diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14217 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.997147385103011,
+  "global_step": 2364,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.105263157894737e-07,
+      "loss": 1.0621,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.210526315789474e-07,
+      "loss": 1.0295,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.315789473684211e-07,
+      "loss": 0.9331,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.421052631578948e-07,
+      "loss": 0.8795,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 0.9076,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2631578947368422e-06,
+      "loss": 0.8873,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4736842105263159e-06,
+      "loss": 0.8685,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6842105263157895e-06,
+      "loss": 0.8587,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8947368421052634e-06,
+      "loss": 0.8205,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 0.8425,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3157894736842105e-06,
+      "loss": 0.8441,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5263157894736844e-06,
+      "loss": 0.8245,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7368421052631583e-06,
+      "loss": 0.7838,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9473684210526317e-06,
+      "loss": 0.7856,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.7518,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.368421052631579e-06,
+      "loss": 0.7465,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.578947368421053e-06,
+      "loss": 0.7388,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.789473684210527e-06,
+      "loss": 0.7522,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7351,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.7259,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4210526315789476e-06,
+      "loss": 0.7253,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.631578947368421e-06,
+      "loss": 0.7163,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.842105263157895e-06,
+      "loss": 0.7194,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.052631578947369e-06,
+      "loss": 0.6971,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.6714,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4736842105263165e-06,
+      "loss": 0.679,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.68421052631579e-06,
+      "loss": 0.6815,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8947368421052634e-06,
+      "loss": 0.6795,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.105263157894738e-06,
+      "loss": 0.7268,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.685,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.526315789473685e-06,
+      "loss": 0.6816,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.736842105263158e-06,
+      "loss": 0.6725,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.947368421052632e-06,
+      "loss": 0.6817,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.157894736842106e-06,
+      "loss": 0.6761,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.6677,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.578947368421054e-06,
+      "loss": 0.683,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.789473684210526e-06,
+      "loss": 0.6819,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6671,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.210526315789475e-06,
+      "loss": 0.6682,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.6734,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.631578947368422e-06,
+      "loss": 0.6638,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.842105263157895e-06,
+      "loss": 0.6456,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.05263157894737e-06,
+      "loss": 0.63,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.263157894736842e-06,
+      "loss": 0.6542,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.6795,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.68421052631579e-06,
+      "loss": 0.6599,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.894736842105264e-06,
+      "loss": 0.6422,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0105263157894738e-05,
+      "loss": 0.6659,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0315789473684213e-05,
+      "loss": 0.6456,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.6354,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.073684210526316e-05,
+      "loss": 0.6608,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0947368421052633e-05,
+      "loss": 0.6422,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1157894736842105e-05,
+      "loss": 0.6449,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.136842105263158e-05,
+      "loss": 0.651,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.6416,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1789473684210527e-05,
+      "loss": 0.594,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6253,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2210526315789475e-05,
+      "loss": 0.6535,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2421052631578949e-05,
+      "loss": 0.6405,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.6335,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2842105263157896e-05,
+      "loss": 0.6411,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.305263157894737e-05,
+      "loss": 0.6601,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3263157894736843e-05,
+      "loss": 0.6366,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3473684210526316e-05,
+      "loss": 0.6419,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.6178,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3894736842105265e-05,
+      "loss": 0.6261,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4105263157894738e-05,
+      "loss": 0.6107,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4315789473684212e-05,
+      "loss": 0.6096,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4526315789473687e-05,
+      "loss": 0.6263,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.6173,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4947368421052632e-05,
+      "loss": 0.6181,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5157894736842107e-05,
+      "loss": 0.6005,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.536842105263158e-05,
+      "loss": 0.599,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5578947368421052e-05,
+      "loss": 0.639,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.6264,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6091,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6210526315789473e-05,
+      "loss": 0.6152,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.642105263157895e-05,
+      "loss": 0.6122,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6631578947368423e-05,
+      "loss": 0.6126,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.6349,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.705263157894737e-05,
+      "loss": 0.6302,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7263157894736843e-05,
+      "loss": 0.6192,
+      "step": 82
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7473684210526317e-05,
+      "loss": 0.628,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.768421052631579e-05,
+      "loss": 0.6063,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.6297,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.810526315789474e-05,
+      "loss": 0.6247,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.831578947368421e-05,
+      "loss": 0.6127,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8526315789473684e-05,
+      "loss": 0.6185,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.873684210526316e-05,
+      "loss": 0.6344,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.6124,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9157894736842108e-05,
+      "loss": 0.5998,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936842105263158e-05,
+      "loss": 0.6161,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578947368421055e-05,
+      "loss": 0.6068,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9789473684210528e-05,
+      "loss": 0.6011,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.6119,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999990414810826e-05,
+      "loss": 0.6175,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999996165926168e-05,
+      "loss": 0.6157,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999913733407686e-05,
+      "loss": 0.5994,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999984663734072e-05,
+      "loss": 0.6222,
+      "step": 99
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999760371189406e-05,
+      "loss": 0.6121,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999965493511912e-05,
+      "loss": 0.6196,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999953032933199e-05,
+      "loss": 0.6012,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999938655406689e-05,
+      "loss": 0.6284,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999223609599437e-05,
+      "loss": 0.5822,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999041496242006e-05,
+      "loss": 0.5853,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9998840214343715e-05,
+      "loss": 0.6216,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999861976429043e-05,
+      "loss": 0.6212,
+      "step": 107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999838014650476e-05,
+      "loss": 0.607,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998121361446064e-05,
+      "loss": 0.607,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9997843409610437e-05,
+      "loss": 0.5782,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9997546291530728e-05,
+      "loss": 0.5909,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9997230007776526e-05,
+      "loss": 0.6034,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999689455895415e-05,
+      "loss": 0.6029,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999653994570668e-05,
+      "loss": 0.6196,
+      "step": 114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996166168713913e-05,
+      "loss": 0.6207,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9995773228692396e-05,
+      "loss": 0.5868,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9995361126395413e-05,
+      "loss": 0.5853,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9994929862612976e-05,
+      "loss": 0.6385,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999447943817184e-05,
+      "loss": 0.5796,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999400985393548e-05,
+      "loss": 0.6038,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999352111080411e-05,
+      "loss": 0.6294,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999301320971466e-05,
