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302330191823467e-05,
+      "loss": 0.2324,
+      "step": 1555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2293340103445409e-05,
+      "loss": 0.3603,
+      "step": 1556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284348058229158e-05,
+      "loss": 0.634,
+      "step": 1557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2275354063847358e-05,
+      "loss": 0.2208,
+      "step": 1558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266358127974312e-05,
+      "loss": 0.8217,
+      "step": 1559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257360258285981e-05,
+      "loss": 0.261,
+      "step": 1560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2248360462459979e-05,
+      "loss": 0.6419,
+      "step": 1561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239358748175556e-05,
+      "loss": 0.1157,
+      "step": 1562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230355123113612e-05,
+      "loss": 0.2895,
+      "step": 1563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2221349594956664e-05,
+      "loss": 0.7699,
+      "step": 1564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221234217138886e-05,
+      "loss": 0.854,
+      "step": 1565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203332860095967e-05,
+      "loss": 0.2786,
+      "step": 1566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194321668765357e-05,
+      "loss": 0.7919,
+      "step": 1567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185308605086004e-05,
+      "loss": 0.3102,
+      "step": 1568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2176293676748494e-05,
+      "loss": 0.3495,
+      "step": 1569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167276891444986e-05,
+      "loss": 0.2768,
+      "step": 1570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2158258256869238e-05,
+      "loss": 0.4896,
+      "step": 1571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2149237780716575e-05,
+      "loss": 0.2383,
+      "step": 1572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21402154706839e-05,
+      "loss": 0.5376,
+      "step": 1573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213119133446968e-05,
+      "loss": 1.2673,
+      "step": 1574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212216537977394e-05,
+      "loss": 0.2763,
+      "step": 1575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2113137614298253e-05,
+      "loss": 0.1787,
+      "step": 1576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104108045745746e-05,
+      "loss": 0.4641,
+      "step": 1577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2095076681821068e-05,
+      "loss": 0.1925,
+      "step": 1578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086043530230421e-05,
+      "loss": 0.3707,
+      "step": 1579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077008598681515e-05,
+      "loss": 0.1334,
+      "step": 1580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206797189488359e-05,
+      "loss": 0.445,
+      "step": 1581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205893342654739e-05,
+      "loss": 0.2741,
+      "step": 1582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204989320138517e-05,
+      "loss": 0.286,
+      "step": 1583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040851227110681e-05,
+      "loss": 0.3721,
+      "step": 1584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031807511439176e-05,
+      "loss": 0.5128,
+      "step": 1585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022762062087372e-05,
+      "loss": 0.7169,
+      "step": 1586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2013714886773492e-05,
+      "loss": 0.2556,
+      "step": 1587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200466599321721e-05,
+      "loss": 0.2081,
+      "step": 1588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1995615389139679e-05,
+      "loss": 0.4697,
+      "step": 1589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986563082263506e-05,
+      "loss": 0.2166,
+      "step": 1590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977509080312755e-05,
+      "loss": 0.2467,
+      "step": 1591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1968453391012928e-05,
+      "loss": 0.5578,
+      "step": 1592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959396022090984e-05,
+      "loss": 0.7259,
+      "step": 1593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1950336981275287e-05,
+      "loss": 0.5781,
+      "step": 1594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1941276276295659e-05,
+      "loss": 0.4387,
+      "step": 1595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1932213914883322e-05,
+      "loss": 0.7901,
+      "step": 1596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923149904770914e-05,
+      "loss": 0.2715,
+      "step": 1597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914084253692486e-05,
+      "loss": 0.5352,
+      "step": 1598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905016969383484e-05,
+      "loss": 0.3969,
+      "step": 1599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189594805958075e-05,
+      "loss": 0.2204,
+      "step": 1600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1886877532022512e-05,
+      "loss": 0.6501,
+      "step": 1601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877805394448378e-05,
+      "loss": 0.086,
+      "step": 1602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1868731654599332e-05,
+      "loss": 0.8418,
+      "step": 1603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859656320217723e-05,
+      "loss": 0.8093,
+      "step": 1604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185057939904726e-05,
+      "loss": 0.5939,
+      "step": 1605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841500898833005e-05,
+      "loss": 0.6849,
+      "step": 1606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1832420827321374e-05,
+      "loss": 1.5583,
+      "step": 1607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823339192260117e-05,
+      "loss": 0.1077,
+      "step": 1608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1814256001398319e-05,
+      "loss": 0.6313,
+      "step": 1609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805171262486397e-05,
+      "loss": 0.1588,
+      "step": 1610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1796084983276084e-05,
+      "loss": 0.7145,
+      "step": 1611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786997171520429e-05,
+      "loss": 0.1958,
+      "step": 1612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1777907834973787e-05,
+      "loss": 0.3471,
+      "step": 1613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176881698139182e-05,
+      "loss": 0.1769,
+      "step": 1614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759724618531475e-05,
+      "loss": 0.3692,
+      "step": 1615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1750630754150995e-05,
+      "loss": 0.6501,
+      "step": 1616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17415353960099e-05,
+      "loss": 0.3588,
+      "step": 1617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1732438551868987e-05,
+      "loss": 0.5295,
+      "step": 1618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172334022949032e-05,
+      "loss": 0.3094,
+      "step": 1619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714240436637224e-05,
+      "loss": 1.0336,
+      "step": 1620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705139181074276e-05,
+      "loss": 0.2226,
+      "step": 1621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696036470567309e-05,
+      "loss": 0.5309,
+      "step": 1622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1686932312883385e-05,
+      "loss": 0.4248,
+      "step": 1623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1677826715790816e-05,
+      "loss": 0.3545,
+      "step": 1624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166871968705913e-05,
+      "loss": 0.7086,
+      "step": 1625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165961123445908e-05,
+      "loss": 0.1383,
+      "step": 1626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1650501365762639e-05,
+      "loss": 0.1877,
+      "step": 1627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164139008874298e-05,
+      "loss": 0.3181,
+      "step": 1628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1632277411174484e-05,
+      "loss": 0.352,
+      "step": 1629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623163340832725e-05,
+      "loss": 0.4789,
+      "step": 1630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1614047885494463e-05,
+      "loss": 0.6181,
+      "step": 1631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160493105293765e-05,
+      "loss": 0.4257,
+      "step": 1632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1595812850941392e-05,
+      "loss": 0.5699,
+      "step": 1633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586693287285989e-05,
+      "loss": 0.3784,
+      "step": 1634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577572369752886e-05,
+      "loss": 0.0726,
+      "step": 1635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568450106124684e-05,
+      "loss": 0.4396,
+      "step": 1636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155932650418514e-05,
+      "loss": 0.6584,
+      "step": 1637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550201571719153e-05,
+      "loss": 0.3099,
+      "step": 1638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1541075316512746e-05,
+      "loss": 0.4678,
+      "step": 1639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1531947746353087e-05,
+      "loss": 0.6482,
+      "step": 1640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1522818869028447e-05,
+      "loss": 0.6951,
+      "step": 1641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151368869232823e-05,
+      "loss": 0.396,
+      "step": 1642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1504557224042943e-05,
+      "loss": 0.8122,
+      "step": 1643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1495424471964187e-05,
+      "loss": 0.4034,
+      "step": 1644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486290443884666e-05,
+      "loss": 0.3465,
+      "step": 1645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147715514759818e-05,
+      "loss": 0.9082,
+      "step": 1646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1468018590899593e-05,
+      "loss": 0.233,
+      "step": 1647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458880781584858e-05,
+      "loss": 0.232,
+      "step": 1648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449741727450994e-05,
+      "loss": 1.0662,
+      "step": 1649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144060143629608e-05,
+      "loss": 0.4462,
+      "step": 1650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.143145991591925e-05,
+      "loss": 0.3738,
+      "step": 1651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422317174120691e-05,
+      "loss": 0.2734,
+      "step": 1652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413173218701629e-05,
+      "loss": 0.1805,
+      "step": 1653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404028057464329e-05,
+      "loss": 0.4917,
+      "step": 1654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394881698212079e-05,
+      "loss": 0.6786,
+      "step": 1655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1385734148749192e-05,
+      "loss": 0.7895,
+      "step": 1656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376585416881002e-05,
+      "loss": 0.1466,
+      "step": 1657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367435510413841e-05,
+      "loss": 0.638,
+      "step": 1658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135828443715505e-05,
+      "loss": 0.9077,
+      "step": 1659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349132204912971e-05,
+      "loss": 0.1088,
+      "step": 1660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133997882149692e-05,
+      "loss": 0.9402,
+      "step": 1661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330824294717214e-05,
+      "loss": 0.1361,
+      "step": 1662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1321668632385123e-05,
+      "loss": 0.5959,
+      "step": 1663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131251184231291e-05,
+      "loss": 0.338,
+      "step": 1664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303353932313784e-05,
+      "loss": 0.5897,
+      "step": 1665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1294194910201913e-05,
+      "loss": 0.585,
+      "step": 1666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285034783792416e-05,
+      "loss": 0.6548,
+      "step": 1667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275873560901358e-05,
+      "loss": 0.1723,
+      "step": 1668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126671124934573e-05,
+      "loss": 0.6993,
+      "step": 1669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257547856943458e-05,
+      "loss": 0.343,
+      "step": 1670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248383391513391e-05,
+      "loss": 0.3124,
+      "step": 1671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1239217860875294e-05,
+      "loss": 0.4779,
+      "step": 1672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1230051272849833e-05,
+      "loss": 0.3736,
+      "step": 1673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1220883635258586e-05,
+      "loss": 0.4343,
+      "step": 1674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1211714955924018e-05,
+      "loss": 0.2803,
+      "step": 1675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1202545242669498e-05,
+      "loss": 0.4734,
+      "step": 1676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193374503319255e-05,
+      "loss": 0.8319,
+      "step": 1677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184202745698414e-05,
+      "loss": 0.3902,
+      "step": 1678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175029977632954e-05,
+      "loss": 0.1623,
+      "step": 1679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165856206949726e-05,
+      "loss": 0.4073,
+      "step": 1680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1156681441476429e-05,
+      "loss": 0.3945,
+      "step": 1681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147505689041624e-05,
+      "loss": 0.6758,
+      "step": 1682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138328957474691e-05,
+      "loss": 0.7108,
+      "step": 1683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1129151254605872e-05,
+      "loss": 0.2137,
+      "step": 1684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119972588266217e-05,
+      "loss": 0.6703,
+      "step": 1685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1110792966287609e-05,
+      "loss": 0.558,
+      "step": 1686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1101612396502743e-05,
+      "loss": 0.1204,
+      "step": 1687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1092430886745124e-05,
+      "loss": 0.3997,
+      "step": 1688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083248444849058e-05,
+      "loss": 0.4719,
+      "step": 1689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074065078649647e-05,
+      "loss": 0.5932,
+      "step": 1690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106488079598278e-05,
+      "loss": 0.455,
+      "step": 1691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055695604685133e-05,
+      "loss": 1.0171,
+      "step": 1692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1046509512594148e-05,
+      "loss": 0.5505,
+      "step": 1693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037322527548046e-05,
+      "loss": 0.4802,
+      "step": 1694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1028134657385804e-05,
+      "loss": 0.3654,
+      "step": 1695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018945909947157e-05,
+      "loss": 0.5186,
+      "step": 1696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1009756293072582e-05,
+      "loss": 0.8237,
+      "step": 1697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100056581460331e-05,
+      "loss": 0.3043,
+      "step": 1698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0991374482381293e-05,
+      "loss": 0.1973,
+      "step": 1699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0982182304249222e-05,
+      "loss": 0.979,
+      "step": 1700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0972989288050511e-05,
+      "loss": 0.5409,
+      "step": 1701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0963795441629275e-05,
+      "loss": 0.1416,
+      "step": 1702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0954600772830352e-05,
+      "loss": 0.2247,
+      "step": 1703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094540528949928e-05,
+      "loss": 0.9576,
+      "step": 1704
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093620899948228e-05,
+      "loss": 0.8673,
+      "step": 1705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092701191062628e-05,
+      "loss": 0.3669,
+      "step": 1706
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091781403077887e-05,
+      "loss": 0.1944,
+      "step": 1707
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908615367788331e-05,
+      "loss": 0.7423,
+      "step": 1708
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0899415929503602e-05,
+      "loss": 0.2325,
+      "step": 1709
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890215723774289e-05,
+      "loss": 1.0509,
+      "step": 1710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088101475845065e-05,
+      "loss": 0.2911,
+      "step": 1711
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0871813041383596e-05,
+      "loss": 0.2298,
+      "step": 1712
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086261058042467e-05,
+      "loss": 0.7847,
+      "step": 1713
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0853407383426058e-05,
+      "loss": 0.4888,
+      "step": 1714
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0844203458240574e-05,
+      "loss": 0.613,
+      "step": 1715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834998812721647e-05,
+      "loss": 0.5125,
+      "step": 1716
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 0.6644,
+      "step": 1717
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816587392100264e-05,
+      "loss": 0.4104,
+      "step": 1718
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080738063270772e-05,
+      "loss": 0.3257,
+      "step": 1719
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798173184401548e-05,
+      "loss": 0.2805,
+      "step": 1720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0788965055038179e-05,
+      "loss": 0.2437,
+      "step": 1721
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077975625247464e-05,
+      "loss": 0.5228,
+      "step": 1722
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770546784568523e-05,
+      "loss": 0.4851,
+      "step": 1723
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761336659177992e-05,
+      "loss": 0.4444,
+      "step": 1724
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752125884161766e-05,
+      "loss": 0.7173,
+      "step": 1725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742914467379126e-05,
+      "loss": 0.7134,
+      "step": 1726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0733702416689895e-05,
+      "loss": 0.3509,
+      "step": 1727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0724489739954447e-05,
+      "loss": 0.5505,
+      "step": 1728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0715276445033667e-05,
+      "loss": 0.3356,
+      "step": 1729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0706062539788995e-05,
+      "loss": 0.7692,
+      "step": 1730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696848032082376e-05,
+      "loss": 0.7251,
+      "step": 1731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0687632929776272e-05,
+      "loss": 0.2165,
+      "step": 1732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678417240733654e-05,
+      "loss": 0.5659,
+      "step": 1733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066920097281799e-05,
+      "loss": 0.2979,
+      "step": 1734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659984133893245e-05,
+      "loss": 0.3755,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0650766731823875e-05,
+      "loss": 1.2607,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641548774474807e-05,
+      "loss": 0.1543,
+      "step": 1737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632330269711449e-05,
+      "loss": 0.6102,
+      "step": 1738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0623111225399674e-05,
+      "loss": 1.092,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613891649405816e-05,
+      "loss": 0.3395,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0604671549596661e-05,
+      "loss": 0.6211,
+      "step": 1741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595450933839444e-05,
+      "loss": 0.1144,
+      "step": 1742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058622981000184e-05,
+      "loss": 0.692,
+      "step": 1743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057700818595195e-05,
+      "loss": 0.3174,
+      "step": 1744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567786069558321e-05,
+      "loss": 0.3368,
+      "step": 1745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0558563468689902e-05,
+      "loss": 0.0985,
+      "step": 1746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549340391216058e-05,
+      "loss": 0.5689,
+      "step": 1747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540116845006568e-05,
+      "loss": 0.9832,
+      "step": 1748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530892837931603e-05,
+      "loss": 0.4641,
+      "step": 1749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0521668377861734e-05,
+      "loss": 0.8304,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512443472667917e-05,
+      "loss": 0.407,
+      "step": 1751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050321813022148e-05,
+      "loss": 0.3282,
+      "step": 1752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0493992358394136e-05,
+      "loss": 0.451,
+      "step": 1753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048476616505796e-05,
+      "loss": 1.0359,
+      "step": 1754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047553955808538e-05,
+      "loss": 1.2757,
+      "step": 1755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046631254534919e-05,
+      "loss": 0.2355,
+      "step": 1756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457085134722516e-05,
+      "loss": 0.4436,
+      "step": 1757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0447857334078828e-05,
+      "loss": 0.5717,
+      "step": 1758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438629151291944e-05,
+      "loss": 0.9789,
+      "step": 1759
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0429400594235978e-05,
+      "loss": 0.4001,
+      "step": 1760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420171670785392e-05,
+      "loss": 0.5228,
+      "step": 1761
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0410942388814949e-05,
+      "loss": 0.2328,
+      "step": 1762
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0401712756199711e-05,
+      "loss": 0.6797,
+      "step": 1763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0392482780815052e-05,
+      "loss": 0.7564,
+      "step": 1764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383252470536631e-05,
+      "loss": 0.9138,
+      "step": 1765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374021833240391e-05,
+      "loss": 0.8152,
+      "step": 1766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0364790876802564e-05,
+      "loss": 0.7219,
+      "step": 1767
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355559609099641e-05,
+      "loss": 0.6621,
+      "step": 1768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034632803800839e-05,
+      "loss": 1.2751,
+      "step": 1769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0337096171405832e-05,
+      "loss": 0.5051,
+      "step": 1770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032786401716924e-05,
+      "loss": 0.1375,
+      "step": 1771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0318631583176136e-05,
+      "loss": 0.1353,
+      "step": 1772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0309398877304278e-05,
+      "loss": 0.3308,
+      "step": 1773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0300165907431652e-05,
+      "loss": 0.4626,
+      "step": 1774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0290932681436482e-05,
+      "loss": 0.2061,
+      "step": 1775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281699207197196e-05,
+      "loss": 0.3251,
+      "step": 1776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027246549259244e-05,
+      "loss": 0.2286,
+      "step": 1777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263231545501068e-05,
+      "loss": 0.6799,
+      "step": 1778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253997373802132e-05,
+      "loss": 0.4321,
+      "step": 1779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244762985374863e-05,
+      "loss": 0.3106,
+      "step": 1780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0235528388098701e-05,
+      "loss": 0.3871,
+      "step": 1781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0226293589853238e-05,
+      "loss": 0.373,
+      "step": 1782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0217058598518259e-05,
+      "loss": 0.8188,
+      "step": 1783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02078234219737e-05,
+      "loss": 0.4146,
+      "step": 1784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0198588068099658e-05,
+      "loss": 0.4664,
+      "step": 1785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0189352544776387e-05,
+      "loss": 0.2935,
+      "step": 1786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018011685988428e-05,
+      "loss": 0.753,
+      "step": 1787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0170881021303867e-05,
+      "loss": 0.3545,
+      "step": 1788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161645036915818e-05,
+      "loss": 0.3849,
+      "step": 1789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0152408914600911e-05,
+      "loss": 0.5602,
+      "step": 1790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143172662240062e-05,
+      "loss": 0.2889,
+      "step": 1791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0133936287714281e-05,
+      "loss": 0.1493,
+      "step": 1792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012469979890469e-05,
+      "loss": 0.3104,
+      "step": 1793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0115463203692507e-05,
+      "loss": 0.823,
+      "step": 1794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106226509959045e-05,
+      "loss": 1.2069,
+      "step": 1795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009698972558569e-05,
+      "loss": 0.1861,
+      "step": 1796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087752858453923e-05,
+      "loss": 0.1901,
+      "step": 1797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0078515916445276e-05,
+      "loss": 0.4054,
+      "step": 1798
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069278907441355e-05,
+      "loss": 0.3749,
+      "step": 1799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060041839323827e-05,