+      "loss": 0.601,
+      "step": 122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999248615164081e-05,
+      "loss": 0.5895,
+      "step": 123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999193993759294e-05,
+      "loss": 0.5768,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9991374568618166e-05,
+      "loss": 0.6074,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9990790045800326e-05,
+      "loss": 0.5561,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9990186370259964e-05,
+      "loss": 0.625,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9989563543154356e-05,
+      "loss": 0.5977,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9988921565677478e-05,
+      "loss": 0.5792,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9988260439060033e-05,
+      "loss": 0.6142,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998758016456942e-05,
+      "loss": 0.5963,
+      "step": 131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9986880743509758e-05,
+      "loss": 0.6284,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9986162177221855e-05,
+      "loss": 0.5909,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9985424467083233e-05,
+      "loss": 0.5801,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998466761450811e-05,
+      "loss": 0.5991,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9983891620947405e-05,
+      "loss": 0.6007,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998309648788872e-05,
+      "loss": 0.6053,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9982282216856363e-05,
+      "loss": 0.5733,
+      "step": 138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9981448809411318e-05,
+      "loss": 0.5798,
+      "step": 139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9980596267151253e-05,
+      "loss": 0.5654,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9979724591710536e-05,
+      "loss": 0.6005,
+      "step": 141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9978833784760192e-05,
+      "loss": 0.5875,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9977923848007937e-05,
+      "loss": 0.5901,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9976994783198148e-05,
+      "loss": 0.6008,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.997604659211189e-05,
+      "loss": 0.584,
+      "step": 145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9975079276566866e-05,
+      "loss": 0.595,
+      "step": 146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9974092838417464e-05,
+      "loss": 0.5594,
+      "step": 147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9973087279554725e-05,
+      "loss": 0.5896,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.997206260190634e-05,
+      "loss": 0.6116,
+      "step": 149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9971018807436656e-05,
+      "loss": 0.585,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9969955898146666e-05,
+      "loss": 0.5912,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9968873876074008e-05,
+      "loss": 0.5674,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9967772743292955e-05,
+      "loss": 0.5737,
+      "step": 153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.996665250191443e-05,
+      "loss": 0.5955,
+      "step": 154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.996551315408597e-05,
+      "loss": 0.5804,
+      "step": 155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9964354701991747e-05,
+      "loss": 0.5757,
+      "step": 156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9963177147852565e-05,
+      "loss": 0.5777,
+      "step": 157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9961980493925835e-05,
+      "loss": 0.5933,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9960764742505584e-05,
+      "loss": 0.5802,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9959529895922466e-05,
+      "loss": 0.569,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9958275956543713e-05,
+      "loss": 0.567,
+      "step": 161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9957002926773182e-05,
+      "loss": 0.5844,
+      "step": 162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.995571080905132e-05,
+      "loss": 0.5729,
+      "step": 163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9954399605855166e-05,
+      "loss": 0.5924,
+      "step": 164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9953069319698345e-05,
+      "loss": 0.591,
+      "step": 165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9951719953131064e-05,
+      "loss": 0.5737,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9950351508740108e-05,
+      "loss": 0.5842,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9948963989148844e-05,
+      "loss": 0.5747,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9947557397017195e-05,
+      "loss": 0.53,
+      "step": 169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.994613173504165e-05,
+      "loss": 0.5498,
+      "step": 170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9944687005955262e-05,
+      "loss": 0.5582,
+      "step": 171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9943223212527623e-05,
+      "loss": 0.5632,
+      "step": 172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9941740357564888e-05,
+      "loss": 0.5699,
+      "step": 173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9940238443909744e-05,
+      "loss": 0.5934,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9938717474441415e-05,
+      "loss": 0.5768,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9937177452075656e-05,
+      "loss": 0.5581,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.993561837976475e-05,
+      "loss": 0.5747,
+      "step": 177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9934040260497504e-05,
+      "loss": 0.572,
+      "step": 178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993244309729922e-05,
+      "loss": 0.5812,
+      "step": 179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9930826893231727e-05,
+      "loss": 0.5415,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992919165139335e-05,
+      "loss": 0.5762,
+      "step": 181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9927537374918903e-05,
+      "loss": 0.586,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9925864066979706e-05,
+      "loss": 0.5929,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992417173078355e-05,
+      "loss": 0.5711,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922460369574703e-05,
+      "loss": 0.5924,
+      "step": 185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9920729986633914e-05,
+      "loss": 0.5936,
+      "step": 186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9918980585278396e-05,
+      "loss": 0.584,
+      "step": 187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9917212168861807e-05,
+      "loss": 0.5773,
+      "step": 188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9915424740774278e-05,
+      "loss": 0.5479,
+      "step": 189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9913618304442372e-05,
+      "loss": 0.5738,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9911792863329095e-05,
+      "loss": 0.5709,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9909948420933886e-05,
+      "loss": 0.5622,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9908084980792615e-05,
+      "loss": 0.5594,
+      "step": 193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9906202546477568e-05,
+      "loss": 0.5684,
+      "step": 194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.990430112159744e-05,
+      "loss": 0.5692,
+      "step": 195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9902380709797326e-05,
+      "loss": 0.5608,
+      "step": 196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9900441314758747e-05,
+      "loss": 0.5311,
+      "step": 197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.989848294019958e-05,
+      "loss": 0.5544,
+      "step": 198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.989650558987411e-05,
+      "loss": 0.5433,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9894509267572987e-05,
+      "loss": 0.551,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9892493977123246e-05,
+      "loss": 0.5752,
+      "step": 201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.989045972238827e-05,
+      "loss": 0.5546,
+      "step": 202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9888406507267802e-05,
+      "loss": 0.5693,
+      "step": 203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9886334335697933e-05,
+      "loss": 0.5631,
+      "step": 204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9884243211651097e-05,
+      "loss": 0.5873,
+      "step": 205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9882133139136052e-05,
+      "loss": 0.5439,
+      "step": 206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9880004122197893e-05,
+      "loss": 0.5722,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9877856164918017e-05,
+      "loss": 0.5644,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9875689271414154e-05,
+      "loss": 0.5477,
+      "step": 209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.987350344584031e-05,
+      "loss": 0.5763,
+      "step": 210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9871298692386796e-05,
+      "loss": 0.5819,
+      "step": 211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.986907501528021e-05,
+      "loss": 0.5831,
+      "step": 212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9866832418783428e-05,
+      "loss": 0.5688,
+      "step": 213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9864570907195587e-05,
+      "loss": 0.5479,
+      "step": 214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.986229048485209e-05,
+      "loss": 0.5596,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9859991156124603e-05,
+      "loss": 0.5374,
+      "step": 216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9857672925421013e-05,
+      "loss": 0.57,
+      "step": 217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.985533579718546e-05,
+      "loss": 0.5638,
+      "step": 218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.985297977589831e-05,
+      "loss": 0.5696,
+      "step": 219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9850604866076146e-05,
+      "loss": 0.5432,
+      "step": 220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.984821107227176e-05,
+      "loss": 0.5599,
+      "step": 221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9845798399074135e-05,
+      "loss": 0.5789,
+      "step": 222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.984336685110847e-05,
+      "loss": 0.5592,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.984091643303613e-05,
+      "loss": 0.5382,
+      "step": 224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9838447149554656e-05,
+      "loss": 0.5562,
+      "step": 225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9835959005397764e-05,
+      "loss": 0.5486,
+      "step": 226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9833452005335313e-05,
+      "loss": 0.5524,
+      "step": 227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.983092615417332e-05,
+      "loss": 0.5477,
+      "step": 228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.982838145675394e-05,
+      "loss": 0.5812,
+      "step": 229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9825817917955455e-05,
+      "loss": 0.5804,
+      "step": 230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9823235542692257e-05,
+      "loss": 0.5571,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9820634335914867e-05,
+      "loss": 0.5525,
+      "step": 232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.981801430260989e-05,
+      "loss": 0.5423,
+      "step": 233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.981537544780004e-05,
+      "loss": 0.5669,
+      "step": 234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9812717776544086e-05,
+      "loss": 0.5443,
+      "step": 235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9810041293936893e-05,
+      "loss": 0.5729,
+      "step": 236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.980734600510938e-05,
+      "loss": 0.5464,
+      "step": 237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9804631915228513e-05,
+      "loss": 0.5694,
+      "step": 238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.980189902949731e-05,
+      "loss": 0.545,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9799147353154812e-05,
+      "loss": 0.5623,
+      "step": 240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9796376891476092e-05,
+      "loss": 0.5405,
+      "step": 241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9793587649772226e-05,
+      "loss": 0.5404,
+      "step": 242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9790779633390292e-05,
+      "loss": 0.5355,
+      "step": 243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9787952847713375e-05,
+      "loss": 0.5423,
+      "step": 244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.978510729816052e-05,
+      "loss": 0.5413,
+      "step": 245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9782242990186757e-05,
+      "loss": 0.5644,
+      "step": 246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9779359929283072e-05,
+      "loss": 0.5429,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97764581209764e-05,
+      "loss": 0.5513,
+      "step": 248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9773537570829622e-05,
+      "loss": 0.5856,
+      "step": 249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9770598284441542e-05,
+      "loss": 0.5861,
+      "step": 250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9767640267446882e-05,
+      "loss": 0.551,
+      "step": 251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.976466352551627e-05,
+      "loss": 0.5646,
+      "step": 252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9761668064356235e-05,
+      "loss": 0.5582,
+      "step": 253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.975865388970919e-05,
+      "loss": 0.5385,
+      "step": 254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.975562100735342e-05,
+      "loss": 0.5545,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9752569423103077e-05,
+      "loss": 0.5581,
+      "step": 256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9749499142808168e-05,
+      "loss": 0.5197,
+      "step": 257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9746410172354524e-05,
+      "loss": 0.5649,
+      "step": 258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.974330251766383e-05,
+      "loss": 0.5476,
+      "step": 259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9740176184693575e-05,
+      "loss": 0.5367,
+      "step": 260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.973703117943705e-05,
+      "loss": 0.5644,
+      "step": 261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.973386750792336e-05,
+      "loss": 0.5413,
+      "step": 262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9730685176217378e-05,
+      "loss": 0.5797,
+      "step": 263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9727484190419752e-05,
+      "loss": 0.5322,
+      "step": 264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9724264556666897e-05,
+      "loss": 0.5509,
+      "step": 265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.972102628113097e-05,
+      "loss": 0.5427,
+      "step": 266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.971776937001987e-05,
+      "loss": 0.5163,
+      "step": 267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9714493829577214e-05,
+      "loss": 0.5419,
+      "step": 268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9711199666082345e-05,
+      "loss": 0.5586,
+      "step": 269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9707886885850292e-05,
+      "loss": 0.5533,
+      "step": 270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9704555495231783e-05,
+      "loss": 0.5544,
+      "step": 271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.970120550061322e-05,
+      "loss": 0.5444,
+      "step": 272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9697836908416668e-05,
+      "loss": 0.5535,
+      "step": 273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9694449725099844e-05,
+      "loss": 0.5651,
+      "step": 274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9691043957156113e-05,
+      "loss": 0.55,
+      "step": 275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9687619611114454e-05,
+      "loss": 0.5333,
+      "step": 276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9684176693539476e-05,
+      "loss": 0.5243,
+      "step": 277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9680715211031375e-05,
+      "loss": 0.5389,
+      "step": 278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9677235170225942e-05,
+      "loss": 0.5429,
+      "step": 279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9673736577794553e-05,
+      "loss": 0.5718,
+      "step": 280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.967021944044414e-05,
+      "loss": 0.547,
+      "step": 281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9666683764917192e-05,
+      "loss": 0.5527,
+      "step": 282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9663129557991727e-05,
+      "loss": 0.544,
+      "step": 283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.96595568264813e-05,
+      "loss": 0.5255,
+      "step": 284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9655965577234963e-05,
+      "loss": 0.5394,
+      "step": 285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9652355817137284e-05,
+      "loss": 0.5621,
+      "step": 286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.964872755310831e-05,
+      "loss": 0.5437,
+      "step": 287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9645080792103557e-05,
+      "loss": 0.5486,
+      "step": 288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9641415541114007e-05,
+      "loss": 0.5507,
+      "step": 289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9637731807166083e-05,
+      "loss": 0.5493,
+      "step": 290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9634029597321646e-05,
+      "loss": 0.538,
+      "step": 291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.963030891867796e-05,
+      "loss": 0.5025,
+      "step": 292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9626569778367725e-05,
+      "loss": 0.5595,
+      "step": 293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9622812183559e-05,
+      "loss": 0.5599,
+      "step": 294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9619036141455242e-05,
+      "loss": 0.5268,
+      "step": 295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.961524165929527e-05,
+      "loss": 0.5293,
+      "step": 296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.961142874435324e-05,
+      "loss": 0.5257,
+      "step": 297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9607597403938666e-05,
+      "loss": 0.5238,
+      "step": 298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9603747645396363e-05,
+      "loss": 0.5261,
+      "step": 299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.959987947610647e-05,
+      "loss": 0.5447,
+      "step": 300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9595992903484408e-05,
+      "loss": 0.5381,
+      "step": 301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.959208793498089e-05,
+      "loss": 0.5244,
+      "step": 302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9588164578081885e-05,
+      "loss": 0.5415,
+      "step": 303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.958422284030862e-05,
+      "loss": 0.5367,
+      "step": 304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9580262729217548e-05,
+      "loss": 0.5506,
+      "step": 305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.957628425240036e-05,
+      "loss": 0.5489,
+      "step": 306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9572287417483942e-05,
+      "loss": 0.5248,
+      "step": 307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9568272232130376e-05,
+      "loss": 0.5302,
+      "step": 308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.956423870403693e-05,
+      "loss": 0.5058,
+      "step": 309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9560186840936026e-05,
+      "loss": 0.5368,
+      "step": 310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9556116650595237e-05,
+      "loss": 0.5238,
+      "step": 311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9552028140817278e-05,
+      "loss": 0.5235,
+      "step": 312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9547921319439974e-05,
+      "loss": 0.5596,
+      "step": 313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9543796194336256e-05,
+      "loss": 0.5566,
+      "step": 314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9539652773414147e-05,
+      "loss": 0.533,
+      "step": 315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9535491064616737e-05,
+      "loss": 0.4995,
+      "step": 316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9531311075922183e-05,
+      "loss": 0.5311,
+      "step": 317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9527112815343684e-05,
+      "loss": 0.5355,
+      "step": 318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.952289629092946e-05,
+      "loss": 0.5235,
+      "step": 319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9518661510762748e-05,
+      "loss": 0.5144,
+      "step": 320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.951440848296178e-05,