+      "loss": 1.2563,
+      "step": 1800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050804719974402e-05,
+      "loss": 0.1829,
+      "step": 1801
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004156755727483e-05,
+      "loss": 0.2088,
+      "step": 1802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032330359106919e-05,
+      "loss": 0.2763,
+      "step": 1803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023093133352478e-05,
+      "loss": 0.5774,
+      "step": 1804
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013855887893362e-05,
+      "loss": 0.5006,
+      "step": 1805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0004618630611435e-05,
+      "loss": 0.2446,
+      "step": 1806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99538136938857e-06,
+      "loss": 0.5574,
+      "step": 1807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98614411210664e-06,
+      "loss": 0.4811,
+      "step": 1808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976906866647526e-06,
+      "loss": 0.3172,
+      "step": 1809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967669640893085e-06,
+      "loss": 0.7546,
+      "step": 1810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95843244272517e-06,
+      "loss": 0.8493,
+      "step": 1811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9491952800256e-06,
+      "loss": 0.5657,
+      "step": 1812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93995816067618e-06,
+      "loss": 0.1002,
+      "step": 1813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930721092558648e-06,
+      "loss": 0.3113,
+      "step": 1814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92148408355473e-06,
+      "loss": 0.5703,
+      "step": 1815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91224714154608e-06,
+      "loss": 0.2937,
+      "step": 1816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90301027441431e-06,
+      "loss": 0.3286,
+      "step": 1817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89377349004096e-06,
+      "loss": 0.2368,
+      "step": 1818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884536796307497e-06,
+      "loss": 0.6937,
+      "step": 1819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875300201095312e-06,
+      "loss": 1.4493,
+      "step": 1820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.866063712285724e-06,
+      "loss": 0.5819,
+      "step": 1821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85682733775994e-06,
+      "loss": 0.2901,
+      "step": 1822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847591085399089e-06,
+      "loss": 0.4432,
+      "step": 1823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838354963084187e-06,
+      "loss": 1.2098,
+      "step": 1824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829118978696136e-06,
+      "loss": 0.3318,
+      "step": 1825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819883140115722e-06,
+      "loss": 0.3462,
+      "step": 1826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810647455223615e-06,
+      "loss": 0.4963,
+      "step": 1827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.801411931900344e-06,
+      "loss": 0.3202,
+      "step": 1828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792176578026307e-06,
+      "loss": 0.3879,
+      "step": 1829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.782941401481745e-06,
+      "loss": 0.5404,
+      "step": 1830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773706410146764e-06,
+      "loss": 0.9879,
+      "step": 1831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.764471611901302e-06,
+      "loss": 0.5852,
+      "step": 1832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.755237014625136e-06,
+      "loss": 0.4381,
+      "step": 1833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746002626197873e-06,
+      "loss": 0.6161,
+      "step": 1834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.736768454498935e-06,
+      "loss": 0.7264,
+      "step": 1835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.727534507407563e-06,
+      "loss": 0.3095,
+      "step": 1836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718300792802808e-06,
+      "loss": 0.3566,
+      "step": 1837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70906731856352e-06,
+      "loss": 0.308,
+      "step": 1838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69983409256835e-06,
+      "loss": 0.2914,
+      "step": 1839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690601122695727e-06,
+      "loss": 0.3,
+      "step": 1840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.681368416823869e-06,
+      "loss": 0.9818,
+      "step": 1841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672135982830761e-06,
+      "loss": 0.2093,
+      "step": 1842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662903828594172e-06,
+      "loss": 0.5878,
+      "step": 1843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.653671961991613e-06,
+      "loss": 0.9939,
+      "step": 1844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64444039090036e-06,
+      "loss": 0.1694,
+      "step": 1845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63520912319744e-06,
+      "loss": 0.5107,
+      "step": 1846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625978166759612e-06,
+      "loss": 0.9991,
+      "step": 1847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616747529463372e-06,
+      "loss": 0.1983,
+      "step": 1848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607517219184951e-06,
+      "loss": 0.269,
+      "step": 1849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.598287243800292e-06,
+      "loss": 0.1587,
+      "step": 1850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.589057611185058e-06,
+      "loss": 0.6279,
+      "step": 1851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57982832921461e-06,
+      "loss": 0.6071,
+      "step": 1852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570599405764023e-06,
+      "loss": 0.4729,
+      "step": 1853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561370848708061e-06,
+      "loss": 0.6719,
+      "step": 1854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552142665921172e-06,
+      "loss": 0.4246,
+      "step": 1855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542914865277488e-06,
+      "loss": 0.1767,
+      "step": 1856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.533687454650816e-06,
+      "loss": 0.7098,
+      "step": 1857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.524460441914621e-06,
+      "loss": 0.6204,
+      "step": 1858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515233834942042e-06,
+      "loss": 0.516,
+      "step": 1859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.506007641605866e-06,
+      "loss": 0.5183,
+      "step": 1860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496781869778521e-06,
+      "loss": 0.3122,
+      "step": 1861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48755652733209e-06,
+      "loss": 0.7072,
+      "step": 1862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47833162213827e-06,
+      "loss": 0.3301,
+      "step": 1863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469107162068399e-06,
+      "loss": 0.5457,
+      "step": 1864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459883154993435e-06,
+      "loss": 0.0791,
+      "step": 1865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450659608783945e-06,
+      "loss": 0.341,
+      "step": 1866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4414365313101e-06,
+      "loss": 1.0575,
+      "step": 1867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43221393044168e-06,
+      "loss": 0.3086,
+      "step": 1868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422991814048051e-06,
+      "loss": 0.1198,
+      "step": 1869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413770189998165e-06,
+      "loss": 1.4348,
+      "step": 1870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40454906616056e-06,
+      "loss": 0.1273,
+      "step": 1871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395328450403342e-06,
+      "loss": 0.1854,
+      "step": 1872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38610835059419e-06,
+      "loss": 0.4726,
+      "step": 1873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37688877460033e-06,
+      "loss": 0.2803,
+      "step": 1874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367669730288555e-06,
+      "loss": 0.3841,
+      "step": 1875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358451225525197e-06,
+      "loss": 0.2185,
+      "step": 1876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.349233268176127e-06,
+      "loss": 0.3447,
+      "step": 1877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340015866106755e-06,
+      "loss": 0.2392,
+      "step": 1878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330799027182015e-06,
+      "loss": 0.1613,
+      "step": 1879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32158275926635e-06,
+      "loss": 0.0594,
+      "step": 1880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31236707022373e-06,
+      "loss": 0.2539,
+      "step": 1881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303151967917626e-06,
+      "loss": 0.2781,
+      "step": 1882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293937460211005e-06,
+      "loss": 0.3412,
+      "step": 1883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284723554966335e-06,
+      "loss": 0.4808,
+      "step": 1884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27551026004556e-06,
+      "loss": 0.4254,
+      "step": 1885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266297583310106e-06,
+      "loss": 0.4481,
+      "step": 1886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257085532620875e-06,
+      "loss": 0.1414,
+      "step": 1887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.247874115838236e-06,
+      "loss": 1.0835,
+      "step": 1888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23866334082201e-06,
+      "loss": 0.4188,
+      "step": 1889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22945321543148e-06,
+      "loss": 0.5582,
+      "step": 1890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220243747525363e-06,
+      "loss": 0.367,
+      "step": 1891
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.211034944961825e-06,
+      "loss": 0.801,
+      "step": 1892
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201826815598455e-06,
+      "loss": 0.4105,
+      "step": 1893
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192619367292281e-06,
+      "loss": 0.9521,
+      "step": 1894
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183412607899741e-06,
+      "loss": 0.3196,
+      "step": 1895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 0.3382,
+      "step": 1896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165001187278357e-06,
+      "loss": 0.3227,
+      "step": 1897
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155796541759429e-06,
+      "loss": 0.2444,
+      "step": 1898
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146592616573942e-06,
+      "loss": 0.2013,
+      "step": 1899
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13738941957533e-06,
+      "loss": 0.8258,
+      "step": 1900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12818695861641e-06,
+      "loss": 0.1978,
+      "step": 1901
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118985241549352e-06,
+      "loss": 0.1101,
+      "step": 1902
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109784276225713e-06,
+      "loss": 0.9038,
+      "step": 1903
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.100584070496401e-06,
+      "loss": 0.5136,
+      "step": 1904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09138463221167e-06,
+      "loss": 0.0898,
+      "step": 1905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082185969221133e-06,
+      "loss": 0.1606,
+      "step": 1906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072988089373726e-06,
+      "loss": 1.0754,
+      "step": 1907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.063791000517722e-06,
+      "loss": 0.2618,
+      "step": 1908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054594710500723e-06,
+      "loss": 0.7029,
+      "step": 1909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04539922716965e-06,
+      "loss": 0.4205,
+      "step": 1910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.036204558370725e-06,
+      "loss": 0.5737,
+      "step": 1911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027010711949494e-06,
+      "loss": 0.374,
+      "step": 1912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01781769575078e-06,
+      "loss": 0.836,
+      "step": 1913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.008625517618709e-06,
+      "loss": 0.6891,
+      "step": 1914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999434185396693e-06,
+      "loss": 0.6136,
+      "step": 1915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.990243706927418e-06,
+      "loss": 0.1659,
+      "step": 1916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981054090052847e-06,
+      "loss": 0.7642,
+      "step": 1917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971865342614199e-06,
+      "loss": 0.4512,
+      "step": 1918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962677472451956e-06,
+      "loss": 0.2934,
+      "step": 1919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.953490487405854e-06,
+      "loss": 0.7253,
+      "step": 1920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944304395314868e-06,
+      "loss": 0.7945,
+      "step": 1921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93511920401722e-06,
+      "loss": 0.5266,
+      "step": 1922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925934921350356e-06,
+      "loss": 0.1454,
+      "step": 1923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916751555150947e-06,
+      "loss": 0.2652,
+      "step": 1924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.907569113254877e-06,
+      "loss": 0.2227,
+      "step": 1925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.898387603497259e-06,
+      "loss": 0.3138,
+      "step": 1926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889207033712391e-06,
+      "loss": 0.4142,
+      "step": 1927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88002741173379e-06,
+      "loss": 0.2549,
+      "step": 1928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870848745394131e-06,
+      "loss": 0.5425,
+      "step": 1929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861671042525312e-06,
+      "loss": 0.3095,
+      "step": 1930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852494310958379e-06,
+      "loss": 0.3617,
+      "step": 1931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84331855852357e-06,
+      "loss": 0.2428,
+      "step": 1932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834143793050275e-06,
+      "loss": 0.3335,
+      "step": 1933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82497002236705e-06,
+      "loss": 0.5209,
+      "step": 1934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81579725430159e-06,
+      "loss": 0.9873,
+      "step": 1935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806625496680747e-06,
+      "loss": 0.4297,
+      "step": 1936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797454757330504e-06,
+      "loss": 0.3849,
+      "step": 1937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788285044075982e-06,
+      "loss": 0.3311,
+      "step": 1938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77911636474142e-06,
+      "loss": 0.1711,
+      "step": 1939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769948727150172e-06,
+      "loss": 0.3717,
+      "step": 1940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760782139124711e-06,
+      "loss": 0.5716,
+      "step": 1941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75161660848661e-06,
+      "loss": 0.3381,
+      "step": 1942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742452143056543e-06,
+      "loss": 0.0657,
+      "step": 1943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733288750654271e-06,
+      "loss": 0.2583,
+      "step": 1944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724126439098645e-06,
+      "loss": 0.0912,
+      "step": 1945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714965216207587e-06,
+      "loss": 0.3709,
+      "step": 1946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705805089798089e-06,
+      "loss": 0.4941,
+      "step": 1947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69664606768622e-06,
+      "loss": 0.6682,
+      "step": 1948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68748815768709e-06,
+      "loss": 0.2392,
+      "step": 1949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67833136761488e-06,
+      "loss": 0.1378,
+      "step": 1950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669175705282791e-06,
+      "loss": 0.3325,
+      "step": 1951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660021178503082e-06,
+      "loss": 0.5819,
+      "step": 1952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650867795087032e-06,
+      "loss": 0.5737,
+      "step": 1953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641715562844952e-06,
+      "loss": 0.4178,
+      "step": 1954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63256448958616e-06,
+      "loss": 1.0417,
+      "step": 1955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623414583119003e-06,
+      "loss": 0.1922,
+      "step": 1956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61426585125081e-06,
+      "loss": 0.2904,
+      "step": 1957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.605118301787925e-06,
+      "loss": 0.2485,
+      "step": 1958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595971942535673e-06,
+      "loss": 0.2191,
+      "step": 1959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586826781298373e-06,
+      "loss": 0.2224,
+      "step": 1960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577682825879312e-06,
+      "loss": 0.5558,
+      "step": 1961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.568540084080755e-06,
+      "loss": 0.1348,
+      "step": 1962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.559398563703924e-06,
+      "loss": 0.8219,
+      "step": 1963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55025827254901e-06,
+      "loss": 0.0509,
+      "step": 1964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541119218415144e-06,
+      "loss": 0.4921,
+      "step": 1965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.531981409100409e-06,
+      "loss": 0.3851,
+      "step": 1966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522844852401824e-06,
+      "loss": 0.5909,
+      "step": 1967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513709556115335e-06,
+      "loss": 0.3923,
+      "step": 1968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504575528035816e-06,
+      "loss": 0.5473,
+      "step": 1969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49544277595706e-06,
+      "loss": 0.392,
+      "step": 1970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486311307671773e-06,
+      "loss": 0.4683,
+      "step": 1971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.477181130971559e-06,
+      "loss": 0.4059,
+      "step": 1972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46805225364692e-06,
+      "loss": 0.4669,
+      "step": 1973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458924683487257e-06,
+      "loss": 0.1506,
+      "step": 1974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44979842828085e-06,
+      "loss": 0.3853,
+      "step": 1975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440673495814862e-06,
+      "loss": 0.301,
+      "step": 1976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.431549893875319e-06,
+      "loss": 0.7295,
+      "step": 1977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42242763024712e-06,
+      "loss": 0.3899,
+      "step": 1978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413306712714014e-06,
+      "loss": 0.4428,
+      "step": 1979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40418714905861e-06,
+      "loss": 0.8604,
+      "step": 1980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.395068947062354e-06,
+      "loss": 0.4804,
+      "step": 1981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385952114505537e-06,
+      "loss": 0.3306,
+      "step": 1982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37683665916728e-06,
+      "loss": 0.4117,
+      "step": 1983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36772258882552e-06,
+      "loss": 0.1979,
+      "step": 1984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358609911257023e-06,
+      "loss": 0.2227,
+      "step": 1985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349498634237366e-06,
+      "loss": 0.3066,
+      "step": 1986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340388765540923e-06,
+      "loss": 0.4425,
+      "step": 1987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.331280312940872e-06,
+      "loss": 0.5891,
+      "step": 1988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.322173284209187e-06,
+      "loss": 0.1983,
+      "step": 1989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.313067687116618e-06,
+      "loss": 0.3935,
+      "step": 1990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.303963529432695e-06,
+      "loss": 1.0187,
+      "step": 1991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294860818925726e-06,
+      "loss": 0.7953,
+      "step": 1992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285759563362778e-06,
+      "loss": 0.1827,
+      "step": 1993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276659770509685e-06,
+      "loss": 0.5949,
+      "step": 1994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.267561448131016e-06,
+      "loss": 0.4719,
+      "step": 1995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258464603990103e-06,
+      "loss": 0.5936,
+      "step": 1996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.249369245849007e-06,
+      "loss": 0.2959,
+      "step": 1997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240275381468528e-06,
+      "loss": 0.3799,
+      "step": 1998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231183018608184e-06,
+      "loss": 0.6286,
+      "step": 1999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222092165026218e-06,
+      "loss": 0.41,
+      "step": 2000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213002828479574e-06,
+      "loss": 0.5375,
+      "step": 2001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.203915016723919e-06,
+      "loss": 0.1438,
+      "step": 2002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194828737513606e-06,
+      "loss": 0.8353,
+      "step": 2003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.185743998601681e-06,
+      "loss": 0.2235,
+      "step": 2004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176660807739886e-06,
+      "loss": 0.2688,
+      "step": 2005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16757917267863e-06,
+      "loss": 0.3904,
+      "step": 2006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158499101166997e-06,
+      "loss": 0.2522,
+      "step": 2007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.149420600952744e-06,
+      "loss": 0.4221,
+      "step": 2008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14034367978228e-06,
+      "loss": 0.6468,
+      "step": 2009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13126834540067e-06,
+      "loss": 0.4395,
+      "step": 2010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122194605551625e-06,
+      "loss": 0.4075,
+      "step": 2011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113122467977491e-06,
+      "loss": 0.3251,
+      "step": 2012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104051940419251e-06,
+      "loss": 0.5516,
+      "step": 2013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094983030616517e-06,
+      "loss": 0.3001,
+      "step": 2014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085915746307515e-06,
+      "loss": 0.1375,
+      "step": 2015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07685009522909e-06,
+      "loss": 0.3996,
+      "step": 2016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.067786085116682e-06,
+      "loss": 0.7781,
+      "step": 2017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058723723704343e-06,
+      "loss": 0.5577,
+      "step": 2018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049663018724714e-06,
+      "loss": 0.659,
+      "step": 2019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040603977909021e-06,
+      "loss": 1.2468,
+      "step": 2020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.031546608987072e-06,
+      "loss": 0.0874,
+      "step": 2021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02249091968725e-06,
+      "loss": 1.0694,
+      "step": 2022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.013436917736495e-06,
+      "loss": 0.3233,
+      "step": 2023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.004384610860324e-06,
+      "loss": 0.6258,
+      "step": 2024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995334006782793e-06,
+      "loss": 0.3916,
+      "step": 2025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98628511322651e-06,
+      "loss": 0.1876,
+      "step": 2026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97723793791263e-06,
+      "loss": 1.2154,
+      "step": 2027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968192488560829e-06,
+      "loss": 0.8171,
+      "step": 2028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95914877288932e-06,
+      "loss": 0.8539,
+      "step": 2029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.950106798614831e-06,
+      "loss": 0.242,
+      "step": 2030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941066573452613e-06,
+      "loss": 0.7691,
+      "step": 2031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932028105116412e-06,
+      "loss": 0.3239,
+      "step": 2032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922991401318487e-06,
+      "loss": 0.4705,
+      "step": 2033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913956469769582e-06,
+      "loss": 0.735,
+      "step": 2034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.904923318178934e-06,
+      "loss": 0.3213,
+      "step": 2035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895891954254258e-06,
+      "loss": 0.4558,
+      "step": 2036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886862385701748e-06,
+      "loss": 0.1707,
+      "step": 2037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87783462022606e-06,
+      "loss": 0.4695,
+      "step": 2038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868808665530323e-06,
+      "loss": 0.4464,
+      "step": 2039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859784529316103e-06,
+      "loss": 0.1301,
+      "step": 2040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85076221928343e-06,
+      "loss": 0.3121,
+      "step": 2041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841741743130765e-06,
+      "loss": 0.2796,
+      "step": 2042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832723108555016e-06,
+      "loss": 0.7807,
+      "step": 2043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823706323251512e-06,
+      "loss": 0.9984,
+      "step": 2044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814691394914001e-06,
+      "loss": 0.3442,
+      "step": 2045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805678331234647e-06,