+      "loss": 0.5421,
+      "step": 321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9510137215679782e-05,
+      "loss": 0.5397,
+      "step": 322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.950584771710492e-05,
+      "loss": 0.5215,
+      "step": 323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9501539995460336e-05,
+      "loss": 0.5329,
+      "step": 324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9497214059004087e-05,
+      "loss": 0.5148,
+      "step": 325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9492869916029164e-05,
+      "loss": 0.5217,
+      "step": 326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.948850757486345e-05,
+      "loss": 0.5155,
+      "step": 327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9484127043869723e-05,
+      "loss": 0.5232,
+      "step": 328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.947972833144562e-05,
+      "loss": 0.5434,
+      "step": 329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.947531144602364e-05,
+      "loss": 0.5391,
+      "step": 330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.947087639607112e-05,
+      "loss": 0.5036,
+      "step": 331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9466423190090223e-05,
+      "loss": 0.5387,
+      "step": 332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.946195183661791e-05,
+      "loss": 0.5271,
+      "step": 333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9457462344225934e-05,
+      "loss": 0.5276,
+      "step": 334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9452954721520824e-05,
+      "loss": 0.5164,
+      "step": 335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9448428977143857e-05,
+      "loss": 0.5296,
+      "step": 336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9443885119771066e-05,
+      "loss": 0.5102,
+      "step": 337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.943932315811319e-05,
+      "loss": 0.5382,
+      "step": 338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9434743100915685e-05,
+      "loss": 0.534,
+      "step": 339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9430144956958694e-05,
+      "loss": 0.5153,
+      "step": 340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9425528735057034e-05,
+      "loss": 0.513,
+      "step": 341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9420894444060173e-05,
+      "loss": 0.5001,
+      "step": 342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9416242092852226e-05,
+      "loss": 0.5296,
+      "step": 343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9411571690351926e-05,
+      "loss": 0.5657,
+      "step": 344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9406883245512607e-05,
+      "loss": 0.5061,
+      "step": 345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9402176767322203e-05,
+      "loss": 0.5091,
+      "step": 346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93974522648032e-05,
+      "loss": 0.4752,
+      "step": 347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.939270974701266e-05,
+      "loss": 0.498,
+      "step": 348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9387949223042164e-05,
+      "loss": 0.5314,
+      "step": 349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9383170702017812e-05,
+      "loss": 0.5267,
+      "step": 350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9378374193100215e-05,
+      "loss": 0.5032,
+      "step": 351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9373559705484463e-05,
+      "loss": 0.4812,
+      "step": 352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9368727248400106e-05,
+      "loss": 0.5105,
+      "step": 353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9363876831111147e-05,
+      "loss": 0.4978,
+      "step": 354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9359008462916026e-05,
+      "loss": 0.5183,
+      "step": 355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9354122153147584e-05,
+      "loss": 0.4997,
+      "step": 356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9349217911173065e-05,
+      "loss": 0.4956,
+      "step": 357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.934429574639408e-05,
+      "loss": 0.5116,
+      "step": 358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.933935566824661e-05,
+      "loss": 0.5203,
+      "step": 359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9334397686200975e-05,
+      "loss": 0.5262,
+      "step": 360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9329421809761806e-05,
+      "loss": 0.5249,
+      "step": 361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9324428048468053e-05,
+      "loss": 0.5111,
+      "step": 362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.931941641189294e-05,
+      "loss": 0.5074,
+      "step": 363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.931438690964397e-05,
+      "loss": 0.5105,
+      "step": 364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.930933955136288e-05,
+      "loss": 0.5243,
+      "step": 365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.930427434672566e-05,
+      "loss": 0.5058,
+      "step": 366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9299191305442488e-05,
+      "loss": 0.5207,
+      "step": 367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9294090437257752e-05,
+      "loss": 0.5373,
+      "step": 368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9288971751950004e-05,
+      "loss": 0.5133,
+      "step": 369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9283835259331963e-05,
+      "loss": 0.543,
+      "step": 370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9278680969250478e-05,
+      "loss": 0.5087,
+      "step": 371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9273508891586517e-05,
+      "loss": 0.5096,
+      "step": 372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.926831903625515e-05,
+      "loss": 0.4976,
+      "step": 373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9263111413205523e-05,
+      "loss": 0.5143,
+      "step": 374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9257886032420853e-05,
+      "loss": 0.5063,
+      "step": 375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9252642903918383e-05,
+      "loss": 0.5053,
+      "step": 376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9247382037749398e-05,
+      "loss": 0.5044,
+      "step": 377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9242103443999172e-05,
+      "loss": 0.4987,
+      "step": 378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.923680713278697e-05,
+      "loss": 0.5056,
+      "step": 379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.923149311426602e-05,
+      "loss": 0.5009,
+      "step": 380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.92261613986235e-05,
+      "loss": 0.5236,
+      "step": 381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9220811996080508e-05,
+      "loss": 0.5323,
+      "step": 382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9215444916892053e-05,
+      "loss": 0.486,
+      "step": 383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9210060171347025e-05,
+      "loss": 0.5399,
+      "step": 384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9204657769768186e-05,
+      "loss": 0.5251,
+      "step": 385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.919923772251215e-05,
+      "loss": 0.5372,
+      "step": 386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.919380003996935e-05,
+      "loss": 0.5241,
+      "step": 387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.918834473256402e-05,
+      "loss": 0.5134,
+      "step": 388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9182871810754204e-05,
+      "loss": 0.5335,
+      "step": 389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.917738128503169e-05,
+      "loss": 0.51,
+      "step": 390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.917187316592203e-05,
+      "loss": 0.4933,
+      "step": 391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9166347463984498e-05,
+      "loss": 0.5079,
+      "step": 392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9160804189812066e-05,
+      "loss": 0.5167,
+      "step": 393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9155243354031407e-05,
+      "loss": 0.4794,
+      "step": 394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.914966496730285e-05,
+      "loss": 0.5303,
+      "step": 395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9144069040320375e-05,
+      "loss": 0.4753,
+      "step": 396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9138455583811584e-05,
+      "loss": 0.4949,
+      "step": 397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.913282460853769e-05,
+      "loss": 0.5007,
+      "step": 398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9127176125293476e-05,
+      "loss": 0.5108,
+      "step": 399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.912151014490731e-05,
+      "loss": 0.4995,
+      "step": 400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9115826678241085e-05,
+      "loss": 0.4843,
+      "step": 401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9110125736190223e-05,
+      "loss": 0.5235,
+      "step": 402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9104407329683643e-05,
+      "loss": 0.5177,
+      "step": 403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9098671469683748e-05,
+      "loss": 0.4963,
+      "step": 404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9092918167186398e-05,
+      "loss": 0.4813,
+      "step": 405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9087147433220894e-05,
+      "loss": 0.5004,
+      "step": 406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.908135927884995e-05,
+      "loss": 0.5014,
+      "step": 407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9075553715169676e-05,
+      "loss": 0.5268,
+      "step": 408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9069730753309558e-05,
+      "loss": 0.4703,
+      "step": 409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9063890404432433e-05,
+      "loss": 0.5249,
+      "step": 410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9058032679734474e-05,
+      "loss": 0.5067,
+      "step": 411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9052157590445156e-05,
+      "loss": 0.4707,
+      "step": 412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9046265147827255e-05,
+      "loss": 0.5288,
+      "step": 413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9040355363176796e-05,
+      "loss": 0.4931,
+      "step": 414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.903442824782307e-05,
+      "loss": 0.5021,
+      "step": 415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9028483813128576e-05,
+      "loss": 0.5239,