+      "loss": 0.3651,
+      "step": 2046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796667139904036e-06,
+      "loss": 0.2922,
+      "step": 2047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78765782861114e-06,
+      "loss": 0.5894,
+      "step": 2048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778650405043336e-06,
+      "loss": 0.142,
+      "step": 2049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769644876886393e-06,
+      "loss": 0.155,
+      "step": 2050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760641251824447e-06,
+      "loss": 0.3711,
+      "step": 2051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.751639537540024e-06,
+      "loss": 0.6391,
+      "step": 2052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74263974171402e-06,
+      "loss": 0.2635,
+      "step": 2053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733641872025688e-06,
+      "loss": 0.6622,
+      "step": 2054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724645936152643e-06,
+      "loss": 0.2139,
+      "step": 2055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715651941770844e-06,
+      "loss": 0.3761,
+      "step": 2056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706659896554594e-06,
+      "loss": 0.4603,
+      "step": 2057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697669808176537e-06,
+      "loss": 0.2127,
+      "step": 2058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688681684307646e-06,
+      "loss": 0.6934,
+      "step": 2059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679695532617214e-06,
+      "loss": 0.4561,
+      "step": 2060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670711360772865e-06,
+      "loss": 0.6475,
+      "step": 2061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.661729176440506e-06,
+      "loss": 0.1375,
+      "step": 2062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652748987284375e-06,
+      "loss": 0.8068,
+      "step": 2063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643770800966994e-06,
+      "loss": 0.593,
+      "step": 2064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634794625149184e-06,
+      "loss": 0.9097,
+      "step": 2065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625820467490047e-06,
+      "loss": 0.127,
+      "step": 2066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6168483356469555e-06,
+      "loss": 0.4516,
+      "step": 2067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607878237275561e-06,
+      "loss": 0.0942,
+      "step": 2068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598910180029783e-06,
+      "loss": 0.85,
+      "step": 2069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5899441715617906e-06,
+      "loss": 0.2062,
+      "step": 2070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580980219522015e-06,
+      "loss": 0.1338,
+      "step": 2071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572018331559126e-06,
+      "loss": 0.6227,
+      "step": 2072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5630585153200286e-06,
+      "loss": 0.7201,
+      "step": 2073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.554100778449866e-06,
+      "loss": 0.7522,
+      "step": 2074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 0.2914,
+      "step": 2075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536191573388042e-06,
+      "loss": 0.684,
+      "step": 2076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.527240120477771e-06,
+      "loss": 0.0764,
+      "step": 2077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51829077749919e-06,
+      "loss": 0.5365,
+      "step": 2078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.509343552088513e-06,
+      "loss": 0.5092,
+      "step": 2079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500398451880133e-06,
+      "loss": 0.2479,
+      "step": 2080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.491455484506643e-06,
+      "loss": 0.3206,
+      "step": 2081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4825146575988e-06,
+      "loss": 0.1472,
+      "step": 2082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4735759787855525e-06,
+      "loss": 0.3303,
+      "step": 2083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.464639455693996e-06,
+      "loss": 0.2266,
+      "step": 2084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455705095949403e-06,
+      "loss": 0.5328,
+      "step": 2085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446772907175191e-06,
+      "loss": 0.561,
+      "step": 2086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.437842896992933e-06,
+      "loss": 0.1689,
+      "step": 2087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4289150730223355e-06,
+      "loss": 0.5399,
+      "step": 2088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4199894428812435e-06,
+      "loss": 0.455,
+      "step": 2089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411066014185624e-06,
+      "loss": 0.5485,
+      "step": 2090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402144794549577e-06,
+      "loss": 0.5582,
+      "step": 2091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3932257915853056e-06,
+      "loss": 0.5346,
+      "step": 2092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3843090129031335e-06,
+      "loss": 0.0715,
+      "step": 2093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375394466111479e-06,
+      "loss": 1.237,
+      "step": 2094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366482158816851e-06,
+      "loss": 0.3052,
+      "step": 2095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357572098623855e-06,
+      "loss": 0.4633,
+      "step": 2096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3486642931351835e-06,
+      "loss": 0.2681,
+      "step": 2097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.339758749951592e-06,
+      "loss": 0.843,
+      "step": 2098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330855476671923e-06,
+      "loss": 0.4302,
+      "step": 2099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321954480893059e-06,
+      "loss": 0.2097,
+      "step": 2100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.313055770209961e-06,
+      "loss": 0.401,
+      "step": 2101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.304159352215625e-06,
+      "loss": 0.4361,
+      "step": 2102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295265234501103e-06,
+      "loss": 0.1067,
+      "step": 2103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2863734246554785e-06,
+      "loss": 0.8222,
+      "step": 2104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.277483930265865e-06,
+      "loss": 0.6234,
+      "step": 2105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.268596758917395e-06,
+      "loss": 0.1992,
+      "step": 2106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259711918193231e-06,
+      "loss": 0.2683,
+      "step": 2107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.250829415674536e-06,
+      "loss": 0.5775,
+      "step": 2108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2419492589404885e-06,
+      "loss": 0.4086,
+      "step": 2109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.233071455568259e-06,
+      "loss": 0.3281,
+      "step": 2110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2241960131330046e-06,
+      "loss": 0.5657,
+      "step": 2111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.215322939207874e-06,
+      "loss": 0.2041,
+      "step": 2112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206452241363999e-06,
+      "loss": 0.6964,
+      "step": 2113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.197583927170478e-06,
+      "loss": 0.2084,
+      "step": 2114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1887180041943746e-06,
+      "loss": 0.614,
+      "step": 2115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1798544800007205e-06,
+      "loss": 0.6934,
+      "step": 2116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.170993362152488e-06,
+      "loss": 0.5556,
+      "step": 2117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162134658210602e-06,
+      "loss": 1.0528,
+      "step": 2118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153278375733935e-06,
+      "loss": 0.3802,
+      "step": 2119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144424522279283e-06,
+      "loss": 0.571,
+      "step": 2120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135573105401375e-06,
+      "loss": 0.5245,
+      "step": 2121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.126724132652854e-06,
+      "loss": 0.4943,
+      "step": 2122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117877611584287e-06,
+      "loss": 0.9022,
+      "step": 2123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109033549744141e-06,
+      "loss": 0.0662,
+      "step": 2124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.100191954678792e-06,
+      "loss": 0.5533,
+      "step": 2125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.091352833932508e-06,
+      "loss": 0.1797,
+      "step": 2126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082516195047444e-06,
+      "loss": 0.2887,
+      "step": 2127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.073682045563632e-06,
+      "loss": 0.6919,
+      "step": 2128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064850393018996e-06,
+      "loss": 0.444,
+      "step": 2129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056021244949315e-06,
+      "loss": 0.432,
+      "step": 2130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047194608888233e-06,
+      "loss": 0.8035,
+      "step": 2131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038370492367261e-06,
+      "loss": 0.6029,
+      "step": 2132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029548902915746e-06,
+      "loss": 0.6482,
+      "step": 2133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.020729848060886e-06,
+      "loss": 0.6117,
+      "step": 2134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011913335327718e-06,
+      "loss": 0.5655,
+      "step": 2135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003099372239105e-06,
+      "loss": 0.0875,
+      "step": 2136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994287966315736e-06,
+      "loss": 0.8337,
+      "step": 2137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985479125076125e-06,
+      "loss": 0.8397,
+      "step": 2138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.976672856036586e-06,
+      "loss": 0.1958,
+      "step": 2139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967869166711243e-06,
+      "loss": 0.1277,
+      "step": 2140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959068064612022e-06,
+      "loss": 0.3671,
+      "step": 2141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950269557248639e-06,
+      "loss": 0.2184,
+      "step": 2142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.941473652128598e-06,
+      "loss": 0.7676,
+      "step": 2143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932680356757173e-06,
+      "loss": 0.4332,
+      "step": 2144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.923889678637425e-06,
+      "loss": 0.2471,
+      "step": 2145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915101625270175e-06,
+      "loss": 0.1123,
+      "step": 2146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906316204154002e-06,
+      "loss": 0.1957,
+      "step": 2147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897533422785245e-06,
+      "loss": 0.2846,
+      "step": 2148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8887532886579896e-06,
+      "loss": 0.2215,
+      "step": 2149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879975809264052e-06,
+      "loss": 0.2054,
+      "step": 2150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871200992092999e-06,
+      "loss": 1.0862,
+      "step": 2151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.862428844632114e-06,
+      "loss": 0.5627,
+      "step": 2152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853659374366408e-06,
+      "loss": 0.7513,
+      "step": 2153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8448925887786114e-06,
+      "loss": 0.5905,
+      "step": 2154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.836128495349152e-06,
+      "loss": 0.1944,
+      "step": 2155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.827367101556168e-06,
+      "loss": 0.3289,
+      "step": 2156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.818608414875498e-06,
+      "loss": 0.6763,
+      "step": 2157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809852442780664e-06,
+      "loss": 0.218,
+      "step": 2158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80109919274287e-06,
+      "loss": 0.3919,
+      "step": 2159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792348672231011e-06,
+      "loss": 0.4869,
+      "step": 2160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.783600888711633e-06,
+      "loss": 0.7862,
+      "step": 2161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774855849648961e-06,
+      "loss": 0.3251,
+      "step": 2162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76611356250487e-06,
+      "loss": 0.3553,
+      "step": 2163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757374034738899e-06,
+      "loss": 0.2447,
+      "step": 2164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74863727380822e-06,
+      "loss": 0.2476,
+      "step": 2165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739903287167646e-06,
+      "loss": 0.2,
+      "step": 2166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73117208226963e-06,
+      "loss": 0.3214,
+      "step": 2167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722443666564244e-06,
+      "loss": 0.4587,
+      "step": 2168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7137180474991825e-06,
+      "loss": 0.3061,
+      "step": 2169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704995232519755e-06,
+      "loss": 0.4128,
+      "step": 2170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69627522906888e-06,
+      "loss": 0.5744,
+      "step": 2171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687558044587072e-06,
+      "loss": 0.5126,
+      "step": 2172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.678843686512437e-06,
+      "loss": 0.35,
+      "step": 2173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670132162280685e-06,
+      "loss": 0.11,
+      "step": 2174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66142347932509e-06,
+      "loss": 0.1399,
+      "step": 2175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652717645076516e-06,
+      "loss": 1.1617,
+      "step": 2176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6440146669633855e-06,
+      "loss": 0.4767,
+      "step": 2177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635314552411687e-06,
+      "loss": 0.196,
+      "step": 2178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626617308844968e-06,
+      "loss": 0.6354,
+      "step": 2179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617922943684327e-06,
+      "loss": 0.2303,
+      "step": 2180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609231464348402e-06,
+      "loss": 0.8564,
+      "step": 2181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.600542878253378e-06,
+      "loss": 0.2515,
+      "step": 2182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591857192812955e-06,
+      "loss": 0.5646,
+      "step": 2183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.583174415438372e-06,
+      "loss": 0.4407,
+      "step": 2184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574494553538379e-06,
+      "loss": 0.3187,
+      "step": 2185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.565817614519245e-06,
+      "loss": 0.9719,
+      "step": 2186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557143605784743e-06,
+      "loss": 0.3585,
+      "step": 2187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5484725347361374e-06,
+      "loss": 0.3806,
+      "step": 2188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5398044087721946e-06,
+      "loss": 0.3976,
+      "step": 2189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5311392352891704e-06,
+      "loss": 0.3451,
+      "step": 2190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522477021680791e-06,
+      "loss": 0.5907,
+      "step": 2191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.513817775338268e-06,
+      "loss": 0.8097,
+      "step": 2192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505161503650277e-06,
+      "loss": 0.9499,
+      "step": 2193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.496508214002948e-06,
+      "loss": 0.6826,
+      "step": 2194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487857913779876e-06,
+      "loss": 0.7373,
+      "step": 2195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479210610362103e-06,
+      "loss": 0.7724,
+      "step": 2196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470566311128113e-06,
+      "loss": 0.2303,
+      "step": 2197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46192502345383e-06,
+      "loss": 0.3151,
+      "step": 2198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4532867547126e-06,
+      "loss": 0.3191,
+      "step": 2199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444651512275198e-06,
+      "loss": 0.7276,
+      "step": 2200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43601930350982e-06,
+      "loss": 0.4496,
+      "step": 2201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427390135782068e-06,
+      "loss": 0.4966,
+      "step": 2202
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418764016454953e-06,
+      "loss": 0.7333,
+      "step": 2203
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410140952888887e-06,
+      "loss": 0.7823,
+      "step": 2204
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401520952441662e-06,
+      "loss": 0.5573,
+      "step": 2205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3929040224684725e-06,
+      "loss": 0.2321,
+      "step": 2206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.384290170321881e-06,
+      "loss": 0.7833,
+      "step": 2207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375679403351834e-06,
+      "loss": 0.3763,
+      "step": 2208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36707172890564e-06,
+      "loss": 0.8738,
+      "step": 2209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3584671543279655e-06,
+      "loss": 0.1922,
+      "step": 2210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349865686960832e-06,
+      "loss": 0.4001,
+      "step": 2211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.341267334143621e-06,
+      "loss": 0.4229,
+      "step": 2212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332672103213042e-06,
+      "loss": 0.7112,
+      "step": 2213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3240800015031544e-06,
+      "loss": 0.2288,
+      "step": 2214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.315491036345338e-06,
+      "loss": 0.2572,
+      "step": 2215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306905215068294e-06,
+      "loss": 0.522,
+      "step": 2216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298322544998048e-06,
+      "loss": 0.2359,
+      "step": 2217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2897430334579355e-06,
+      "loss": 0.8938,
+      "step": 2218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281166687768596e-06,
+      "loss": 0.4045,
+      "step": 2219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272593515247971e-06,
+      "loss": 0.6413,
+      "step": 2220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.264023523211283e-06,
+      "loss": 0.404,
+      "step": 2221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.255456718971053e-06,
+      "loss": 0.477,
+      "step": 2222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246893109837076e-06,
+      "loss": 0.7949,
+      "step": 2223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238332703116425e-06,
+      "loss": 0.3958,
+      "step": 2224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2297755061134354e-06,
+      "loss": 0.6234,
+      "step": 2225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221221526129715e-06,
+      "loss": 0.4659,
+      "step": 2226
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.212670770464102e-06,
+      "loss": 0.4282,
+      "step": 2227
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.20412324641271e-06,
+      "loss": 0.3047,
+      "step": 2228
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.195578961268881e-06,
+      "loss": 0.7585,
+      "step": 2229
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187037922323198e-06,
+      "loss": 0.4759,
+      "step": 2230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.178500136863477e-06,
+      "loss": 0.3728,
+      "step": 2231
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.169965612174744e-06,
+      "loss": 0.0843,
+      "step": 2232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161434355539258e-06,
+      "loss": 1.2831,
+      "step": 2233
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1529063742364844e-06,
+      "loss": 0.8554,
+      "step": 2234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144381675543092e-06,
+      "loss": 0.5415,
+      "step": 2235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.135860266732952e-06,
+      "loss": 0.2921,
+      "step": 2236
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127342155077127e-06,
+      "loss": 0.443,
+      "step": 2237
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118827347843862e-06,
+      "loss": 0.9841,
+      "step": 2238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.110315852298586e-06,
+      "loss": 0.645,
+      "step": 2239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101807675703906e-06,
+      "loss": 1.1439,
+      "step": 2240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.093302825319589e-06,
+      "loss": 0.0771,
+      "step": 2241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084801308402579e-06,
+      "loss": 0.1194,
+      "step": 2242
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07630313220695e-06,
+      "loss": 0.2109,
+      "step": 2243
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.067808303983949e-06,
+      "loss": 0.6468,
+      "step": 2244
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059316830981954e-06,
+      "loss": 0.1938,
+      "step": 2245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050828720446487e-06,
+      "loss": 0.2435,
+      "step": 2246
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042343979620198e-06,
+      "loss": 0.5509,
+      "step": 2247
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033862615742859e-06,
+      "loss": 1.0045,
+      "step": 2248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025384636051361e-06,
+      "loss": 0.3905,
+      "step": 2249
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016910047779714e-06,
+      "loss": 0.2055,
+      "step": 2250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008438858159025e-06,
+      "loss": 0.7599,
+      "step": 2251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.99997107441751e-06,
+      "loss": 0.7247,
+      "step": 2252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991506703780475e-06,
+      "loss": 0.2329,
+      "step": 2253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 0.165,
+      "step": 2254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974588230706484e-06,
+      "loss": 0.8686,
+      "step": 2255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966134142705557e-06,
+      "loss": 0.8378,
+      "step": 2256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957683496681143e-06,
+      "loss": 0.5268,
+      "step": 2257
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949236299843925e-06,
+      "loss": 0.5213,
+      "step": 2258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.940792559401648e-06,
+      "loss": 0.4147,
+      "step": 2259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932352282559093e-06,
+      "loss": 0.398,
+      "step": 2260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923915476518097e-06,
+      "loss": 0.614,
+      "step": 2261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.915482148477537e-06,
+      "loss": 0.3956,
+      "step": 2262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907052305633315e-06,
+      "loss": 0.3809,
+      "step": 2263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898625955178362e-06,
+      "loss": 0.4004,
+      "step": 2264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.890203104302634e-06,
+      "loss": 0.1534,
+      "step": 2265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881783760193093e-06,
+      "loss": 0.7212,
+      "step": 2266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87336793003371e-06,
+      "loss": 0.398,
+      "step": 2267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864955621005465e-06,
+      "loss": 0.3433,
+      "step": 2268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856546840286325e-06,
+      "loss": 0.2855,
+      "step": 2269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848141595051256e-06,
+      "loss": 0.6157,
+      "step": 2270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.839739892472192e-06,
+      "loss": 0.2244,
+      "step": 2271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.831341739718055e-06,
+      "loss": 0.0903,
+      "step": 2272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8229471439547436e-06,
+      "loss": 0.5654,
+      "step": 2273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8145561123451086e-06,
+      "loss": 0.1009,
+      "step": 2274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806168652048967e-06,
+      "loss": 0.724,
+      "step": 2275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797784770223085e-06,
+      "loss": 0.1806,
+      "step": 2276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789404474021178e-06,
+      "loss": 0.5044,
+      "step": 2277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.781027770593901e-06,
+      "loss": 0.5813,
+      "step": 2278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772654667088842e-06,
+      "loss": 0.2444,
+      "step": 2279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764285170650521e-06,
+      "loss": 0.4204,
+      "step": 2280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7559192884203756e-06,
+      "loss": 0.2169,
+      "step": 2281
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747557027536763e-06,
+      "loss": 0.7114,
+      "step": 2282
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.739198395134947e-06,
+      "loss": 0.2784,
+      "step": 2283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730843398347101e-06,
+      "loss": 0.6747,
+      "step": 2284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722492044302286e-06,
+      "loss": 0.4397,
+      "step": 2285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714144340126471e-06,
+      "loss": 0.3126,
+      "step": 2286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705800292942498e-06,
+      "loss": 0.4421,
+      "step": 2287
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697459909870084e-06,
+      "loss": 0.7063,
+      "step": 2288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689123198025836e-06,
+      "loss": 1.0006,
+      "step": 2289
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6807901645232175e-06,
+      "loss": 0.2464,
+      "step": 2290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672460816472556e-06,
+      "loss": 0.2597,
+      "step": 2291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664135160981032e-06,
+      "loss": 0.3239,
+      "step": 2292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655813205152678e-06,
+      "loss": 0.6595,
+      "step": 2293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64749495608837e-06,
+      "loss": 0.4048,
+      "step": 2294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.639180420885817e-06,
+      "loss": 0.6938,
+      "step": 2295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630869606639566e-06,
+      "loss": 0.6033,
+      "step": 2296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622562520440977e-06,
+      "loss": 0.2656,
+      "step": 2297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614259169378251e-06,
+      "loss": 0.2738,
+      "step": 2298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605959560536376e-06,
+      "loss": 0.1603,
+      "step": 2299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5976637009971634e-06,
+      "loss": 0.4897,
+      "step": 2300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.589371597839215e-06,
+      "loss": 0.3017,
+      "step": 2301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581083258137943e-06,
+      "loss": 0.253,
+      "step": 2302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572798688965539e-06,
+      "loss": 0.3692,
+      "step": 2303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564517897390962e-06,
+      "loss": 0.548,
+      "step": 2304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.556240890479978e-06,
+      "loss": 0.4101,
+      "step": 2305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547967675295102e-06,
+      "loss": 1.0742,
+      "step": 2306
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.53969825889562e-06,
+      "loss": 0.5327,
+      "step": 2307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.531432648337578e-06,
+      "loss": 0.3978,
+      "step": 2308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523170850673772e-06,
+      "loss": 0.1587,
+      "step": 2309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.514912872953746e-06,
+      "loss": 0.2793,
+      "step": 2310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5066587222237845e-06,
+      "loss": 0.2132,
+      "step": 2311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498408405526905e-06,
+      "loss": 0.3549,
+      "step": 2312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.490161929902853e-06,
+      "loss": 0.5956,
+      "step": 2313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481919302388108e-06,
+      "loss": 0.8393,
+      "step": 2314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4736805300158455e-06,
+      "loss": 0.1807,
+      "step": 2315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465445619815965e-06,
+      "loss": 0.3536,
+      "step": 2316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457214578815068e-06,
+      "loss": 0.1575,
+      "step": 2317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448987414036457e-06,
+      "loss": 0.4684,
+      "step": 2318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440764132500125e-06,
+      "loss": 0.5111,
+      "step": 2319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43254474122275e-06,
+      "loss": 0.3227,
+      "step": 2320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424329247217688e-06,
+      "loss": 0.6993,
+      "step": 2321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416117657494977e-06,
+      "loss": 0.2296,
+      "step": 2322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.407909979061319e-06,
+      "loss": 0.6036,
+      "step": 2323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399706218920078e-06,
+      "loss": 0.2103,
+      "step": 2324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391506384071278e-06,
+      "loss": 0.9135,
+      "step": 2325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38331048151159e-06,
+      "loss": 0.2192,
+      "step": 2326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3751185182343326e-06,
+      "loss": 0.143,
+      "step": 2327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.366930501229459e-06,
+      "loss": 0.3265,
+      "step": 2328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35874643748356e-06,
+      "loss": 0.4615,
+      "step": 2329
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.350566333979852e-06,
+      "loss": 0.1319,
+      "step": 2330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342390197698178e-06,
+      "loss": 0.3459,
+      "step": 2331
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3342180356149756e-06,
+      "loss": 0.0652,
+      "step": 2332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32604985470331e-06,
+      "loss": 0.2086,
+      "step": 2333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31788566193285e-06,
+      "loss": 0.351,
+      "step": 2334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309725464269852e-06,
+      "loss": 0.8181,
+      "step": 2335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3015692686771725e-06,
+      "loss": 1.0106,
+      "step": 2336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293417082114235e-06,
+      "loss": 0.2015,
+      "step": 2337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2852689115370685e-06,
+      "loss": 1.0468,
+      "step": 2338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2771247638982556e-06,
+      "loss": 0.4765,
+      "step": 2339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.268984646146957e-06,
+      "loss": 0.252,
+      "step": 2340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.260848565228882e-06,
+      "loss": 0.2756,
+      "step": 2341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252716528086319e-06,
+      "loss": 1.0596,
+      "step": 2342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244588541658078e-06,
+      "loss": 1.6129,
+      "step": 2343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.236464612879529e-06,
+      "loss": 0.2892,
+      "step": 2344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228344748682574e-06,
+      "loss": 0.9569,
+      "step": 2345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.220228955995654e-06,
+      "loss": 0.204,
+      "step": 2346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2121172417437345e-06,
+      "loss": 0.9523,
+      "step": 2347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.204009612848288e-06,
+      "loss": 0.2095,
+      "step": 2348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.19590607622731e-06,
+      "loss": 0.719,
+      "step": 2349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.187806638795313e-06,
+      "loss": 0.378,
+      "step": 2350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1797113074633e-06,
+      "loss": 0.3722,
+      "step": 2351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171620089138774e-06,
+      "loss": 1.0227,
+      "step": 2352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163532990725728e-06,
+      "loss": 0.5596,
+      "step": 2353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15545001912464e-06,
+      "loss": 0.4512,
+      "step": 2354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.147371181232468e-06,
+      "loss": 0.27,
+      "step": 2355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.139296483942639e-06,
+      "loss": 0.8225,
+      "step": 2356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13122593414505e-06,
+      "loss": 0.4869,
+      "step": 2357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1231595387260655e-06,
+      "loss": 0.2415,
+      "step": 2358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11509730456849e-06,
+      "loss": 0.8108,
+      "step": 2359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107039238551588e-06,
+      "loss": 0.1921,
+      "step": 2360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098985347551061e-06,
+      "loss": 0.7365,
+      "step": 2361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090935638439061e-06,
+      "loss": 0.1707,
+      "step": 2362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082890118084159e-06,
+      "loss": 0.14,
+      "step": 2363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0748487933513564e-06,
+      "loss": 0.4158,
+      "step": 2364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0668116711020675e-06,
+      "loss": 0.3557,
+      "step": 2365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058778758194134e-06,
+      "loss": 0.6417,
+      "step": 2366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050750061481799e-06,
+      "loss": 0.5867,
+      "step": 2367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042725587815707e-06,
+      "loss": 0.1333,
+      "step": 2368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.034705344042898e-06,
+      "loss": 0.1946,
+      "step": 2369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0266893370068096e-06,
+      "loss": 0.1708,
+      "step": 2370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018677573547255e-06,
+      "loss": 0.2627,
+      "step": 2371
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010670060500433e-06,
+      "loss": 0.1293,
+      "step": 2372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002666804698911e-06,
+      "loss": 0.4852,
+      "step": 2373
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994667812971633e-06,
+      "loss": 0.5008,
+      "step": 2374
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9866730921439e-06,
+      "loss": 0.2142,
+      "step": 2375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978682649037356e-06,
+      "loss": 0.6047,
+      "step": 2376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9706964904700096e-06,
+      "loss": 0.4391,
+      "step": 2377
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962714623256217e-06,
+      "loss": 0.1899,
+      "step": 2378
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954737054206658e-06,
+      "loss": 0.9023,
+      "step": 2379
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946763790128362e-06,
+      "loss": 1.4613,
+      "step": 2380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93879483782466e-06,
+      "loss": 0.8208,
+      "step": 2381
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.930830204095233e-06,
+      "loss": 0.3266,
+      "step": 2382
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922869895736058e-06,
+      "loss": 0.3452,
+      "step": 2383
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914913919539429e-06,
+      "loss": 0.8321,
+      "step": 2384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906962282293941e-06,
+      "loss": 0.7127,
+      "step": 2385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899014990784485e-06,
+      "loss": 1.1055,
+      "step": 2386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891072051792249e-06,
+      "loss": 0.9572,
+      "step": 2387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8831334720947035e-06,
+      "loss": 0.3635,
+      "step": 2388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875199258465594e-06,
+      "loss": 0.1555,
+      "step": 2389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867269417674956e-06,
+      "loss": 0.2178,
+      "step": 2390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8593439564890844e-06,
+      "loss": 0.0847,
+      "step": 2391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851422881670529e-06,
+      "loss": 0.7499,
+      "step": 2392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.843506199978104e-06,
+      "loss": 0.4546,
+      "step": 2393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835593918166885e-06,
+      "loss": 0.1872,
+      "step": 2394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.827686042988181e-06,
+      "loss": 1.187,
+      "step": 2395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8197825811895425e-06,
+      "loss": 0.9172,
+      "step": 2396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8118835395147565e-06,
+      "loss": 0.5064,
+      "step": 2397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803988924703839e-06,
+      "loss": 0.265,
+      "step": 2398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796098743493025e-06,
+      "loss": 0.705,
+      "step": 2399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788213002614772e-06,
+      "loss": 0.8738,
+      "step": 2400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.780331708797744e-06,
+      "loss": 0.3878,
+      "step": 2401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772454868766814e-06,
+      "loss": 0.2919,
+      "step": 2402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764582489243049e-06,
+      "loss": 0.389,
+      "step": 2403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7567145769437184e-06,
+      "loss": 0.1643,
+      "step": 2404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748851138582269e-06,
+      "loss": 0.3532,
+      "step": 2405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740992180868344e-06,
+      "loss": 0.2736,
+      "step": 2406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733137710507753e-06,
+      "loss": 0.2896,
+      "step": 2407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7252877342024825e-06,
+      "loss": 0.4831,
+      "step": 2408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.717442258650672e-06,
+      "loss": 0.6584,
+      "step": 2409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709601290546638e-06,
+      "loss": 0.9879,
+      "step": 2410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701764836580841e-06,
+      "loss": 0.1284,
+      "step": 2411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693932903439893e-06,
+      "loss": 0.2469,
+      "step": 2412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686105497806545e-06,
+      "loss": 0.5351,
+      "step": 2413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678282626359688e-06,
+      "loss": 0.5464,
+      "step": 2414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.670464295774343e-06,
+      "loss": 0.8074,
+      "step": 2415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662650512721656e-06,
+      "loss": 0.8688,
+      "step": 2416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654841283868894e-06,
+      "loss": 0.5274,
+      "step": 2417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647036615879434e-06,
+      "loss": 0.5416,
+      "step": 2418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6392365154127735e-06,
+      "loss": 0.1645,
+      "step": 2419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631440989124496e-06,
+      "loss": 0.3623,
+      "step": 2420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623650043666293e-06,
+      "loss": 0.138,
+      "step": 2421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615863685685936e-06,
+      "loss": 0.3549,
+      "step": 2422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.608081921827303e-06,
+      "loss": 0.3786,
+      "step": 2423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6003047587303376e-06,
+      "loss": 0.1193,
+      "step": 2424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592532203031047e-06,
+      "loss": 0.1771,
+      "step": 2425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584764261361532e-06,
+      "loss": 0.0817,
+      "step": 2426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577000940349939e-06,
+      "loss": 1.0217,
+      "step": 2427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569242246620477e-06,
+      "loss": 0.5144,
+      "step": 2428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561488186793407e-06,
+      "loss": 0.3582,
+      "step": 2429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.553738767485034e-06,
+      "loss": 0.5733,
+      "step": 2430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545993995307705e-06,
+      "loss": 1.0937,
+      "step": 2431
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538253876869801e-06,
+      "loss": 0.5608,
+      "step": 2432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530518418775734e-06,
+      "loss": 0.6618,
+      "step": 2433
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522787627625932e-06,
+      "loss": 0.638,
+      "step": 2434
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515061510016859e-06,
+      "loss": 0.377,
+      "step": 2435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507340072540969e-06,
+      "loss": 0.507,
+      "step": 2436
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499623321786735e-06,
+      "loss": 0.1218,
+      "step": 2437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491911264338625e-06,
+      "loss": 0.7504,
+      "step": 2438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484203906777112e-06,
+      "loss": 0.5426,
+      "step": 2439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47650125567865e-06,
+      "loss": 0.2788,
+      "step": 2440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468803317615681e-06,
+      "loss": 0.8723,
+      "step": 2441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461110099156624e-06,
+      "loss": 0.5582,
+      "step": 2442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453421606865869e-06,
+      "loss": 0.2396,
+      "step": 2443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445737847303776e-06,
+      "loss": 0.9396,
+      "step": 2444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438058827026667e-06,
+      "loss": 0.4423,
+      "step": 2445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430384552586819e-06,
+      "loss": 0.4914,
+      "step": 2446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422715030532461e-06,
+      "loss": 0.717,
+      "step": 2447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415050267407762e-06,
+      "loss": 0.4051,
+      "step": 2448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407390269752838e-06,
+      "loss": 0.882,
+      "step": 2449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39973504410373e-06,
+      "loss": 0.5345,
+      "step": 2450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392084596992419e-06,
+      "loss": 0.5697,
+      "step": 2451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384438934946801e-06,
+      "loss": 0.4005,
+      "step": 2452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376798064490683e-06,
+      "loss": 1.1076,
+      "step": 2453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36916199214379e-06,
+      "loss": 0.6767,
+      "step": 2454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3615307244217595e-06,
+      "loss": 0.4405,
+      "step": 2455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353904267836121e-06,
+      "loss": 0.657,
+      "step": 2456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3462826288943e-06,
+      "loss": 0.4952,
+      "step": 2457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3386658140996114e-06,
+      "loss": 0.4811,
+      "step": 2458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331053829951256e-06,
+      "loss": 0.2048,
+      "step": 2459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.323446682944309e-06,
+      "loss": 0.757,
+      "step": 2460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3158443795697215e-06,
+      "loss": 0.3974,
+      "step": 2461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308246926314307e-06,
+      "loss": 0.4402,
+      "step": 2462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300654329660755e-06,
+      "loss": 0.606,
+      "step": 2463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.293066596087587e-06,
+      "loss": 0.1819,
+      "step": 2464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2854837320691956e-06,
+      "loss": 0.4415,
+      "step": 2465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277905744075804e-06,
+      "loss": 0.443,
+      "step": 2466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.27033263857349e-06,
+      "loss": 0.954,
+      "step": 2467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262764422024157e-06,
+      "loss": 1.1069,
+      "step": 2468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255201100885529e-06,
+      "loss": 0.4507,
+      "step": 2469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247642681611161e-06,
+      "loss": 0.1013,
+      "step": 2470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240089170650433e-06,
+      "loss": 0.6042,
+      "step": 2471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232540574448524e-06,
+      "loss": 0.2755,
+      "step": 2472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224996899446425e-06,
+      "loss": 0.5435,
+      "step": 2473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217458152080927e-06,
+      "loss": 0.4584,
+      "step": 2474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209924338784617e-06,
+      "loss": 0.9287,
+      "step": 2475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20239546598587e-06,
+      "loss": 0.2728,
+      "step": 2476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194871540108849e-06,
+      "loss": 0.5076,
+      "step": 2477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187352567573489e-06,
+      "loss": 0.3909,
+      "step": 2478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179838554795515e-06,
+      "loss": 0.1655,
+      "step": 2479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172329508186396e-06,
+      "loss": 0.3523,
+      "step": 2480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.164825434153381e-06,
+      "loss": 0.2029,
+      "step": 2481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157326339099467e-06,
+      "loss": 0.135,
+      "step": 2482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149832229423412e-06,
+      "loss": 0.7176,
+      "step": 2483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142343111519712e-06,
+      "loss": 0.2849,
+      "step": 2484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1348589917786105e-06,
+      "loss": 0.1904,
+      "step": 2485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127379876586071e-06,
+      "loss": 0.3412,
+      "step": 2486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119905772323809e-06,
+      "loss": 0.1872,
+      "step": 2487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112436685369248e-06,
+      "loss": 0.4595,
+      "step": 2488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1049726220955365e-06,
+      "loss": 0.3407,
+      "step": 2489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0975135888715316e-06,
+      "loss": 0.2527,
+      "step": 2490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090059592061811e-06,
+      "loss": 0.0825,
+      "step": 2491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0826106380266395e-06,
+      "loss": 0.1367,
+      "step": 2492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075166733121985e-06,
+      "loss": 0.7319,
+      "step": 2493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067727883699508e-06,
+      "loss": 0.4003,
+      "step": 2494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060294096106561e-06,
+      "loss": 0.1299,
+      "step": 2495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05286537668617e-06,
+      "loss": 0.1407,
+      "step": 2496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0454417317770334e-06,
+      "loss": 1.0542,
+      "step": 2497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038023167713522e-06,
+      "loss": 0.4597,
+      "step": 2498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030609690825682e-06,
+      "loss": 0.139,
+      "step": 2499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0232013074392065e-06,
+      "loss": 0.2997,
+      "step": 2500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0157980238754465e-06,
+      "loss": 0.3288,
+      "step": 2501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.008399846451402e-06,
+      "loss": 0.9255,
+      "step": 2502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001006781479715e-06,
+      "loss": 0.6731,
+      "step": 2503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9936188352686645e-06,
+      "loss": 0.6588,
+      "step": 2504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986236014122165e-06,
+      "loss": 0.8307,
+      "step": 2505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978858324339752e-06,
+      "loss": 0.1876,
+      "step": 2506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971485772216595e-06,
+      "loss": 1.1246,
+      "step": 2507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964118364043463e-06,
+      "loss": 0.4212,
+      "step": 2508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956756106106746e-06,
+      "loss": 0.249,
+      "step": 2509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949399004688435e-06,
+      "loss": 0.1748,
+      "step": 2510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942047066066131e-06,
+      "loss": 1.0374,
+      "step": 2511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9347002965130165e-06,
+      "loss": 0.4996,
+      "step": 2512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9273587022978754e-06,
+      "loss": 0.2021,
+      "step": 2513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920022289685057e-06,
+      "loss": 0.2048,
+      "step": 2514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912691064934513e-06,
+      "loss": 0.6029,
+      "step": 2515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905365034301754e-06,
+      "loss": 0.1833,
+      "step": 2516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898044204037861e-06,
+      "loss": 0.262,
+      "step": 2517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890728580389478e-06,
+      "loss": 0.5067,
+      "step": 2518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883418169598808e-06,
+      "loss": 0.4284,
+      "step": 2519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8761129779036054e-06,
+      "loss": 0.4972,
+      "step": 2520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868813011537169e-06,
+      "loss": 0.3598,
+      "step": 2521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861518276728341e-06,
+      "loss": 0.7376,
+      "step": 2522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854228779701498e-06,
+      "loss": 0.7484,
+      "step": 2523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846944526676556e-06,
+      "loss": 0.9729,
+      "step": 2524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839665523868942e-06,
+      "loss": 0.8678,
+      "step": 2525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832391777489607e-06,
+      "loss": 0.5405,
+      "step": 2526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.82512329374503e-06,
+      "loss": 0.8692,
+      "step": 2527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817860078837186e-06,
+      "loss": 0.1085,
+      "step": 2528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8106021389635583e-06,
+      "loss": 0.3708,
+      "step": 2529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8033494803171224e-06,
+      "loss": 0.7071,
+      "step": 2530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7961021090863625e-06,
+      "loss": 0.657,
+      "step": 2531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78886003145524e-06,
+      "loss": 0.1733,
+      "step": 2532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7816232536032017e-06,
+      "loss": 0.5104,
+      "step": 2533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7743917817051723e-06,
+      "loss": 0.3636,
+      "step": 2534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.76716562193155e-06,
+      "loss": 0.4569,
+      "step": 2535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.759944780448199e-06,
+      "loss": 0.3151,