+      "step": 416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902252207048902e-05,
+      "loss": 0.5114,
+      "step": 417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9016543031333288e-05,
+      "loss": 0.4889,
+      "step": 418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9010546707123426e-05,
+      "loss": 0.4856,
+      "step": 419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9004533109354618e-05,
+      "loss": 0.5175,
+      "step": 420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8998502249555147e-05,
+      "loss": 0.5069,
+      "step": 421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8992454139286415e-05,
+      "loss": 0.5245,
+      "step": 422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8986388790142866e-05,
+      "loss": 0.4987,
+      "step": 423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8980306213752005e-05,
+      "loss": 0.4864,
+      "step": 424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8974206421774366e-05,
+      "loss": 0.4936,
+      "step": 425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.896808942590348e-05,
+      "loss": 0.5244,
+      "step": 426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8961955237865858e-05,
+      "loss": 0.5304,
+      "step": 427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.895580386942097e-05,
+      "loss": 0.5019,
+      "step": 428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8949635332361222e-05,
+      "loss": 0.5069,
+      "step": 429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8943449638511934e-05,
+      "loss": 0.4657,
+      "step": 430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.893724679973132e-05,
+      "loss": 0.4933,
+      "step": 431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8931026827910447e-05,
+      "loss": 0.4911,
+      "step": 432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.892478973497324e-05,
+      "loss": 0.4859,
+      "step": 433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8918535532876444e-05,
+      "loss": 0.4997,
+      "step": 434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8912264233609603e-05,
+      "loss": 0.4896,
+      "step": 435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.890597584919503e-05,
+      "loss": 0.4891,
+      "step": 436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88996703916878e-05,
+      "loss": 0.5035,
+      "step": 437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.889334787317571e-05,
+      "loss": 0.5242,
+      "step": 438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.888700830577927e-05,
+      "loss": 0.495,
+      "step": 439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8880651701651668e-05,
+      "loss": 0.5303,
+      "step": 440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8874278072978756e-05,
+      "loss": 0.4986,
+      "step": 441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.886788743197902e-05,
+      "loss": 0.5022,
+      "step": 442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8861479790903562e-05,
+      "loss": 0.5014,
+      "step": 443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8855055162036073e-05,
+      "loss": 0.5198,
+      "step": 444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.884861355769281e-05,
+      "loss": 0.4739,
+      "step": 445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.884215499022257e-05,
+      "loss": 0.4837,
+      "step": 446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8835679472006673e-05,
+      "loss": 0.508,
+      "step": 447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.882918701545893e-05,
+      "loss": 0.5189,
+      "step": 448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.882267763302563e-05,
+      "loss": 0.4951,
+      "step": 449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.88161513371855e-05,
+      "loss": 0.5053,
+      "step": 450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8809608140449703e-05,
+      "loss": 0.5074,
+      "step": 451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8803048055361786e-05,
+      "loss": 0.4807,
+      "step": 452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.879647109449769e-05,
+      "loss": 0.4767,
+      "step": 453
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.878987727046569e-05,
+      "loss": 0.4888,
+      "step": 454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.87832665959064e-05,
+      "loss": 0.4772,
+      "step": 455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8776639083492728e-05,
+      "loss": 0.4946,
+      "step": 456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8769994745929873e-05,
+      "loss": 0.5071,
+      "step": 457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.876333359595528e-05,
+      "loss": 0.5063,
+      "step": 458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8756655646338624e-05,
+      "loss": 0.4664,
+      "step": 459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8749960909881788e-05,
+      "loss": 0.4911,
+      "step": 460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.874324939941883e-05,
+      "loss": 0.4927,
+      "step": 461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8736521127815978e-05,
+      "loss": 0.4888,
+      "step": 462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.872977610797158e-05,
+      "loss": 0.4868,
+      "step": 463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8723014352816092e-05,
+      "loss": 0.4878,
+      "step": 464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8716235875312058e-05,
+      "loss": 0.4884,
+      "step": 465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.870944068845407e-05,
+      "loss": 0.4478,
+      "step": 466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.870262880526876e-05,
+      "loss": 0.5093,
+      "step": 467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8695800238814775e-05,
+      "loss": 0.466,
+      "step": 468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.868895500218272e-05,
+      "loss": 0.4834,
+      "step": 469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8682093108495186e-05,
+      "loss": 0.5018,
+      "step": 470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8675214570906676e-05,
+      "loss": 0.4842,
+      "step": 471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8668319402603607e-05,
+      "loss": 0.4873,
+      "step": 472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8661407616804283e-05,
+      "loss": 0.5112,
+      "step": 473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8654479226758853e-05,
+      "loss": 0.4967,
+      "step": 474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.86475342457493e-05,
+      "loss": 0.4942,
+      "step": 475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8640572687089426e-05,
+      "loss": 0.4984,
+      "step": 476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.863359456412479e-05,
+      "loss": 0.4937,
+      "step": 477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.862659989023273e-05,
+      "loss": 0.4555,
+      "step": 478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.861958867882229e-05,
+      "loss": 0.5226,
+      "step": 479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8612560943334233e-05,
+      "loss": 0.4804,
+      "step": 480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8605516697240995e-05,
+      "loss": 0.5174,
+      "step": 481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8598455954046658e-05,
+      "loss": 0.4824,
+      "step": 482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8591378727286937e-05,
+      "loss": 0.4887,
+      "step": 483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8584285030529144e-05,
+      "loss": 0.5137,
+      "step": 484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.857717487737216e-05,
+      "loss": 0.4893,
+      "step": 485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.857004828144642e-05,
+      "loss": 0.4968,
+      "step": 486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.856290525641388e-05,
+      "loss": 0.4549,
+      "step": 487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8555745815967986e-05,
+      "loss": 0.454,
+      "step": 488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8548569973833657e-05,
+      "loss": 0.4674,
+      "step": 489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8541377743767253e-05,
+      "loss": 0.4525,
+      "step": 490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8534169139556553e-05,
+      "loss": 0.4912,
+      "step": 491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8526944175020728e-05,
+      "loss": 0.5091,
+      "step": 492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8519702864010303e-05,
+      "loss": 0.5042,
+      "step": 493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8512445220407143e-05,
+      "loss": 0.4652,
+      "step": 494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8505171258124428e-05,
+      "loss": 0.4914,
+      "step": 495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8497880991106624e-05,
+      "loss": 0.4979,
+      "step": 496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.849057443332944e-05,
+      "loss": 0.4719,
+      "step": 497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8483251598799827e-05,
+      "loss": 0.4571,
+      "step": 498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.847591250155594e-05,
+      "loss": 0.4942,
+      "step": 499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8468557155667102e-05,
+      "loss": 0.4612,
+      "step": 500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8461185575233787e-05,
+      "loss": 0.4519,
+      "step": 501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.84537977743876e-05,
+      "loss": 0.4869,
+      "step": 502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8446393767291234e-05,
+      "loss": 0.4922,
+      "step": 503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8438973568138446e-05,
+      "loss": 0.4945,
+      "step": 504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8431537191154038e-05,
+      "loss": 0.5007,
+      "step": 505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.842408465059383e-05,
+      "loss": 0.4665,
+      "step": 506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8416615960744624e-05,
+      "loss": 0.4905,
+      "step": 507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8409131135924178e-05,
+      "loss": 0.4943,
+      "step": 508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.840163019048119e-05,
+      "loss": 0.4644,
+      "step": 509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.839411313879525e-05,
+      "loss": 0.4856,
+      "step": 510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8386579995276836e-05,
+      "loss": 0.4578,
+      "step": 511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.837903077436726e-05,
+      "loss": 0.489,