+      "step": 2536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7527292634164468e-06,
+      "loss": 0.3425,
+      "step": 2537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.745519076993078e-06,
+      "loss": 0.8592,
+      "step": 2538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738314227330324e-06,
+      "loss": 0.612,
+      "step": 2539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7311147205758767e-06,
+      "loss": 0.4476,
+      "step": 2540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7239205628728483e-06,
+      "loss": 0.4324,
+      "step": 2541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7167317603597975e-06,
+      "loss": 0.8044,
+      "step": 2542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7095483191707206e-06,
+      "loss": 0.5255,
+      "step": 2543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7023702454350284e-06,
+      "loss": 1.1665,
+      "step": 2544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6951975452775567e-06,
+      "loss": 0.4031,
+      "step": 2545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6880302248185528e-06,
+      "loss": 0.6015,
+      "step": 2546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680868290173677e-06,
+      "loss": 0.2944,
+      "step": 2547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.673711747453994e-06,
+      "loss": 0.3458,
+      "step": 2548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666560602765965e-06,
+      "loss": 0.2846,
+      "step": 2549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6594148622114465e-06,
+      "loss": 0.1694,
+      "step": 2550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.652274531887686e-06,
+      "loss": 0.8621,
+      "step": 2551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645139617887312e-06,
+      "loss": 0.3391,
+      "step": 2552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6380101262983325e-06,
+      "loss": 0.1652,
+      "step": 2553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6308860632041275e-06,
+      "loss": 0.4014,
+      "step": 2554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.623767434683444e-06,
+      "loss": 0.1897,
+      "step": 2555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6166542468103982e-06,
+      "loss": 0.3581,
+      "step": 2556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.609546505654462e-06,
+      "loss": 0.3195,
+      "step": 2557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.602444217280445e-06,
+      "loss": 0.4585,
+      "step": 2558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595347387748529e-06,
+      "loss": 0.5355,
+      "step": 2559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5882560231142205e-06,
+      "loss": 0.4943,
+      "step": 2560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5811701294283684e-06,
+      "loss": 0.3331,
+      "step": 2561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574089712737152e-06,
+      "loss": 0.4038,
+      "step": 2562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5670147790820786e-06,
+      "loss": 0.111,
+      "step": 2563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.559945334499978e-06,
+      "loss": 0.4786,
+      "step": 2564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5528813850229915e-06,
+      "loss": 0.211,
+      "step": 2565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5458229366785778e-06,
+      "loss": 0.9218,
+      "step": 2566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.538769995489494e-06,
+      "loss": 0.2535,
+      "step": 2567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531722567473813e-06,
+      "loss": 0.9102,
+      "step": 2568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5246806586448845e-06,
+      "loss": 0.9315,
+      "step": 2569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5176442750113593e-06,
+      "loss": 0.145,
+      "step": 2570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510613422577169e-06,
+      "loss": 0.4009,
+      "step": 2571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.503588107341538e-06,
+      "loss": 0.1709,
+      "step": 2572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.496568335298953e-06,
+      "loss": 0.2438,
+      "step": 2573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4895541124391663e-06,
+      "loss": 0.369,
+      "step": 2574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482545444747214e-06,
+      "loss": 0.5492,
+      "step": 2575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.475542338203377e-06,
+      "loss": 0.3708,
+      "step": 2576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4685447987831967e-06,
+      "loss": 0.412,
+      "step": 2577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.461552832457462e-06,
+      "loss": 0.1957,
+      "step": 2578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.45456644519221e-06,
+      "loss": 0.5748,
+      "step": 2579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.447585642948712e-06,
+      "loss": 0.2465,
+      "step": 2580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440610431683479e-06,
+      "loss": 0.8258,
+      "step": 2581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4336408173482485e-06,
+      "loss": 0.7513,
+      "step": 2582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.426676805889979e-06,
+      "loss": 0.2464,
+      "step": 2583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4197184032508636e-06,
+      "loss": 0.4672,
+      "step": 2584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4127656153682866e-06,
+      "loss": 0.7142,
+      "step": 2585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.405818448174857e-06,
+      "loss": 0.5815,
+      "step": 2586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3988769075983796e-06,
+      "loss": 0.3723,
+      "step": 2587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.391940999561871e-06,
+      "loss": 0.3277,
+      "step": 2588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.385010729983529e-06,
+      "loss": 0.1971,
+      "step": 2589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.378086104776743e-06,
+      "loss": 0.7748,
+      "step": 2590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.371167129850089e-06,
+      "loss": 0.4773,
+      "step": 2591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3642538111073207e-06,
+      "loss": 0.6452,
+      "step": 2592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.357346154447364e-06,
+      "loss": 0.8553,
+      "step": 2593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.350444165764315e-06,
+      "loss": 0.8722,
+      "step": 2594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.343547850947434e-06,
+      "loss": 0.6605,
+      "step": 2595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3366572158811384e-06,
+      "loss": 0.6067,
+      "step": 2596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3297722664450005e-06,
+      "loss": 0.5699,
+      "step": 2597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.32289300851374e-06,
+      "loss": 0.2802,
+      "step": 2598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3160194479572193e-06,
+      "loss": 0.7199,
+      "step": 2599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.309151590640446e-06,
+      "loss": 0.3595,
+      "step": 2600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3022894424235573e-06,
+      "loss": 0.6795,
+      "step": 2601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2954330091618104e-06,
+      "loss": 0.5189,
+      "step": 2602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2885822967055957e-06,
+      "loss": 0.4163,
+      "step": 2603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2817373109004247e-06,
+      "loss": 1.3277,
+      "step": 2604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.274898057586916e-06,
+      "loss": 0.1723,
+      "step": 2605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2680645426007984e-06,
+      "loss": 0.1842,
+      "step": 2606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2612367717729056e-06,
+      "loss": 0.4348,
+      "step": 2607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.254414750929169e-06,
+      "loss": 0.7194,
+      "step": 2608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247598485890614e-06,
+      "loss": 1.5466,
+      "step": 2609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2407879824733535e-06,
+      "loss": 0.6882,
+      "step": 2610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2339832464885846e-06,
+      "loss": 0.5292,
+      "step": 2611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2271842837425917e-06,
+      "loss": 0.0862,
+      "step": 2612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.220391100036716e-06,
+      "loss": 0.5655,
+      "step": 2613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2136037011673803e-06,
+      "loss": 0.3313,
+      "step": 2614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.206822092926065e-06,
+      "loss": 0.3397,
+      "step": 2615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2000462810993205e-06,
+      "loss": 0.1773,
+      "step": 2616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1932762714687417e-06,
+      "loss": 0.2476,
+      "step": 2617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1865120698109675e-06,
+      "loss": 0.1364,
+      "step": 2618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1797536818976894e-06,
+      "loss": 0.3412,
+      "step": 2619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.173001113495643e-06,
+      "loss": 0.4178,
+      "step": 2620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1662543703665873e-06,
+      "loss": 0.8176,
+      "step": 2621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159513458267317e-06,
+      "loss": 0.1725,
+      "step": 2622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1527783829496483e-06,
+      "loss": 0.4008,
+      "step": 2623
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1460491501604207e-06,
+      "loss": 0.2511,
+      "step": 2624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1393257656414842e-06,
+      "loss": 0.5218,
+      "step": 2625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1326082351297025e-06,
+      "loss": 0.4271,
+      "step": 2626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1258965643569382e-06,
+      "loss": 0.1541,
+      "step": 2627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119190759050069e-06,
+      "loss": 0.5952,
+      "step": 2628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.112490824930946e-06,
+      "loss": 0.3563,
+      "step": 2629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1057967677164258e-06,
+      "loss": 0.521,
+      "step": 2630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0991085931183418e-06,
+      "loss": 0.6732,
+      "step": 2631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0924263068435213e-06,
+      "loss": 0.451,
+      "step": 2632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.085749914593752e-06,
+      "loss": 0.3367,
+      "step": 2633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0790794220658047e-06,
+      "loss": 0.2151,
+      "step": 2634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0724148349513995e-06,
+      "loss": 0.2614,
+      "step": 2635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0657561589372377e-06,
+      "loss": 0.6332,
+      "step": 2636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0591033997049646e-06,
+      "loss": 0.486,
+      "step": 2637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0524565629311787e-06,
+      "loss": 0.2151,
+      "step": 2638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0458156542874283e-06,
+      "loss": 1.2802,
+      "step": 2639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039180679440199e-06,
+      "loss": 0.4212,
+      "step": 2640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032551644050917e-06,
+      "loss": 0.6584,
+      "step": 2641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0259285537759375e-06,
+      "loss": 0.8318,
+      "step": 2642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0193114142665424e-06,
+      "loss": 0.1632,
+      "step": 2643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0127002311689446e-06,
+      "loss": 0.327,
+      "step": 2644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.006095010124267e-06,
+      "loss": 0.2035,
+      "step": 2645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99949575676854e-06,
+      "loss": 0.593,
+      "step": 2646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9929024767327088e-06,
+      "loss": 0.8479,
+      "step": 2647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9863151756426255e-06,
+      "loss": 0.5075,
+      "step": 2648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9797338591190362e-06,
+      "loss": 0.5196,
+      "step": 2649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9731585327775814e-06,
+      "loss": 0.2626,
+      "step": 2650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.966589202228781e-06,
+      "loss": 0.3817,
+      "step": 2651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9600258730780564e-06,
+      "loss": 0.5428,
+      "step": 2652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9534685509256954e-06,
+      "loss": 1.1203,
+      "step": 2653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9469172413668647e-06,
+      "loss": 0.8896,
+      "step": 2654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9403719499916008e-06,
+      "loss": 0.4786,
+      "step": 2655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.933832682384802e-06,
+      "loss": 0.3299,
+      "step": 2656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.927299444126229e-06,
+      "loss": 0.8985,
+      "step": 2657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9207722407905004e-06,
+      "loss": 0.1818,
+      "step": 2658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.914251077947077e-06,
+      "loss": 0.4581,
+      "step": 2659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9077359611602773e-06,
+      "loss": 0.6572,
+      "step": 2660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9012268959892562e-06,
+      "loss": 1.14,
+      "step": 2661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.894723887987997e-06,
+      "loss": 0.6955,
+      "step": 2662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.888226942705319e-06,
+      "loss": 0.9261,
+      "step": 2663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.881736065684878e-06,
+      "loss": 0.0841,
+      "step": 2664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.875251262465142e-06,
+      "loss": 0.3344,
+      "step": 2665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8687725385793973e-06,
+      "loss": 0.788,
+      "step": 2666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.862299899555746e-06,
+      "loss": 0.3933,
+      "step": 2667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8558333509170943e-06,
+      "loss": 0.443,
+      "step": 2668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8493728981811553e-06,
+      "loss": 0.2674,
+      "step": 2669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.842918546860438e-06,
+      "loss": 0.3281,
+      "step": 2670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8364703024622474e-06,
+      "loss": 0.6084,
+      "step": 2671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8300281704886778e-06,
+      "loss": 0.0771,
+      "step": 2672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8235921564366043e-06,
+      "loss": 0.4822,
+      "step": 2673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.817162265797685e-06,
+      "loss": 0.414,
+      "step": 2674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.81073850405835e-06,
+      "loss": 0.7023,
+      "step": 2675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8043208766998088e-06,
+      "loss": 0.2086,
+      "step": 2676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7979093891980257e-06,
+      "loss": 0.5554,
+      "step": 2677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791504047023734e-06,
+      "loss": 0.6256,
+      "step": 2678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.78510485564241e-06,
+      "loss": 0.2584,
+      "step": 2679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7787118205143005e-06,
+      "loss": 0.8273,
+      "step": 2680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.772324947094388e-06,
+      "loss": 0.3443,
+      "step": 2681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7659442408324e-06,
+      "loss": 0.3268,
+      "step": 2682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.759569707172799e-06,
+      "loss": 0.4404,
+      "step": 2683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7532013515547863e-06,
+      "loss": 0.3048,
+      "step": 2684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.746839179412286e-06,
+      "loss": 0.476,
+      "step": 2685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7404831961739487e-06,
+      "loss": 0.8498,
+      "step": 2686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7341334072631456e-06,
+      "loss": 0.2882,
+      "step": 2687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7277898180979544e-06,
+      "loss": 0.3826,
+      "step": 2688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.721452434091182e-06,
+      "loss": 0.1598,
+      "step": 2689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7151212606503164e-06,
+      "loss": 0.4298,
+      "step": 2690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7087963031775576e-06,
+      "loss": 0.5985,
+      "step": 2691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.702477567069809e-06,
+      "loss": 0.407,
+      "step": 2692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696165057718655e-06,
+      "loss": 0.5407,
+      "step": 2693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6898587805103715e-06,
+      "loss": 0.3108,
+      "step": 2694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.683558740825908e-06,
+      "loss": 0.4546,
+      "step": 2695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6772649440409084e-06,
+      "loss": 0.7508,
+      "step": 2696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6709773955256748e-06,
+      "loss": 0.2906,
+      "step": 2697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6646961006451866e-06,
+      "loss": 0.3345,
+      "step": 2698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6584210647590813e-06,
+      "loss": 0.0561,
+      "step": 2699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6521522932216603e-06,
+      "loss": 0.5976,
+      "step": 2700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.645889791381877e-06,
+      "loss": 0.1916,
+      "step": 2701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.639633564583337e-06,
+      "loss": 0.654,
+      "step": 2702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633383618164289e-06,
+      "loss": 0.1623,
+      "step": 2703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.627139957457623e-06,
+      "loss": 0.5446,
+      "step": 2704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6209025877908746e-06,
+      "loss": 0.1643,
+      "step": 2705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.614671514486197e-06,
+      "loss": 0.6236,
+      "step": 2706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6084467428603786e-06,
+      "loss": 0.2473,
+      "step": 2707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6022282782248277e-06,
+      "loss": 0.2256,
+      "step": 2708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5960161258855807e-06,
+      "loss": 0.319,
+      "step": 2709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5898102911432755e-06,
+      "loss": 0.4527,
+      "step": 2710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5836107792931653e-06,
+      "loss": 0.3455,
+      "step": 2711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.577417595625107e-06,
+      "loss": 0.2213,
+      "step": 2712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5712307454235585e-06,
+      "loss": 0.3587,
+      "step": 2713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565050233967573e-06,
+      "loss": 0.3905,
+      "step": 2714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5588760665307953e-06,
+      "loss": 0.3828,
+      "step": 2715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5527082483814537e-06,
+      "loss": 1.0979,
+      "step": 2716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546546784782371e-06,
+      "loss": 0.6272,
+      "step": 2717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.54039168099093e-06,
+      "loss": 1.0196,
+      "step": 2718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5342429422590984e-06,
+      "loss": 0.4573,
+      "step": 2719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5281005738334087e-06,
+      "loss": 0.2521,
+      "step": 2720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.521964580954964e-06,
+      "loss": 0.6572,
+      "step": 2721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.515834968859423e-06,
+      "loss": 0.3006,
+      "step": 2722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5097117427769925e-06,
+      "loss": 0.5791,
+      "step": 2723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5035949079324396e-06,
+      "loss": 0.1887,
+      "step": 2724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4974844695450794e-06,
+      "loss": 0.4171,
+      "step": 2725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4913804328287626e-06,
+      "loss": 0.4763,
+      "step": 2726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4852828029918818e-06,
+      "loss": 0.6915,
+      "step": 2727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4791915852373604e-06,
+      "loss": 0.1152,
+      "step": 2728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4731067847626512e-06,
+      "loss": 0.3072,
+      "step": 2729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4670284067597316e-06,
+      "loss": 0.6578,
+      "step": 2730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4609564564151e-06,
+      "loss": 0.5307,
+      "step": 2731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.454890938909764e-06,
+      "loss": 0.1255,
+      "step": 2732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4488318594192582e-06,
+      "loss": 1.0802,
+      "step": 2733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4427792231136047e-06,
+      "loss": 0.3624,
+      "step": 2734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.436733035157337e-06,
+      "loss": 0.0966,
+      "step": 2735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4306933007094834e-06,
+      "loss": 0.8253,
+      "step": 2736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.424660024923575e-06,
+      "loss": 1.3565,
+      "step": 2737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4186332129476196e-06,
+      "loss": 0.8132,
+      "step": 2738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4126128699241193e-06,
+      "loss": 0.2675,
+      "step": 2739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406599000990043e-06,
+      "loss": 0.5471,
+      "step": 2740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4005916112768524e-06,
+      "loss": 0.1999,
+      "step": 2741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39459070591047e-06,
+      "loss": 0.3318,
+      "step": 2742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.388596290011288e-06,
+      "loss": 0.279,
+      "step": 2743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3826083686941614e-06,
+      "loss": 0.5495,
+      "step": 2744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3766269470684045e-06,
+      "loss": 0.6233,
+      "step": 2745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3706520302377823e-06,
+      "loss": 0.2474,
+      "step": 2746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3646836233005133e-06,
+      "loss": 0.2805,
+      "step": 2747
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3587217313492572e-06,
+      "loss": 0.2733,
+      "step": 2748
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3527663594711225e-06,
+      "loss": 0.4747,
+      "step": 2749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.346817512747649e-06,
+      "loss": 0.5605,
+      "step": 2750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3408751962548037e-06,
+      "loss": 0.2898,
+      "step": 2751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3349394150629856e-06,
+      "loss": 0.5622,
+      "step": 2752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3290101742370243e-06,
+      "loss": 0.4029,
+      "step": 2753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3230874788361612e-06,
+      "loss": 0.3653,
+      "step": 2754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3171713339140554e-06,
+      "loss": 0.6298,
+      "step": 2755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.311261744518769e-06,
+      "loss": 0.702,
+      "step": 2756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.305358715692784e-06,
+      "loss": 0.1523,
+      "step": 2757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2994622524729748e-06,
+      "loss": 0.7494,
+      "step": 2758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2935723598906168e-06,
+      "loss": 0.4289,
+      "step": 2759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287689042971376e-06,
+      "loss": 0.1902,
+      "step": 2760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2818123067353172e-06,
+      "loss": 0.3553,
+      "step": 2761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.275942156196875e-06,
+      "loss": 0.7529,
+      "step": 2762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270078596364875e-06,
+      "loss": 0.275,
+      "step": 2763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.264221632242515e-06,
+      "loss": 0.2919,
+      "step": 2764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2583712688273706e-06,
+      "loss": 0.1605,
+      "step": 2765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.252527511111381e-06,
+      "loss": 0.1476,
+      "step": 2766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2466903640808444e-06,
+      "loss": 0.298,
+      "step": 2767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2408598327164234e-06,
+      "loss": 0.2826,
+      "step": 2768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2350359219931393e-06,
+      "loss": 0.2343,
+      "step": 2769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2292186368803582e-06,
+      "loss": 0.6979,
+      "step": 2770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.223407982341793e-06,
+      "loss": 0.125,
+      "step": 2771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217603963335504e-06,
+      "loss": 0.4834,
+      "step": 2772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2118065848138838e-06,
+      "loss": 0.5449,
+      "step": 2773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2060158517236606e-06,
+      "loss": 0.8641,
+      "step": 2774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.200231769005895e-06,
+      "loss": 0.2491,
+      "step": 2775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1944543415959675e-06,
+      "loss": 0.5606,
+      "step": 2776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1886835744235913e-06,
+      "loss": 0.2482,
+      "step": 2777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18291947241278e-06,
+      "loss": 0.6772,
+      "step": 2778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1771620404818716e-06,
+      "loss": 0.2814,
+      "step": 2779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1714112835435076e-06,
+      "loss": 0.4646,
+      "step": 2780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165667206504641e-06,
+      "loss": 0.0977,
+      "step": 2781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159929814266517e-06,
+      "loss": 0.2098,
+      "step": 2782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.154199111724684e-06,
+      "loss": 0.2865,
+      "step": 2783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.148475103768969e-06,
+      "loss": 0.3172,
+      "step": 2784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1427577952835044e-06,
+      "loss": 0.3712,
+      "step": 2785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.137047191146696e-06,
+      "loss": 0.1368,
+      "step": 2786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1313432962312287e-06,
+      "loss": 0.4392,
+      "step": 2787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1256461154040653e-06,
+      "loss": 0.1441,
+      "step": 2788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.11995565352644e-06,
+      "loss": 0.1903,
+      "step": 2789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1142719154538526e-06,
+      "loss": 0.4259,
+      "step": 2790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1085949060360654e-06,
+      "loss": 0.1339,
+      "step": 2791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102924630117097e-06,
+      "loss": 0.6465,
+      "step": 2792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09726109253523e-06,
+      "loss": 0.6413,
+      "step": 2793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09160429812299e-06,
+      "loss": 0.1473,
+      "step": 2794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0859542517071452e-06,
+      "loss": 0.2451,
+      "step": 2795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.080310958108709e-06,
+      "loss": 0.4625,
+      "step": 2796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0746744221429393e-06,
+      "loss": 0.6306,
+      "step": 2797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0690446486193227e-06,
+      "loss": 0.207,
+      "step": 2798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0634216423415766e-06,
+      "loss": 0.383,
+      "step": 2799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0578054081076347e-06,
+      "loss": 0.3404,
+      "step": 2800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0521959507096712e-06,
+      "loss": 0.6692,
+      "step": 2801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046593274934062e-06,
+      "loss": 0.0796,
+      "step": 2802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040997385561405e-06,
+      "loss": 0.6737,
+      "step": 2803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0354082873665015e-06,
+      "loss": 0.5278,
+      "step": 2804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0298259851183633e-06,
+      "loss": 0.6476,
+      "step": 2805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0242504835802e-06,
+      "loss": 0.4029,
+      "step": 2806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.01868178750942e-06,
+      "loss": 0.3605,
+      "step": 2807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013119901657624e-06,
+      "loss": 0.6976,
+      "step": 2808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0075648307705986e-06,
+      "loss": 0.3881,
+      "step": 2809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0020165795883285e-06,
+      "loss": 0.8939,
+      "step": 2810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.996475152844961e-06,
+      "loss": 0.6305,
+      "step": 2811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990940555268829e-06,
+      "loss": 0.7085,
+      "step": 2812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9854127915824427e-06,
+      "loss": 0.4041,
+      "step": 2813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.979891866502476e-06,
+      "loss": 0.1591,
+      "step": 2814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9743777847397672e-06,
+      "loss": 0.1603,
+      "step": 2815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9688705509993155e-06,
+      "loss": 0.2365,
+      "step": 2816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9633701699802808e-06,
+      "loss": 0.3073,
+      "step": 2817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95787664637597e-06,
+      "loss": 0.2662,
+      "step": 2818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9523899848738435e-06,
+      "loss": 0.6567,
+      "step": 2819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9469101901555045e-06,
+      "loss": 0.5323,
+      "step": 2820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9414372668966954e-06,
+      "loss": 0.9163,
+      "step": 2821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9359712197672997e-06,
+      "loss": 0.1387,
+      "step": 2822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9305120534313295e-06,
+      "loss": 0.3105,
+      "step": 2823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925059772546929e-06,
+      "loss": 0.3035,
+      "step": 2824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9196143817663604e-06,
+      "loss": 0.9518,
+      "step": 2825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9141758857360194e-06,
+      "loss": 0.1613,
+      "step": 2826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087442890964102e-06,
+      "loss": 1.4111,
+      "step": 2827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9033195964821438e-06,
+      "loss": 0.6956,
+      "step": 2828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8979018125219551e-06,
+      "loss": 0.1349,
+      "step": 2829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.892490941838674e-06,
+      "loss": 0.4514,
+      "step": 2830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8870869890492328e-06,
+      "loss": 0.1965,
+      "step": 2831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8816899587646631e-06,
+      "loss": 0.6784,
+      "step": 2832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.876299855590088e-06,
+      "loss": 0.4588,
+      "step": 2833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8709166841247206e-06,
+      "loss": 0.5838,
+      "step": 2834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.865540448961859e-06,
+      "loss": 0.8011,
+      "step": 2835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8601711546888844e-06,
+      "loss": 0.5683,
+      "step": 2836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8548088058872504e-06,
+      "loss": 0.4172,
+      "step": 2837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8494534071324966e-06,
+      "loss": 0.3628,
+      "step": 2838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8441049629942164e-06,
+      "loss": 0.4079,
+      "step": 2839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8387634780360774e-06,
+      "loss": 0.4785,
+      "step": 2840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.833428956815807e-06,
+      "loss": 0.7582,
+      "step": 2841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8281014038851963e-06,
+      "loss": 0.5137,
+      "step": 2842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.822780823790088e-06,
+      "loss": 0.1978,
+      "step": 2843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8174672210703626e-06,
+      "loss": 0.8673,
+      "step": 2844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8121606002599667e-06,
+      "loss": 0.347,
+      "step": 2845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8068609658868774e-06,
+      "loss": 0.6565,
+      "step": 2846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.801568322473115e-06,
+      "loss": 0.6129,
+      "step": 2847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7962826745347318e-06,
+      "loss": 0.1691,
+      "step": 2848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7910040265818118e-06,
+      "loss": 0.2563,
+      "step": 2849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.785732383118467e-06,
+      "loss": 0.1954,
+      "step": 2850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7804677486428335e-06,
+      "loss": 0.5878,
+      "step": 2851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7752101276470645e-06,
+      "loss": 0.3468,
+      "step": 2852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7699595246173285e-06,
+      "loss": 0.4344,
+      "step": 2853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7647159440338136e-06,
+      "loss": 0.8082,
+      "step": 2854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.759479390370703e-06,
+      "loss": 0.447,
+      "step": 2855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7542498680961917e-06,
+      "loss": 0.4194,
+      "step": 2856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7490273816724734e-06,
+      "loss": 0.3886,
+      "step": 2857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7438119355557425e-06,
+      "loss": 0.5085,
+      "step": 2858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7386035341961805e-06,
+      "loss": 0.2425,
+      "step": 2859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7334021820379588e-06,
+      "loss": 0.8268,
+      "step": 2860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7282078835192362e-06,
+      "loss": 0.5128,
+      "step": 2861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7230206430721508e-06,
+      "loss": 0.3984,
+      "step": 2862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7178404651228187e-06,
+      "loss": 0.3189,
+      "step": 2863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7126673540913308e-06,
+      "loss": 0.2327,
+      "step": 2864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075013143917473e-06,
+      "loss": 0.0589,
+      "step": 2865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7023423504320934e-06,
+      "loss": 0.6573,
+      "step": 2866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6971904666143602e-06,
+      "loss": 0.2029,
+      "step": 2867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6920456673344931e-06,
+      "loss": 0.1926,
+      "step": 2868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6869079569823932e-06,
+      "loss": 1.0207,
+      "step": 2869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6817773399419201e-06,
+      "loss": 1.0384,
+      "step": 2870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6766538205908734e-06,
+      "loss": 0.5954,
+      "step": 2871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6715374033009945e-06,
+      "loss": 0.6178,
+      "step": 2872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6664280924379682e-06,
+      "loss": 0.644,
+      "step": 2873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6613258923614217e-06,
+      "loss": 0.686,
+      "step": 2874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6562308074249045e-06,
+      "loss": 1.277,
+      "step": 2875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6511428419759012e-06,
+      "loss": 0.1427,
+      "step": 2876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6460620003558193e-06,
+      "loss": 0.6644,
+      "step": 2877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6409882868999883e-06,
+      "loss": 0.9175,
+      "step": 2878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6359217059376552e-06,
+      "loss": 0.7213,
+      "step": 2879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6308622617919823e-06,
+      "loss": 0.1825,
+      "step": 2880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.625809958780037e-06,
+      "loss": 0.3662,
+      "step": 2881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6207648012128063e-06,
+      "loss": 0.1685,
+      "step": 2882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6157267933951637e-06,
+      "loss": 0.7539,
+      "step": 2883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6106959396258926e-06,
+      "loss": 0.3077,
+      "step": 2884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6056722441976668e-06,
+      "loss": 0.6523,
+      "step": 2885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.600655711397059e-06,
+      "loss": 0.1373,
+      "step": 2886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5956463455045268e-06,
+      "loss": 1.3338,
+      "step": 2887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5906441507944059e-06,
+      "loss": 0.181,
+      "step": 2888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5856491315349199e-06,
+      "loss": 0.2415,
+      "step": 2889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5806612919881726e-06,
+      "loss": 0.1877,
+      "step": 2890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.575680636410134e-06,
+      "loss": 0.1269,
+      "step": 2891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5707071690506504e-06,
+      "loss": 0.1897,
+      "step": 2892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5657408941534303e-06,
+      "loss": 0.4561,
+      "step": 2893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5607818159560473e-06,
+      "loss": 0.1715,
+      "step": 2894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5558299386899333e-06,
+      "loss": 0.933,
+      "step": 2895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5508852665803776e-06,
+      "loss": 0.5304,
+      "step": 2896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5459478038465158e-06,
+      "loss": 0.1532,
+      "step": 2897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5410175547013461e-06,
+      "loss": 0.2809,
+      "step": 2898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5360945233516933e-06,
+      "loss": 0.2863,
+      "step": 2899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.531178713998235e-06,
+      "loss": 0.1213,
+      "step": 2900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526270130835481e-06,
+      "loss": 0.1484,
+      "step": 2901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5213687780517827e-06,
+      "loss": 0.1136,
+      "step": 2902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5164746598293157e-06,
+      "loss": 0.9091,
+      "step": 2903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5115877803440836e-06,
+      "loss": 0.7219,
+      "step": 2904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5067081437659093e-06,
+      "loss": 0.864,
+      "step": 2905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5018357542584461e-06,
+      "loss": 0.1431,
+      "step": 2906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4969706159791564e-06,
+      "loss": 0.4302,
+      "step": 2907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4921127330793138e-06,
+      "loss": 0.22,
+      "step": 2908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4872621097040074e-06,
+      "loss": 0.4644,
+      "step": 2909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.482418749992125e-06,
+      "loss": 0.6865,
+      "step": 2910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.477582658076362e-06,
+      "loss": 0.4325,
+      "step": 2911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4727538380832095e-06,
+      "loss": 0.4527,
+      "step": 2912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4679322941329522e-06,
+      "loss": 0.4294,
+      "step": 2913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4631180303396742e-06,
+      "loss": 0.6191,
+      "step": 2914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4583110508112396e-06,
+      "loss": 0.5553,
+      "step": 2915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4535113596492977e-06,
+      "loss": 0.1841,
+      "step": 2916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4487189609492802e-06,
+      "loss": 0.3909,
+      "step": 2917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4439338588004005e-06,
+      "loss": 0.3638,
+      "step": 2918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4391560572856412e-06,
+      "loss": 0.4135,
+      "step": 2919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.434385560481758e-06,
+      "loss": 1.0027,
+      "step": 2920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4296223724592662e-06,
+      "loss": 0.3135,
+      "step": 2921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4248664972824578e-06,
+      "loss": 0.4216,
+      "step": 2922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4201179390093766e-06,
+      "loss": 0.1923,
+      "step": 2923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.415376701691823e-06,
+      "loss": 0.4358,
+      "step": 2924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4106427893753537e-06,
+      "loss": 0.8978,
+      "step": 2925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4059162060992736e-06,
+      "loss": 0.9287,
+      "step": 2926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4011969558966332e-06,
+      "loss": 0.6779,
+      "step": 2927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.396485042794229e-06,
+      "loss": 0.5497,
+      "step": 2928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3917804708125903e-06,
+      "loss": 0.2655,
+      "step": 2929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.387083243965992e-06,
+      "loss": 0.3886,
+      "step": 2930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3823933662624379e-06,
+      "loss": 0.5727,
+      "step": 2931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3777108417036544e-06,
+      "loss": 0.6326,
+      "step": 2932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.373035674285098e-06,
+      "loss": 0.2231,
+      "step": 2933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3683678679959556e-06,
+      "loss": 0.7899,
+      "step": 2934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3637074268191209e-06,
+      "loss": 0.2318,
+      "step": 2935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3590543547312108e-06,
+      "loss": 0.6944,
+      "step": 2936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3544086557025493e-06,
+      "loss": 0.2826,
+      "step": 2937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3497703336971746e-06,
+      "loss": 0.5849,
+      "step": 2938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3451393926728252e-06,
+      "loss": 0.4211,
+      "step": 2939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3405158365809445e-06,
+      "loss": 0.5118,
+      "step": 2940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3358996693666748e-06,
+      "loss": 0.4355,
+      "step": 2941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3312908949688497e-06,
+      "loss": 0.2487,
+      "step": 2942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3266895173200056e-06,
+      "loss": 1.0261,
+      "step": 2943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.322095540346352e-06,
+      "loss": 0.3785,
+      "step": 2944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3175089679677922e-06,
+      "loss": 0.1614,
+      "step": 2945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3129298040979133e-06,
+      "loss": 0.4962,
+      "step": 2946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3083580526439787e-06,
+      "loss": 0.746,
+      "step": 2947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303793717506927e-06,
+      "loss": 0.1074,
+      "step": 2948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2992368025813628e-06,
+      "loss": 0.1346,
+      "step": 2949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2946873117555692e-06,
+      "loss": 0.5185,
+      "step": 2950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2901452489114896e-06,
+      "loss": 0.6071,
+      "step": 2951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2856106179247297e-06,
+      "loss": 0.6263,
+      "step": 2952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.281083422664553e-06,
+      "loss": 0.381,
+      "step": 2953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2765636669938798e-06,
+      "loss": 0.5439,
+      "step": 2954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2720513547692804e-06,
+      "loss": 0.8867,
+      "step": 2955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2675464898409772e-06,
+      "loss": 0.3748,
+      "step": 2956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2630490760528358e-06,
+      "loss": 0.2928,
+      "step": 2957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2585591172423606e-06,
+      "loss": 0.2137,
+      "step": 2958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.254076617240706e-06,
+      "loss": 0.2009,
+      "step": 2959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.249601579872648e-06,
+      "loss": 0.6304,
+      "step": 2960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2451340089566022e-06,
+      "loss": 0.1017,
+      "step": 2961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.240673908304615e-06,
+      "loss": 0.4035,
+      "step": 2962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2362212817223562e-06,
+      "loss": 0.7977,
+      "step": 2963
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2317761330091172e-06,
+      "loss": 0.6039,
+      "step": 2964
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2273384659578092e-06,
+      "loss": 0.4996,
+      "step": 2965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2229082843549622e-06,
+      "loss": 0.1319,
+      "step": 2966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2184855919807149e-06,
+      "loss": 0.6316,
+      "step": 2967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2140703926088182e-06,
+      "loss": 0.4433,
+      "step": 2968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.20966269000663e-06,
+      "loss": 0.2338,
+      "step": 2969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2052624879351105e-06,
+      "loss": 0.6004,
+      "step": 2970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2008697901488187e-06,
+      "loss": 0.2005,
+      "step": 2971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1964846003959118e-06,
+      "loss": 0.4373,
+      "step": 2972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1921069224181413e-06,
+      "loss": 0.5428,
+      "step": 2973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.187736759950846e-06,
+      "loss": 0.1092,
+      "step": 2974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1833741167229584e-06,
+      "loss": 0.1035,
+      "step": 2975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1790189964569899e-06,
+      "loss": 0.621,
+      "step": 2976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1746714028690287e-06,
+      "loss": 0.5422,
+      "step": 2977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1703313396687521e-06,
+      "loss": 0.3729,
+      "step": 2978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1659988105594022e-06,
+      "loss": 0.2218,
+      "step": 2979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1616738192377963e-06,
+      "loss": 0.152,
+      "step": 2980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1573563693943202e-06,
+      "loss": 0.6722,
+      "step": 2981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1530464647129235e-06,
+      "loss": 0.809,
+      "step": 2982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1487441088711194e-06,
+      "loss": 0.091,
+      "step": 2983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1444493055399774e-06,
+      "loss": 0.7194,
+      "step": 2984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1401620583841255e-06,
+      "loss": 0.3971,
+      "step": 2985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1358823710617395e-06,
+      "loss": 0.2721,
+      "step": 2986
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.131610247224555e-06,
+      "loss": 0.0778,
+      "step": 2987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1273456905178392e-06,
+      "loss": 1.0566,
+      "step": 2988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1230887045804151e-06,
+      "loss": 0.4217,
+      "step": 2989
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1188392930446368e-06,
+      "loss": 0.8682,
+      "step": 2990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.114597459536404e-06,
+      "loss": 0.1709,
+      "step": 2991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1103632076751459e-06,
+      "loss": 0.4549,
+      "step": 2992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1061365410738168e-06,
+      "loss": 0.2418,
+      "step": 2993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1019174633389073e-06,
+      "loss": 0.5846,
+      "step": 2994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0977059780704314e-06,
+      "loss": 0.9675,
+      "step": 2995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0935020888619218e-06,
+      "loss": 0.3654,
+      "step": 2996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0893057993004297e-06,
+      "loss": 0.5139,
+      "step": 2997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.085117112966525e-06,
+      "loss": 0.7085,
+      "step": 2998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0809360334342855e-06,
+      "loss": 0.6113,
+      "step": 2999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.076762564271302e-06,
+      "loss": 0.4454,
+      "step": 3000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0725967090386702e-06,
+      "loss": 0.3418,
+      "step": 3001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.068438471290988e-06,
+      "loss": 0.4498,
+      "step": 3002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.064287854576359e-06,
+      "loss": 0.3492,
+      "step": 3003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0601448624363752e-06,
+      "loss": 0.5284,
+      "step": 3004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0560094984061276e-06,
+      "loss": 0.4955,
+      "step": 3005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0518817660141977e-06,
+      "loss": 0.3498,
+      "step": 3006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0477616687826597e-06,
+      "loss": 0.4906,
+      "step": 3007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0436492102270646e-06,
+      "loss": 0.4547,
+      "step": 3008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0395443938564542e-06,
+      "loss": 0.2597,
+      "step": 3009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.035447223173337e-06,
+      "loss": 0.3177,
+      "step": 3010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.031357701673713e-06,
+      "loss": 1.0358,
+      "step": 3011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0272758328470445e-06,
+      "loss": 0.3802,
+      "step": 3012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0232016201762696e-06,
+      "loss": 0.3371,
+      "step": 3013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0191350671377898e-06,
+      "loss": 0.3837,
+      "step": 3014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0150761772014739e-06,
+      "loss": 0.2609,
+      "step": 3015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0110249538306493e-06,
+      "loss": 1.0096,
+      "step": 3016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0069814004821033e-06,
+      "loss": 0.5259,
+      "step": 3017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0029455206060778e-06,
+      "loss": 0.3823,
+      "step": 3018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.989173176462708e-07,
+      "loss": 0.2672,
+      "step": 3019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.94896795039827e-07,
+      "loss": 0.6352,
+      "step": 3020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.908839562173344e-07,
+      "loss": 0.3759,
+      "step": 3021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.868788046028266e-07,
+      "loss": 0.4234,
+      "step": 3022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828813436137829e-07,
+      "loss": 0.1006,