+      "step": 512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8371465490538677e-05,
+      "loss": 0.4773,
+      "step": 513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8363884158294014e-05,
+      "loss": 0.4532,
+      "step": 514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8356286792166968e-05,
+      "loss": 0.4589,
+      "step": 515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8348673406721988e-05,
+      "loss": 0.4844,
+      "step": 516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8341044016554215e-05,
+      "loss": 0.4646,
+      "step": 517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.833339863628948e-05,
+      "loss": 0.5153,
+      "step": 518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8325737280584266e-05,
+      "loss": 0.4582,
+      "step": 519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8318059964125682e-05,
+      "loss": 0.4663,
+      "step": 520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8310366701631438e-05,
+      "loss": 0.4764,
+      "step": 521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8302657507849804e-05,
+      "loss": 0.4915,
+      "step": 522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8294932397559594e-05,
+      "loss": 0.4773,
+      "step": 523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8287191385570145e-05,
+      "loss": 0.5073,
+      "step": 524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8279434486721265e-05,
+      "loss": 0.4685,
+      "step": 525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.827166171588322e-05,
+      "loss": 0.4718,
+      "step": 526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.826387308795671e-05,
+      "loss": 0.4787,
+      "step": 527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8256068617872827e-05,
+      "loss": 0.4615,
+      "step": 528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.824824832059304e-05,
+      "loss": 0.4784,
+      "step": 529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8240412211109147e-05,
+      "loss": 0.4909,
+      "step": 530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8232560304443275e-05,
+      "loss": 0.4791,
+      "step": 531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.822469261564782e-05,
+      "loss": 0.4597,
+      "step": 532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8216809159805443e-05,
+      "loss": 0.4891,
+      "step": 533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.820890995202902e-05,
+      "loss": 0.4967,
+      "step": 534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.820099500746164e-05,
+      "loss": 0.4782,
+      "step": 535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8193064341276546e-05,
+      "loss": 0.4973,
+      "step": 536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8185117968677127e-05,
+      "loss": 0.4877,
+      "step": 537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8177155904896878e-05,
+      "loss": 0.4878,
+      "step": 538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.816917816519938e-05,
+      "loss": 0.5028,
+      "step": 539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.816118476487826e-05,
+      "loss": 0.4776,
+      "step": 540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8153175719257166e-05,
+      "loss": 0.4687,
+      "step": 541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8145151043689746e-05,
+      "loss": 0.4818,
+      "step": 542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8137110753559602e-05,
+      "loss": 0.4475,
+      "step": 543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8129054864280283e-05,
+      "loss": 0.4828,
+      "step": 544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8120983391295227e-05,
+      "loss": 0.4724,
+      "step": 545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8112896350077748e-05,
+      "loss": 0.4707,
+      "step": 546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8104793756131022e-05,
+      "loss": 0.4473,
+      "step": 547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8096675624988014e-05,
+      "loss": 0.4585,
+      "step": 548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8088541972211505e-05,
+      "loss": 0.4888,
+      "step": 549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8080392813394005e-05,
+      "loss": 0.4295,
+      "step": 550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.807222816415776e-05,
+      "loss": 0.4768,
+      "step": 551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.806404804015471e-05,
+      "loss": 0.4744,
+      "step": 552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.805585245706647e-05,
+      "loss": 0.4589,
+      "step": 553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8047641430604275e-05,
+      "loss": 0.4495,
+      "step": 554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8039414976508974e-05,
+      "loss": 0.4643,
+      "step": 555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8031173110551e-05,
+      "loss": 0.4548,
+      "step": 556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.802291584853031e-05,
+      "loss": 0.4384,
+      "step": 557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8014643206276395e-05,
+      "loss": 0.4889,
+      "step": 558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8006355199648218e-05,
+      "loss": 0.4891,
+      "step": 559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7998051844534206e-05,
+      "loss": 0.477,
+      "step": 560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.79897331568522e-05,
+      "loss": 0.4758,
+      "step": 561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.798139915254945e-05,
+      "loss": 0.4623,
+      "step": 562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7973049847602543e-05,
+      "loss": 0.4375,
+      "step": 563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7964685258017416e-05,
+      "loss": 0.436,
+      "step": 564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7956305399829316e-05,
+      "loss": 0.4832,
+      "step": 565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7947910289102735e-05,
+      "loss": 0.449,
+      "step": 566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.793949994193142e-05,
+      "loss": 0.5002,
+      "step": 567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7931074374438328e-05,
+      "loss": 0.4709,
+      "step": 568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7922633602775588e-05,
+      "loss": 0.4836,
+      "step": 569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7914177643124482e-05,
+      "loss": 0.4786,
+      "step": 570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7905706511695402e-05,
+      "loss": 0.458,
+      "step": 571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7897220224727826e-05,
+      "loss": 0.455,
+      "step": 572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.788871879849029e-05,
+      "loss": 0.4534,
+      "step": 573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7880202249280354e-05,
+      "loss": 0.4704,
+      "step": 574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7871670593424556e-05,
+      "loss": 0.4837,
+      "step": 575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7863123847278405e-05,
+      "loss": 0.4695,
+      "step": 576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7854562027226342e-05,
+      "loss": 0.4724,
+      "step": 577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7845985149681696e-05,
+      "loss": 0.4594,
+      "step": 578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7837393231086665e-05,
+      "loss": 0.4557,
+      "step": 579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7828786287912282e-05,
+      "loss": 0.4394,
+      "step": 580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.782016433665839e-05,
+      "loss": 0.4711,
+      "step": 581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7811527393853584e-05,
+      "loss": 0.4505,
+      "step": 582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.780287547605522e-05,
+      "loss": 0.4695,
+      "step": 583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7794208599849348e-05,
+      "loss": 0.4673,
+      "step": 584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7785526781850695e-05,
+      "loss": 0.4513,
+      "step": 585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.777683003870264e-05,
+      "loss": 0.4495,
+      "step": 586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7768118387077163e-05,
+      "loss": 0.4337,
+      "step": 587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7759391843674834e-05,
+      "loss": 0.4569,
+      "step": 588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7750650425224765e-05,
+      "loss": 0.4334,
+      "step": 589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7741894148484585e-05,
+      "loss": 0.4348,
+      "step": 590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7733123030240412e-05,
+      "loss": 0.4589,
+      "step": 591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7724337087306803e-05,
+      "loss": 0.4565,
+      "step": 592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7715536336526754e-05,
+      "loss": 0.4689,
+      "step": 593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.770672079477163e-05,
+      "loss": 0.4683,
+      "step": 594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.769789047894116e-05,
+      "loss": 0.4353,
+      "step": 595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7689045405963396e-05,
+      "loss": 0.4801,
+      "step": 596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7680185592794673e-05,
+      "loss": 0.4497,
+      "step": 597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.767131105641959e-05,
+      "loss": 0.46,
+      "step": 598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7662421813850966e-05,
+      "loss": 0.4447,
+      "step": 599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.765351788212982e-05,
+      "loss": 0.4172,
+      "step": 600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7644599278325327e-05,
+      "loss": 0.4324,
+      "step": 601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7635666019534784e-05,
+      "loss": 0.4447,
+      "step": 602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7626718122883587e-05,
+      "loss": 0.4447,
+      "step": 603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.761775560552519e-05,
+      "loss": 0.4414,
+      "step": 604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7608778484641088e-05,
+      "loss": 0.4437,
+      "step": 605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.759978677744075e-05,
+      "loss": 0.4884,
+      "step": 606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7590780501161622e-05,
+      "loss": 0.4498,
+      "step": 607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.758175967306908e-05,