+      "step": 3023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.788915766611151e-07,
+      "loss": 0.1852,
+      "step": 3024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.749095071491744e-07,
+      "loss": 0.0946,
+      "step": 3025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.709351384757338e-07,
+      "loss": 0.5805,
+      "step": 3026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.669684740320096e-07,
+      "loss": 0.1263,
+      "step": 3027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.630095172026345e-07,
+      "loss": 0.6096,
+      "step": 3028
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59058271365667e-07,
+      "loss": 0.4793,
+      "step": 3029
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.551147398925853e-07,
+      "loss": 0.193,
+      "step": 3030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.511789261482929e-07,
+      "loss": 0.986,
+      "step": 3031
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.472508334910946e-07,
+      "loss": 0.7166,
+      "step": 3032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.433304652727149e-07,
+      "loss": 0.5168,
+      "step": 3033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.394178248382868e-07,
+      "loss": 0.5399,
+      "step": 3034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.355129155263498e-07,
+      "loss": 0.7725,
+      "step": 3035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.316157406688475e-07,
+      "loss": 0.1617,
+      "step": 3036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.277263035911177e-07,
+      "loss": 0.9066,
+      "step": 3037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.238446076119001e-07,
+      "loss": 0.6502,
+      "step": 3038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.19970656043333e-07,
+      "loss": 0.2981,
+      "step": 3039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.161044521909412e-07,
+      "loss": 0.4119,
+      "step": 3040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.122459993536392e-07,
+      "loss": 0.2653,
+      "step": 3041
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.083953008237311e-07,
+      "loss": 0.3631,
+      "step": 3042
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.045523598869011e-07,
+      "loss": 0.3217,
+      "step": 3043
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.007171798222136e-07,
+      "loss": 0.6791,
+      "step": 3044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.968897639021157e-07,
+      "loss": 0.9069,
+      "step": 3045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.930701153924215e-07,
+      "loss": 0.5389,
+      "step": 3046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.892582375523296e-07,
+      "loss": 0.1555,
+      "step": 3047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.854541336343947e-07,
+      "loss": 0.1171,
+      "step": 3048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.816578068845472e-07,
+      "loss": 0.8605,
+      "step": 3049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.778692605420747e-07,
+      "loss": 0.4138,
+      "step": 3050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.740884978396358e-07,
+      "loss": 0.2557,
+      "step": 3051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.703155220032378e-07,
+      "loss": 0.1487,
+      "step": 3052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.665503362522509e-07,
+      "loss": 0.3668,
+      "step": 3053
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.627929437993898e-07,
+      "loss": 0.4541,
+      "step": 3054
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.590433478507287e-07,
+      "loss": 0.6625,
+      "step": 3055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.553015516056839e-07,
+      "loss": 0.8121,
+      "step": 3056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.515675582570181e-07,
+      "loss": 0.4635,
+      "step": 3057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.478413709908351e-07,
+      "loss": 0.7067,
+      "step": 3058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.44122992986578e-07,
+      "loss": 0.4707,
+      "step": 3059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.404124274170278e-07,
+      "loss": 0.363,
+      "step": 3060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.367096774482996e-07,
+      "loss": 0.7477,
+      "step": 3061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.330147462398353e-07,
+      "loss": 0.74,
+      "step": 3062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.293276369444114e-07,
+      "loss": 0.2897,
+      "step": 3063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.256483527081305e-07,
+      "loss": 0.3499,
+      "step": 3064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.21976896670409e-07,
+      "loss": 0.3529,
+      "step": 3065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.183132719639908e-07,
+      "loss": 0.4447,
+      "step": 3066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.146574817149411e-07,
+      "loss": 0.5645,
+      "step": 3067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.110095290426334e-07,
+      "loss": 0.4013,
+      "step": 3068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.073694170597579e-07,
+      "loss": 0.248,
+      "step": 3069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.037371488723078e-07,
+      "loss": 0.3818,
+      "step": 3070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.001127275795928e-07,
+      "loss": 0.685,
+      "step": 3071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.964961562742212e-07,
+      "loss": 0.4431,
+      "step": 3072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.928874380421059e-07,
+      "loss": 1.1852,
+      "step": 3073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.892865759624569e-07,
+      "loss": 0.206,
+      "step": 3074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.856935731077808e-07,
+      "loss": 0.1192,
+      "step": 3075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.821084325438788e-07,
+      "loss": 0.4057,
+      "step": 3076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.785311573298459e-07,
+      "loss": 0.2351,
+      "step": 3077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.749617505180596e-07,
+      "loss": 0.2316,
+      "step": 3078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.714002151541911e-07,
+      "loss": 0.2792,
+      "step": 3079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.678465542771929e-07,
+      "loss": 0.5803,
+      "step": 3080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.643007709192918e-07,
+      "loss": 0.4693,
+      "step": 3081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.607628681059998e-07,
+      "loss": 0.2162,
+      "step": 3082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.572328488561064e-07,
+      "loss": 0.6214,
+      "step": 3083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.537107161816681e-07,
+      "loss": 0.4785,
+      "step": 3084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.501964730880151e-07,
+      "loss": 0.3521,
+      "step": 3085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.466901225737455e-07,
+      "loss": 0.7873,
+      "step": 3086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.431916676307238e-07,
+      "loss": 0.6029,
+      "step": 3087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.397011112440744e-07,
+      "loss": 0.3034,
+      "step": 3088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.36218456392187e-07,
+      "loss": 0.5407,
+      "step": 3089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.327437060467047e-07,
+      "loss": 0.633,
+      "step": 3090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.292768631725266e-07,
+      "loss": 0.2982,
+      "step": 3091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.258179307278068e-07,
+      "loss": 0.2559,
+      "step": 3092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.223669116639487e-07,
+      "loss": 0.3704,
+      "step": 3093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.189238089256034e-07,
+      "loss": 0.7993,
+      "step": 3094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.154886254506632e-07,
+      "loss": 0.3126,
+      "step": 3095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.120613641702723e-07,
+      "loss": 0.4239,
+      "step": 3096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.086420280088091e-07,
+      "loss": 0.3201,
+      "step": 3097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.052306198838854e-07,
+      "loss": 0.5193,
+      "step": 3098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.018271427063583e-07,
+      "loss": 0.4905,
+      "step": 3099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.984315993803104e-07,
+      "loss": 1.0417,
+      "step": 3100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.950439928030583e-07,
+      "loss": 0.546,
+      "step": 3101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.916643258651434e-07,
+      "loss": 0.1284,
+      "step": 3102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.882926014503344e-07,
+      "loss": 0.3676,
+      "step": 3103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.849288224356221e-07,
+      "loss": 0.4688,
+      "step": 3104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.815729916912184e-07,
+      "loss": 0.1862,
+      "step": 3105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.782251120805528e-07,
+      "loss": 0.8067,
+      "step": 3106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.748851864602691e-07,
+      "loss": 0.1092,
+      "step": 3107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.715532176802298e-07,
+      "loss": 0.4329,
+      "step": 3108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.682292085834985e-07,
+      "loss": 0.2216,
+      "step": 3109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.649131620063554e-07,
+      "loss": 0.1725,
+      "step": 3110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.616050807782803e-07,
+      "loss": 0.3414,
+      "step": 3111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.583049677219633e-07,
+      "loss": 0.3964,
+      "step": 3112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.550128256532906e-07,
+      "loss": 0.0718,
+      "step": 3113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.517286573813453e-07,
+      "loss": 0.3956,
+      "step": 3114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.484524657084134e-07,
+      "loss": 0.905,
+      "step": 3115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.45184253429969e-07,
+      "loss": 0.282,
+      "step": 3116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.419240233346801e-07,
+      "loss": 0.5808,
+      "step": 3117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.386717782044016e-07,
+      "loss": 0.4393,
+      "step": 3118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.35427520814178e-07,
+      "loss": 0.2507,
+      "step": 3119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.32191253932235e-07,
+      "loss": 0.0819,
+      "step": 3120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.289629803199837e-07,
+      "loss": 0.231,
+      "step": 3121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.257427027320129e-07,
+      "loss": 0.2356,
+      "step": 3122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.225304239160856e-07,
+      "loss": 0.5866,
+      "step": 3123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.193261466131484e-07,
+      "loss": 0.6016,
+      "step": 3124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.161298735573107e-07,
+      "loss": 0.4389,
+      "step": 3125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.129416074758565e-07,
+      "loss": 0.9209,
+      "step": 3126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.097613510892364e-07,
+      "loss": 0.3433,
+      "step": 3127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.065891071110708e-07,
+      "loss": 0.8736,
+      "step": 3128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.034248782481389e-07,
+      "loss": 0.3451,
+      "step": 3129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.002686672003821e-07,
+      "loss": 0.6143,
+      "step": 3130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.971204766609007e-07,
+      "loss": 0.3584,
+      "step": 3131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.939803093159502e-07,
+      "loss": 0.8005,
+      "step": 3132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.908481678449407e-07,
+      "loss": 0.4243,
+      "step": 3133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.877240549204355e-07,
+      "loss": 0.679,
+      "step": 3134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.846079732081455e-07,
+      "loss": 0.433,
+      "step": 3135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.814999253669307e-07,
+      "loss": 0.4159,
+      "step": 3136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.783999140487939e-07,
+      "loss": 0.7405,
+      "step": 3137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.753079418988817e-07,
+      "loss": 0.1599,
+      "step": 3138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7222401155548e-07,
+      "loss": 0.4637,
+      "step": 3139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.691481256500164e-07,
+      "loss": 1.0772,
+      "step": 3140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.660802868070525e-07,
+      "loss": 0.4674,
+      "step": 3141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.630204976442787e-07,
+      "loss": 0.1126,
+      "step": 3142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.599687607725235e-07,
+      "loss": 0.4014,
+      "step": 3143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.569250787957425e-07,
+      "loss": 1.0177,
+      "step": 3144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.538894543110185e-07,
+      "loss": 0.5509,
+      "step": 3145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.508618899085583e-07,
+      "loss": 0.3254,
+      "step": 3146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.478423881716899e-07,
+      "loss": 0.3184,
+      "step": 3147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.448309516768657e-07,
+      "loss": 0.1536,
+      "step": 3148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.418275829936537e-07,
+      "loss": 0.6397,
+      "step": 3149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.388322846847371e-07,
+      "loss": 0.5167,
+      "step": 3150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.358450593059128e-07,
+      "loss": 0.3594,
+      "step": 3151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.32865909406095e-07,
+      "loss": 0.8182,
+      "step": 3152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.298948375272984e-07,
+      "loss": 0.3073,
+      "step": 3153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.269318462046502e-07,
+      "loss": 0.2284,
+      "step": 3154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.239769379663818e-07,
+      "loss": 0.1235,
+      "step": 3155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.210301153338293e-07,
+      "loss": 0.9044,
+      "step": 3156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.180913808214283e-07,
+      "loss": 0.1177,
+      "step": 3157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.151607369367095e-07,
+      "loss": 0.0957,
+      "step": 3158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.122381861803039e-07,
+      "loss": 0.1114,
+      "step": 3159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.093237310459387e-07,
+      "loss": 0.4629,
+      "step": 3160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.064173740204292e-07,
+      "loss": 0.6846,
+      "step": 3161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.035191175836829e-07,
+      "loss": 0.2305,
+      "step": 3162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.006289642086948e-07,
+      "loss": 0.4344,
+      "step": 3163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.977469163615456e-07,
+      "loss": 0.7026,
+      "step": 3164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.948729765014004e-07,
+      "loss": 0.5258,
+      "step": 3165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.920071470805055e-07,
+      "loss": 0.3031,
+      "step": 3166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.891494305441869e-07,
+      "loss": 0.2518,
+      "step": 3167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.862998293308485e-07,
+      "loss": 0.2708,
+      "step": 3168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.834583458719721e-07,
+      "loss": 0.5062,
+      "step": 3169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.806249825921061e-07,
+      "loss": 0.4113,
+      "step": 3170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.777997419088731e-07,
+      "loss": 0.3584,
+      "step": 3171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.749826262329715e-07,
+      "loss": 0.0763,
+      "step": 3172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.721736379681574e-07,
+      "loss": 0.9065,
+      "step": 3173
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.69372779511259e-07,
+      "loss": 0.4688,
+      "step": 3174
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6658005325216136e-07,
+      "loss": 1.2444,
+      "step": 3175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6379546157381496e-07,
+      "loss": 0.0605,
+      "step": 3176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.610190068522302e-07,
+      "loss": 0.2868,
+      "step": 3177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5825069145646996e-07,
+      "loss": 0.8566,
+      "step": 3178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.554905177486557e-07,
+      "loss": 0.2537,
+      "step": 3179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5273848808396027e-07,
+      "loss": 0.5242,
+      "step": 3180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.499946048106085e-07,
+      "loss": 0.6213,
+      "step": 3181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4725887026987325e-07,
+      "loss": 0.816,
+      "step": 3182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.445312867960727e-07,
+      "loss": 0.4172,
+      "step": 3183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4181185671657634e-07,
+      "loss": 0.3887,
+      "step": 3184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.391005823517891e-07,
+      "loss": 0.3569,
+      "step": 3185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3639746601516044e-07,
+      "loss": 0.3868,
+      "step": 3186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.337025100131764e-07,
+      "loss": 0.8333,
+      "step": 3187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3101571664536433e-07,
+      "loss": 1.1801,
+      "step": 3188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2833708820428366e-07,
+      "loss": 0.5384,
+      "step": 3189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.256666269755283e-07,
+      "loss": 0.3206,
+      "step": 3190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.230043352377222e-07,
+      "loss": 0.1815,
+      "step": 3191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.203502152625172e-07,
+      "loss": 0.1703,
+      "step": 3192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1770426931459605e-07,
+      "loss": 0.5808,
+      "step": 3193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1506649965166403e-07,
+      "loss": 0.4924,
+      "step": 3194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1243690852445174e-07,
+      "loss": 0.1271,
+      "step": 3195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0981549817670883e-07,
+      "loss": 0.4351,
+      "step": 3196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0720227084520613e-07,
+      "loss": 0.5773,
+      "step": 3197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.045972287597333e-07,
+      "loss": 0.0412,
+      "step": 3198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0200037414309225e-07,
+      "loss": 0.5143,
+      "step": 3199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9941170921110386e-07,
+      "loss": 0.4758,
+      "step": 3200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.968312361725968e-07,
+      "loss": 0.5903,
+      "step": 3201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.94258957229412e-07,
+      "loss": 0.2327,
+      "step": 3202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.916948745763938e-07,
+      "loss": 0.3603,
+      "step": 3203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.89138990401402e-07,
+      "loss": 0.3749,
+      "step": 3204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.865913068852933e-07,
+      "loss": 0.4324,
+      "step": 3205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.840518262019299e-07,
+      "loss": 0.5024,
+      "step": 3206
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.815205505181741e-07,
+      "loss": 0.3577,
+      "step": 3207
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.789974819938869e-07,
+      "loss": 0.2331,
+      "step": 3208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.764826227819285e-07,
+      "loss": 0.3507,
+      "step": 3209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7397597502815133e-07,
+      "loss": 0.4767,
+      "step": 3210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.714775408714033e-07,
+      "loss": 0.5395,
+      "step": 3211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6898732244352143e-07,
+      "loss": 0.3307,
+      "step": 3212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6650532186933817e-07,
+      "loss": 0.3552,
+      "step": 3213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.640315412666662e-07,
+      "loss": 0.2963,
+      "step": 3214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6156598274630915e-07,
+      "loss": 0.5494,
+      "step": 3215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.591086484120543e-07,
+      "loss": 0.4979,
+      "step": 3216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5665954036067207e-07,
+      "loss": 0.3025,
+      "step": 3217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5421866068191315e-07,
+      "loss": 0.1178,
+      "step": 3218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.517860114585037e-07,
+      "loss": 0.4353,
+      "step": 3219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4936159476615216e-07,
+      "loss": 0.274,
+      "step": 3220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4694541267354165e-07,
+      "loss": 0.6204,
+      "step": 3221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.445374672423252e-07,
+      "loss": 0.6575,
+      "step": 3222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.421377605271325e-07,
+      "loss": 0.215,
+      "step": 3223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.397462945755603e-07,
+      "loss": 1.1108,
+      "step": 3224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.373630714281739e-07,
+      "loss": 0.5021,
+      "step": 3225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3498809311850677e-07,
+      "loss": 0.2015,
+      "step": 3226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.326213616730578e-07,
+      "loss": 0.1329,
+      "step": 3227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3026287911128383e-07,
+      "loss": 0.1278,
+      "step": 3228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.279126474456129e-07,
+      "loss": 0.1643,
+      "step": 3229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.25570668681422e-07,
+      "loss": 0.72,
+      "step": 3230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.232369448170525e-07,
+      "loss": 0.1316,
+      "step": 3231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.209114778438027e-07,
+      "loss": 0.2526,
+      "step": 3232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1859426974592323e-07,
+      "loss": 0.3257,
+      "step": 3233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.162853225006168e-07,
+      "loss": 0.2931,
+      "step": 3234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.139846380780387e-07,
+      "loss": 0.2548,
+      "step": 3235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1169221844129517e-07,
+      "loss": 0.8259,
+      "step": 3236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.094080655464382e-07,
+      "loss": 0.7719,
+      "step": 3237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.071321813424666e-07,
+      "loss": 0.1641,
+      "step": 3238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0486456777132465e-07,
+      "loss": 0.4577,
+      "step": 3239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.026052267678981e-07,
+      "loss": 0.2927,
+      "step": 3240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0035416026001573e-07,
+      "loss": 0.7574,
+      "step": 3241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9811137016844347e-07,
+      "loss": 0.3939,
+      "step": 3242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9587685840688716e-07,
+      "loss": 0.4283,
+      "step": 3243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.936506268819894e-07,
+      "loss": 0.7296,
+      "step": 3244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9143267749332626e-07,
+      "loss": 0.3296,
+      "step": 3245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.892230121334083e-07,
+      "loss": 0.7506,
+      "step": 3246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8702163268767294e-07,
+      "loss": 0.5201,
+      "step": 3247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.848285410344953e-07,
+      "loss": 0.4586,
+      "step": 3248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8264373904517307e-07,
+      "loss": 0.1796,
+      "step": 3249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.804672285839316e-07,
+      "loss": 0.264,
+      "step": 3250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7829901150792205e-07,
+      "loss": 0.5836,
+      "step": 3251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7613908966722004e-07,
+      "loss": 0.2848,
+      "step": 3252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.739874649048202e-07,
+      "loss": 0.483,
+      "step": 3253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7184413905664063e-07,
+      "loss": 0.2777,
+      "step": 3254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.697091139515151e-07,
+      "loss": 0.2268,
+      "step": 3255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6758239141119745e-07,
+      "loss": 0.3943,
+      "step": 3256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6546397325035833e-07,
+      "loss": 0.3703,
+      "step": 3257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6335386127657734e-07,
+      "loss": 0.6099,
+      "step": 3258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6125205729035097e-07,
+      "loss": 0.3529,
+      "step": 3259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.591585630850835e-07,
+      "loss": 0.2289,
+      "step": 3260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.570733804470926e-07,
+      "loss": 0.4459,
+      "step": 3261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5499651115560296e-07,
+      "loss": 0.2682,
+      "step": 3262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.529279569827414e-07,
+      "loss": 0.9663,
+      "step": 3263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5086771969354497e-07,
+      "loss": 0.1486,
+      "step": 3264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4881580104595296e-07,
+      "loss": 0.1055,
+      "step": 3265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.467722027908048e-07,
+      "loss": 0.2701,
+      "step": 3266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4473692667184136e-07,
+      "loss": 0.4828,
+      "step": 3267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4270997442570335e-07,
+      "loss": 0.4001,
+      "step": 3268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.406913477819273e-07,
+      "loss": 0.2183,
+      "step": 3269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.386810484629476e-07,
+      "loss": 0.6708,
+      "step": 3270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3667907818409109e-07,
+      "loss": 0.0859,
+      "step": 3271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3468543865358017e-07,
+      "loss": 1.0439,
+      "step": 3272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3270013157252747e-07,
+      "loss": 0.6171,
+      "step": 3273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3072315863493456e-07,
+      "loss": 0.2367,
+      "step": 3274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.287545215276943e-07,
+      "loss": 0.2085,
+      "step": 3275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2679422193058297e-07,
+      "loss": 0.624,
+      "step": 3276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2484226151626932e-07,
+      "loss": 0.5026,
+      "step": 3277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2289864195030097e-07,
+      "loss": 0.2918,
+      "step": 3278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2096336489111025e-07,
+      "loss": 0.1347,
+      "step": 3279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1903643199000846e-07,
+      "loss": 0.3107,
+      "step": 3280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1711784489119146e-07,
+      "loss": 0.6183,
+      "step": 3281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1520760523173313e-07,
+      "loss": 0.357,
+      "step": 3282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.133057146415829e-07,
+      "loss": 1.0704,
+      "step": 3283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.114121747435649e-07,
+      "loss": 0.5997,
+      "step": 3284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0952698715338226e-07,
+      "loss": 0.8914,
+      "step": 3285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0765015347960716e-07,
+      "loss": 0.8981,
+      "step": 3286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0578167532368742e-07,
+      "loss": 0.3142,
+      "step": 3287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0392155427993554e-07,
+      "loss": 0.4018,
+      "step": 3288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0206979193554187e-07,
+      "loss": 0.6319,
+      "step": 3289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0022638987055698e-07,
+      "loss": 0.5867,
+      "step": 3290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.983913496578993e-07,
+      "loss": 0.4264,
+      "step": 3291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9656467286335523e-07,
+      "loss": 0.6637,
+      "step": 3292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9474636104557244e-07,
+      "loss": 0.201,
+      "step": 3293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9293641575606203e-07,
+      "loss": 0.4301,
+      "step": 3294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9113483853919756e-07,
+      "loss": 0.6203,
+      "step": 3295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8934163093220715e-07,
+      "loss": 0.443,
+      "step": 3296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.875567944651835e-07,
+      "loss": 0.4433,
+      "step": 3297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8578033066107392e-07,
+      "loss": 0.153,
+      "step": 3298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8401224103568038e-07,
+      "loss": 0.1347,
+      "step": 3299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.822525270976605e-07,
+      "loss": 1.0795,
+      "step": 3300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8050119034852765e-07,
+      "loss": 0.6878,
+      "step": 3301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.787582322826431e-07,
+      "loss": 0.4432,
+      "step": 3302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7702365438722058e-07,
+      "loss": 0.346,
+      "step": 3303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7529745814232168e-07,
+      "loss": 0.6478,
+      "step": 3304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7357964502086155e-07,
+      "loss": 0.304,
+      "step": 3305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.718702164885966e-07,
+      "loss": 0.2115,
+      "step": 3306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7016917400413002e-07,
+      "loss": 0.4776,
+      "step": 3307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6847651901891081e-07,
+      "loss": 0.1824,
+      "step": 3308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6679225297723146e-07,
+      "loss": 0.2061,
+      "step": 3309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6511637731622453e-07,
+      "loss": 0.0909,
+      "step": 3310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6344889346586402e-07,
+      "loss": 0.4458,
+      "step": 3311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6178980284896507e-07,
+      "loss": 0.2238,
+      "step": 3312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6013910688117972e-07,
+      "loss": 0.2882,
+      "step": 3313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.584968069709958e-07,
+      "loss": 0.7846,
+      "step": 3314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5686290451974007e-07,
+      "loss": 0.2641,
+      "step": 3315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5523740092157068e-07,
+      "loss": 0.9445,
+      "step": 3316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5362029756348373e-07,
+      "loss": 0.641,
+      "step": 3317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5201159582530323e-07,
+      "loss": 0.4782,
+      "step": 3318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.504112970796856e-07,
+      "loss": 0.3934,
+      "step": 3319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4881940269211637e-07,
+      "loss": 0.6534,
+      "step": 3320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4723591402091453e-07,
+      "loss": 0.2397,
+      "step": 3321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4566083241722262e-07,
+      "loss": 0.8282,
+      "step": 3322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4409415922500892e-07,
+      "loss": 0.2352,
+      "step": 3323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4253589578106853e-07,
+      "loss": 0.2955,
+      "step": 3324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.409860434150223e-07,
+      "loss": 0.3577,
+      "step": 3325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3944460344931133e-07,
+      "loss": 0.5081,
+      "step": 3326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3791157719920124e-07,
+      "loss": 0.4509,
+      "step": 3327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3638696597277678e-07,
+      "loss": 0.2776,
+      "step": 3328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3487077107094182e-07,
+      "loss": 0.57,
+      "step": 3329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3336299378742147e-07,
+      "loss": 0.799,
+      "step": 3330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3186363540875658e-07,
+      "loss": 0.2123,
+      "step": 3331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3037269721430268e-07,
+      "loss": 0.3061,
+      "step": 3332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2889018047623546e-07,
+      "loss": 0.3454,
+      "step": 3333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2741608645954084e-07,
+      "loss": 0.1991,
+      "step": 3334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2595041642201822e-07,
+      "loss": 0.9226,
+      "step": 3335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2449317161427942e-07,
+      "loss": 0.1231,
+      "step": 3336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2304435327974873e-07,
+      "loss": 0.1086,
+      "step": 3337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2160396265465835e-07,
+      "loss": 0.6008,
+      "step": 3338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2017200096805294e-07,
+      "loss": 0.4057,
+      "step": 3339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1874846944177732e-07,
+      "loss": 0.3772,
+      "step": 3340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1733336929049322e-07,
+      "loss": 0.5391,
+      "step": 3341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1592670172166032e-07,
+      "loss": 0.4972,
+      "step": 3342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1452846793554739e-07,
+      "loss": 0.3332,
+      "step": 3343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1313866912522343e-07,
+      "loss": 0.1267,
+      "step": 3344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1175730647656313e-07,
+      "loss": 0.5742,
+      "step": 3345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1038438116824258e-07,
+      "loss": 0.3348,
+      "step": 3346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0901989437173577e-07,
+      "loss": 0.6279,
+      "step": 3347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0766384725131807e-07,
+      "loss": 0.617,
+      "step": 3348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0631624096406612e-07,
+      "loss": 0.4161,
+      "step": 3349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0497707665985235e-07,
+      "loss": 0.568,
+      "step": 3350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.036463554813416e-07,
+      "loss": 0.5617,
+      "step": 3351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0232407856400007e-07,
+      "loss": 0.6874,
+      "step": 3352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0101024703608741e-07,
+      "loss": 0.2118,
+      "step": 3353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.970486201865693e-08,
+      "loss": 0.3295,
+      "step": 3354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.840792462555426e-08,
+      "loss": 0.7405,
+      "step": 3355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.711943596341644e-08,
+      "loss": 0.4235,
+      "step": 3356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.583939713167179e-08,
+      "loss": 0.4252,
+      "step": 3357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.456780922253995e-08,
+      "loss": 0.3624,
+      "step": 3358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.330467332102855e-08,
+      "loss": 0.9345,
+      "step": 3359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.204999050493213e-08,
+      "loss": 0.1537,
+      "step": 3360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.080376184483653e-08,
+      "loss": 0.5072,
+      "step": 3361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.95659884041089e-08,
+      "loss": 0.4435,
+      "step": 3362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.833667123890444e-08,
+      "loss": 0.6293,
+      "step": 3363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.711581139816294e-08,
+      "loss": 0.6438,
+      "step": 3364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.59034099236078e-08,
+      "loss": 0.8262,
+      "step": 3365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.469946784974481e-08,
+      "loss": 0.124,
+      "step": 3366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.350398620386113e-08,
+      "loss": 0.253,
+      "step": 3367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.231696600602523e-08,
+      "loss": 0.5502,
+      "step": 3368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.113840826908582e-08,
+      "loss": 0.3486,
+      "step": 3369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.996831399867067e-08,
+      "loss": 0.1985,
+      "step": 3370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.88066841931856e-08,
+      "loss": 0.1503,
+      "step": 3371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.765351984381663e-08,
+      "loss": 0.3757,
+      "step": 3372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.650882193452114e-08,
+      "loss": 0.4162,
+      "step": 3373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.53725914420378e-08,
+      "loss": 0.1726,
+      "step": 3374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.424482933587774e-08,
+      "loss": 0.7147,
+      "step": 3375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.312553657832567e-08,
+      "loss": 0.2259,
+      "step": 3376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.201471412443983e-08,
+      "loss": 0.4781,
+      "step": 3377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.091236292205317e-08,
+      "loss": 0.6066,
+      "step": 3378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.981848391176771e-08,
+      "loss": 0.3425,
+      "step": 3379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.873307802695795e-08,
+      "loss": 0.1342,
+      "step": 3380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.765614619376859e-08,
+      "loss": 0.7865,
+      "step": 3381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.658768933111238e-08,
+      "loss": 0.3902,
+      "step": 3382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.552770835067224e-08,
+      "loss": 0.6891,
+      "step": 3383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.447620415689693e-08,
+      "loss": 0.3139,
+      "step": 3384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.34331776470054e-08,
+      "loss": 0.6291,
+      "step": 3385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.239862971097909e-08,
+      "loss": 0.1256,
+      "step": 3386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.137256123156631e-08,
+      "loss": 0.1212,
+      "step": 3387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.035497308428229e-08,
+      "loss": 0.6287,
+      "step": 3388
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.934586613740245e-08,
+      "loss": 0.7758,
+      "step": 3389
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8345241251969165e-08,
+      "loss": 0.6998,
+      "step": 3390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7353099281785004e-08,
+      "loss": 0.5153,
+      "step": 3391
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.636944107341391e-08,
+      "loss": 0.4989,
+      "step": 3392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.539426746618337e-08,
+      "loss": 0.6204,
+      "step": 3393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.442757929217779e-08,
+      "loss": 1.072,
+      "step": 3394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.346937737624624e-08,
+      "loss": 0.2936,
+      "step": 3395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.251966253599028e-08,
+      "loss": 0.2182,
+      "step": 3396
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1578435581775e-08,
+      "loss": 0.3352,
+      "step": 3397
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.06456973167202e-08,
+      "loss": 0.1389,
+      "step": 3398
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.972144853670369e-08,
+      "loss": 0.9909,
+      "step": 3399
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8805690030360176e-08,
+      "loss": 0.3091,
+      "step": 3400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.789842257907795e-08,
+      "loss": 0.4326,
+      "step": 3401
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.699964695699999e-08,
+      "loss": 0.458,
+      "step": 3402
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.610936393102616e-08,
+      "loss": 0.2077,
+      "step": 3403
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.522757426080771e-08,
+      "loss": 0.4628,
+      "step": 3404
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.435427869874942e-08,
+      "loss": 0.6317,
+      "step": 3405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3489477990007466e-08,
+      "loss": 0.6267,
+      "step": 3406
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.263317287249158e-08,
+      "loss": 0.7146,
+      "step": 3407
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1785364076859515e-08,
+      "loss": 0.403,
+      "step": 3408
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0946052326522603e-08,
+      "loss": 0.4473,
+      "step": 3409
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.011523833763909e-08,
+      "loss": 0.5415,
+      "step": 3410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.929292281911856e-08,
+      "loss": 0.6456,
+      "step": 3411
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.847910647261754e-08,
+      "loss": 0.2464,
+      "step": 3412
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.767378999254168e-08,
+      "loss": 0.5539,
+      "step": 3413
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.68769740660424e-08,
+      "loss": 0.1042,
+      "step": 3414
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6088659373019195e-08,
+      "loss": 0.2637,
+      "step": 3415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.530884658611733e-08,
+      "loss": 0.7727,
+      "step": 3416
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.453753637072788e-08,
+      "loss": 0.1596,
+      "step": 3417
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3774729384986605e-08,
+      "loss": 0.062,
+      "step": 3418
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3020426279773974e-08,
+      "loss": 0.606,
+      "step": 3419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.227462769871404e-08,
+      "loss": 0.5132,
+      "step": 3420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.153733427817329e-08,
+      "loss": 0.3075,
+      "step": 3421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.080854664726296e-08,
+      "loss": 0.8488,
+      "step": 3422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.008826542783561e-08,
+      "loss": 0.5545,
+      "step": 3423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.937649123448627e-08,
+      "loss": 0.6243,
+      "step": 3424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8673224674548028e-08,
+      "loss": 0.1504,
+      "step": 3425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7978466348100863e-08,
+      "loss": 0.6896,
+      "step": 3426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7292216847957242e-08,
+      "loss": 0.397,
+      "step": 3427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6614476759676546e-08,
+      "loss": 0.1611,
+      "step": 3428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5945246661551738e-08,
+      "loss": 0.378,
+      "step": 3429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5284527124618262e-08,
+      "loss": 1.1266,
+      "step": 3430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4632318712646264e-08,
+      "loss": 0.142,
+      "step": 3431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3988621982148353e-08,
+      "loss": 0.2314,
+      "step": 3432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3353437482369624e-08,
+      "loss": 0.6081,
+      "step": 3433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.272676575529431e-08,
+      "loss": 0.5439,
+      "step": 3434
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2108607335642463e-08,
+      "loss": 0.148,
+      "step": 3435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1498962750869933e-08,
+      "loss": 0.4668,
+      "step": 3436
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0897832521169505e-08,
+      "loss": 0.598,
+      "step": 3437
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0305217159466428e-08,
+      "loss": 0.4789,
+      "step": 3438
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9721117171420668e-08,
+      "loss": 0.8924,
+      "step": 3439
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.91455330554291e-08,
+      "loss": 0.3555,
+      "step": 3440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8578465302618864e-08,
+      "loss": 0.2468,
+      "step": 3441
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.801991439685291e-08,
+      "loss": 0.1638,
+      "step": 3442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.746988081472556e-08,
+      "loss": 0.4189,
+      "step": 3443
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.692836502556472e-08,
+      "loss": 0.8092,
+      "step": 3444
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.639536749142745e-08,
+      "loss": 0.5936,
+      "step": 3445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.587088866710551e-08,
+      "loss": 0.6812,
+      "step": 3446
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.535492900012203e-08,
+      "loss": 0.1604,
+      "step": 3447
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4847488930728182e-08,
+      "loss": 0.3876,
+      "step": 3448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4348568891908721e-08,
+      "loss": 0.4053,
+      "step": 3449
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3858169309376446e-08,
+      "loss": 0.191,
+      "step": 3450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3376290601574416e-08,
+      "loss": 0.388,
+      "step": 3451
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2902933179675947e-08,
+      "loss": 0.2158,
+      "step": 3452
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2438097447581288e-08,
+      "loss": 0.83,
+      "step": 3453
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1981783801923163e-08,
+      "loss": 0.2578,
+      "step": 3454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.153399263205901e-08,
+      "loss": 0.2417,
+      "step": 3455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1094724320074301e-08,
+      "loss": 0.1445,
+      "step": 3456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0663979240784772e-08,
+      "loss": 0.3383,
+      "step": 3457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0241757761733084e-08,
+      "loss": 0.435,
+      "step": 3458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.82806024318661e-09,
+      "loss": 0.4017,
+      "step": 3459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.42288703814187e-09,
+      "loss": 0.4539,
+      "step": 3460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.026238492321204e-09,
+      "loss": 0.4565,
+      "step": 3461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.638114944171661e-09,
+      "loss": 0.3836,
+      "step": 3462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.258516724868326e-09,
+      "loss": 0.2089,
+      "step": 3463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.887444158310998e-09,
+      "loss": 0.3926,
+      "step": 3464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.524897561124179e-09,
+      "loss": 0.5146,
+      "step": 3465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.170877242658192e-09,
+      "loss": 0.5855,
+      "step": 3466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.82538350498918e-09,
+      "loss": 0.3306,
+      "step": 3467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.488416642914663e-09,
+      "loss": 0.1595,
+      "step": 3468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.1599769439590896e-09,
+      "loss": 0.1855,
+      "step": 3469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.840064688370506e-09,
+      "loss": 0.2628,
+      "step": 3470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.528680149120557e-09,
+      "loss": 0.2729,
+      "step": 3471
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.225823591903378e-09,
+      "loss": 0.8123,
+      "step": 3472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.93149527513781e-09,
+      "loss": 0.0727,
+      "step": 3473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.645695449965182e-09,
+      "loss": 0.2807,
+      "step": 3474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.368424360251533e-09,
+      "loss": 0.4565,
+      "step": 3475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.099682242580949e-09,
+      "loss": 0.2044,
+      "step": 3476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.839469326265555e-09,
+      "loss": 0.5837,
+      "step": 3477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5877858333366323e-09,
+      "loss": 0.3283,
+      "step": 3478
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3446319785468418e-09,
+      "loss": 0.408,
+      "step": 3479
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1100079693735517e-09,
+      "loss": 0.6993,
+      "step": 3480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.883914006014399e-09,
+      "loss": 0.5955,
+      "step": 3481
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6663502813872865e-09,
+      "loss": 0.3154,
+      "step": 3482
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4573169811337173e-09,
+      "loss": 0.624,
+      "step": 3483
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2568142836154607e-09,
+      "loss": 1.3292,
+      "step": 3484
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0648423599156642e-09,
+      "loss": 0.1611,
+      "step": 3485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8814013738377436e-09,
+      "loss": 0.4124,
+      "step": 3486
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7064914819064914e-09,
+      "loss": 0.1735,
+      "step": 3487
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.540112833366969e-09,
+      "loss": 0.5188,
+      "step": 3488
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3822655701856147e-09,
+      "loss": 0.4373,
+      "step": 3489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2329498270480245e-09,
+      "loss": 1.0257,
+      "step": 3490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0921657313622825e-09,
+      "loss": 0.3341,
+      "step": 3491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.599134032534096e-10,
+      "loss": 0.1077,
+      "step": 3492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.361929555700255e-10,
+      "loss": 0.324,
+      "step": 3493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.210044938776862e-10,
+      "loss": 0.4412,
+      "step": 3494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.14348116464436e-10,
+      "loss": 0.3659,
+      "step": 3495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.162239143352565e-10,
+      "loss": 0.6461,
+      "step": 3496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.266319712187272e-10,
+      "loss": 0.2729,
+      "step": 3497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.455723635592545e-10,
+      "loss": 0.6661,
+      "step": 3498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7304516052373277e-10,
+      "loss": 0.3117,
+      "step": 3499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.090504239959934e-10,
+      "loss": 0.1163,
+      "step": 3500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.535882085823559e-10,
+      "loss": 0.7944,
+      "step": 3501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.066585616071869e-10,
+      "loss": 0.7776,
+      "step": 3502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.826152311290024e-11,
+      "loss": 0.1178,
+      "step": 3503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8397125862177363e-11,
+      "loss": 0.1046,
+      "step": 3504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7065395339077583e-11,
+      "loss": 0.6657,
+      "step": 3505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2663497445971646e-12,
+      "loss": 0.3002,
+      "step": 3506
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.2335,
+      "step": 3507
+    },
+    {
+      "epoch": 1.0,
+      "step": 3507,
+      "total_flos": 0.0,
+      "train_loss": 0.5562806533278309,
+      "train_runtime": 5235.0373,
+      "train_samples_per_second": 5.359,
+      "train_steps_per_second": 0.67
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3507,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}