+      "loss": 0.4182,
+      "step": 608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7572724310456387e-05,
+      "loss": 0.4457,
+      "step": 609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.756367443064468e-05,
+      "loss": 0.4131,
+      "step": 610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7554610050982917e-05,
+      "loss": 0.4987,
+      "step": 611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.754553118884786e-05,
+      "loss": 0.4674,
+      "step": 612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.753643786164403e-05,
+      "loss": 0.4434,
+      "step": 613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7527330086803682e-05,
+      "loss": 0.4574,
+      "step": 614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7518207881786757e-05,
+      "loss": 0.4463,
+      "step": 615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7509071264080874e-05,
+      "loss": 0.4391,
+      "step": 616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7499920251201278e-05,
+      "loss": 0.4593,
+      "step": 617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7490754860690797e-05,
+      "loss": 0.4378,
+      "step": 618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.748157511011984e-05,
+      "loss": 0.4625,
+      "step": 619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7472381017086332e-05,
+      "loss": 0.4543,
+      "step": 620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7463172599215697e-05,
+      "loss": 0.452,
+      "step": 621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.745394987416082e-05,
+      "loss": 0.4678,
+      "step": 622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7444712859602017e-05,
+      "loss": 0.4316,
+      "step": 623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7435461573246992e-05,
+      "loss": 0.4519,
+      "step": 624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7426196032830812e-05,
+      "loss": 0.4964,
+      "step": 625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7416916256115868e-05,
+      "loss": 0.4422,
+      "step": 626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7407622260891846e-05,
+      "loss": 0.44,
+      "step": 627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7398314064975678e-05,
+      "loss": 0.4467,
+      "step": 628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.738899168621154e-05,
+      "loss": 0.4439,
+      "step": 629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.737965514247077e-05,
+      "loss": 0.4828,
+      "step": 630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7370304451651892e-05,
+      "loss": 0.4325,
+      "step": 631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.736093963168052e-05,
+      "loss": 0.4629,
+      "step": 632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.735156070050938e-05,
+      "loss": 0.4191,
+      "step": 633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7342167676118227e-05,
+      "loss": 0.4678,
+      "step": 634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.733276057651385e-05,
+      "loss": 0.4704,
+      "step": 635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7323339419730017e-05,
+      "loss": 0.4239,
+      "step": 636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7313904223827438e-05,
+      "loss": 0.4251,
+      "step": 637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.730445500689374e-05,
+      "loss": 0.417,
+      "step": 638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7294991787043433e-05,
+      "loss": 0.4485,
+      "step": 639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7285514582417863e-05,
+      "loss": 0.4635,
+      "step": 640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7276023411185193e-05,
+      "loss": 0.4629,
+      "step": 641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7266518291540356e-05,
+      "loss": 0.4446,
+      "step": 642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7256999241705027e-05,
+      "loss": 0.4295,
+      "step": 643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7247466279927585e-05,
+      "loss": 0.4771,
+      "step": 644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.723791942448308e-05,
+      "loss": 0.4328,
+      "step": 645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7228358693673185e-05,
+      "loss": 0.4564,
+      "step": 646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7218784105826197e-05,
+      "loss": 0.4601,
+      "step": 647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.720919567929696e-05,
+      "loss": 0.4681,
+      "step": 648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.719959343246684e-05,
+      "loss": 0.4157,
+      "step": 649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7189977383743713e-05,
+      "loss": 0.4321,
+      "step": 650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7180347551561916e-05,
+      "loss": 0.4402,
+      "step": 651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7170703954382193e-05,
+      "loss": 0.4471,
+      "step": 652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7161046610691686e-05,
+      "loss": 0.4539,
+      "step": 653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.715137553900389e-05,
+      "loss": 0.435,
+      "step": 654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7141690757858618e-05,
+      "loss": 0.4595,
+      "step": 655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7131992285821953e-05,
+      "loss": 0.4331,
+      "step": 656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.712228014148624e-05,
+      "loss": 0.4514,
+      "step": 657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.711255434347003e-05,
+      "loss": 0.4314,
+      "step": 658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7102814910418035e-05,
+      "loss": 0.4474,
+      "step": 659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7093061861001124e-05,
+      "loss": 0.4285,
+      "step": 660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7083295213916266e-05,
+      "loss": 0.4458,
+      "step": 661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7073514987886483e-05,
+      "loss": 0.4403,
+      "step": 662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7063721201660843e-05,
+      "loss": 0.4429,
+      "step": 663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7053913874014404e-05,
+      "loss": 0.4631,
+      "step": 664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7044093023748187e-05,
+      "loss": 0.4254,
+      "step": 665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.703425866968913e-05,
+      "loss": 0.4343,
+      "step": 666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7024410830690063e-05,
+      "loss": 0.4319,
+      "step": 667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.701454952562967e-05,
+      "loss": 0.4534,
+      "step": 668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7004674773412437e-05,
+      "loss": 0.4355,
+      "step": 669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6994786592968644e-05,
+      "loss": 0.4529,
+      "step": 670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6984885003254307e-05,
+      "loss": 0.428,
+      "step": 671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6974970023251146e-05,
+      "loss": 0.454,
+      "step": 672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6965041671966555e-05,
+      "loss": 0.4562,
+      "step": 673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6955099968433555e-05,
+      "loss": 0.4239,
+      "step": 674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6945144931710772e-05,
+      "loss": 0.4454,
+      "step": 675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6935176580882386e-05,
+      "loss": 0.4341,
+      "step": 676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6925194935058104e-05,
+      "loss": 0.452,
+      "step": 677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6915200013373116e-05,
+      "loss": 0.4621,
+      "step": 678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.690519183498807e-05,
+      "loss": 0.4188,
+      "step": 679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6895170419089017e-05,
+      "loss": 0.461,
+      "step": 680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.688513578488739e-05,
+      "loss": 0.4527,
+      "step": 681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6875087951619973e-05,
+      "loss": 0.429,
+      "step": 682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.686502693854883e-05,
+      "loss": 0.4224,
+      "step": 683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.685495276496131e-05,
+      "loss": 0.43,
+      "step": 684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.684486545016998e-05,
+      "loss": 0.4199,
+      "step": 685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6834765013512613e-05,
+      "loss": 0.4535,
+      "step": 686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6824651474352123e-05,
+      "loss": 0.4004,
+      "step": 687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6814524852076543e-05,
+      "loss": 0.4337,
+      "step": 688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6804385166098997e-05,
+      "loss": 0.4412,
+      "step": 689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6794232435857646e-05,
+      "loss": 0.4649,
+      "step": 690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6784066680815655e-05,
+      "loss": 0.4433,
+      "step": 691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6773887920461166e-05,
+      "loss": 0.4279,
+      "step": 692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.676369617430724e-05,
+      "loss": 0.4247,
+      "step": 693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6753491461891847e-05,
+      "loss": 0.4688,
+      "step": 694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6743273802777806e-05,
+      "loss": 0.4338,
+      "step": 695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.673304321655275e-05,
+      "loss": 0.4434,
+      "step": 696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6722799722829107e-05,
+      "loss": 0.3884,
+      "step": 697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6712543341244035e-05,
+      "loss": 0.4586,
+      "step": 698
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.670227409145942e-05,
+      "loss": 0.4247,
+      "step": 699
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.669199199316178e-05,
+      "loss": 0.4217,
+      "step": 700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6681697066062307e-05,
+      "loss": 0.4179,
+      "step": 701
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6671389329896755e-05,
+      "loss": 0.4384,
+      "step": 702
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.666106880442545e-05,
+      "loss": 0.4237,
+      "step": 703
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.665073550943322e-05,
+      "loss": 0.4329,
+      "step": 704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6640389464729394e-05,
+      "loss": 0.4214,
+      "step": 705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6630030690147724e-05,
+      "loss": 0.414,
+      "step": 706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6619659205546376e-05,
+      "loss": 0.4342,
+      "step": 707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6609275030807874e-05,
+      "loss": 0.4147,
+      "step": 708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6598878185839084e-05,
+      "loss": 0.4157,
+      "step": 709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6588468690571137e-05,
+      "loss": 0.4385,
+      "step": 710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.657804656495944e-05,
+      "loss": 0.4046,
+      "step": 711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6567611828983597e-05,
+      "loss": 0.4061,
+      "step": 712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6557164502647392e-05,
+      "loss": 0.4363,
+      "step": 713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.654670460597875e-05,
+      "loss": 0.4395,
+      "step": 714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6536232159029684e-05,
+      "loss": 0.4436,
+      "step": 715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.652574718187627e-05,
+      "loss": 0.4024,
+      "step": 716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.651524969461861e-05,
+      "loss": 0.4374,
+      "step": 717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6504739717380783e-05,
+      "loss": 0.4305,
+      "step": 718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.649421727031081e-05,
+      "loss": 0.4285,
+      "step": 719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6483682373580622e-05,
+      "loss": 0.4322,
+      "step": 720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6473135047386014e-05,
+      "loss": 0.4158,
+      "step": 721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.646257531194661e-05,
+      "loss": 0.4265,
+      "step": 722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6452003187505826e-05,
+      "loss": 0.444,
+      "step": 723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.644141869433082e-05,
+      "loss": 0.4209,
+      "step": 724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6430821852712465e-05,
+      "loss": 0.4377,
+      "step": 725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.642021268296531e-05,
+      "loss": 0.4087,
+      "step": 726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6409591205427535e-05,
+      "loss": 0.4256,
+      "step": 727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6398957440460913e-05,
+      "loss": 0.4158,
+      "step": 728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6388311408450773e-05,
+      "loss": 0.4295,
+      "step": 729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.637765312980596e-05,
+      "loss": 0.4256,
+      "step": 730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6366982624958803e-05,
+      "loss": 0.43,
+      "step": 731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6356299914365062e-05,
+      "loss": 0.4107,
+      "step": 732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.634560501850389e-05,
+      "loss": 0.4099,
+      "step": 733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.633489795787782e-05,
+      "loss": 0.4424,
+      "step": 734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6324178753012678e-05,
+      "loss": 0.4286,
+      "step": 735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6313447424457596e-05,
+      "loss": 0.4257,
+      "step": 736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6302703992784933e-05,
+      "loss": 0.4027,
+      "step": 737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6291948478590255e-05,
+      "loss": 0.4077,
+      "step": 738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6281180902492285e-05,
+      "loss": 0.4351,
+      "step": 739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6270401285132878e-05,
+      "loss": 0.4354,
+      "step": 740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.625960964717697e-05,
+      "loss": 0.4348,
+      "step": 741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.624880600931254e-05,
+      "loss": 0.3969,
+      "step": 742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6237990392250562e-05,
+      "loss": 0.4194,
+      "step": 743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.622716281672499e-05,
+      "loss": 0.3814,
+      "step": 744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6216323303492697e-05,
+      "loss": 0.4161,
+      "step": 745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6205471873333435e-05,
+      "loss": 0.4197,
+      "step": 746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.619460854704981e-05,
+      "loss": 0.4613,
+      "step": 747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6183733345467225e-05,
+      "loss": 0.4307,
+      "step": 748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.617284628943386e-05,
+      "loss": 0.4158,
+      "step": 749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.616194739982061e-05,
+      "loss": 0.4503,
+      "step": 750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6151036697521058e-05,
+      "loss": 0.4281,
+      "step": 751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6140114203451432e-05,
+      "loss": 0.4306,
+      "step": 752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.612917993855057e-05,
+      "loss": 0.4238,
+      "step": 753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6118233923779866e-05,
+      "loss": 0.3881,
+      "step": 754
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.610727618012325e-05,
+      "loss": 0.4061,
+      "step": 755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.609630672858713e-05,
+      "loss": 0.4208,
+      "step": 756
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6085325590200355e-05,
+      "loss": 0.4096,
+      "step": 757
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.607433278601419e-05,
+      "loss": 0.4021,
+      "step": 758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.606332833710225e-05,
+      "loss": 0.4391,
+      "step": 759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6052312264560483e-05,
+      "loss": 0.4411,
+      "step": 760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6041284589507116e-05,
+      "loss": 0.3753,
+      "step": 761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6030245333082622e-05,
+      "loss": 0.4081,
+      "step": 762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.601919451644967e-05,
+      "loss": 0.3866,
+      "step": 763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.60081321607931e-05,
+      "loss": 0.4049,
+      "step": 764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5997058287319852e-05,
+      "loss": 0.3736,
+      "step": 765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.598597291725897e-05,
+      "loss": 0.4448,
+      "step": 766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5974876071861536e-05,
+      "loss": 0.4348,
+      "step": 767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5963767772400608e-05,
+      "loss": 0.4151,
+      "step": 768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.595264804017122e-05,
+      "loss": 0.4356,
+      "step": 769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5941516896490327e-05,
+      "loss": 0.4039,
+      "step": 770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5930374362696745e-05,
+      "loss": 0.4383,
+      "step": 771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5919220460151136e-05,
+      "loss": 0.4005,
+      "step": 772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.590805521023595e-05,
+      "loss": 0.4072,
+      "step": 773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.58968786343554e-05,
+      "loss": 0.418,
+      "step": 774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5885690753935392e-05,
+      "loss": 0.4215,
+      "step": 775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5874491590423527e-05,
+      "loss": 0.3988,
+      "step": 776
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5863281165289026e-05,
+      "loss": 0.4458,
+      "step": 777
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5852059500022697e-05,
+      "loss": 0.3886,
+      "step": 778
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.584082661613689e-05,
+      "loss": 0.4162,
+      "step": 779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5829582535165475e-05,
+      "loss": 0.4311,
+      "step": 780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5818327278663778e-05,
+      "loss": 0.4139,
+      "step": 781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5807060868208555e-05,
+      "loss": 0.416,
+      "step": 782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5795783325397938e-05,
+      "loss": 0.4263,
+      "step": 783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.57844946718514e-05,
+      "loss": 0.4088,
+      "step": 784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5773194929209723e-05,
+      "loss": 0.3904,
+      "step": 785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.576188411913494e-05,
+      "loss": 0.3703,
+      "step": 786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.57505622633103e-05,
+      "loss": 0.4546,
+      "step": 787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5739229383440227e-05,
+      "loss": 0.4187,
+      "step": 788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5727885501250288e-05,
+      "loss": 0.4135,
+      "step": 789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5716530638487127e-05,
+      "loss": 0.4252,
+      "step": 790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5705164816918447e-05,
+      "loss": 0.3882,
+      "step": 791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.569378805833296e-05,
+      "loss": 0.3794,
+      "step": 792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5682400384540338e-05,
+      "loss": 0.391,
+      "step": 793
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.567100181737119e-05,
+      "loss": 0.4017,
+      "step": 794
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5659592378677e-05,
+      "loss": 0.3756,
+      "step": 795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5648172090330084e-05,
+      "loss": 0.4062,
+      "step": 796
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5636740974223576e-05,
+      "loss": 0.4028,
+      "step": 797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5625299052271357e-05,
+      "loss": 0.3799,
+      "step": 798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.561384634640802e-05,
+      "loss": 0.3824,
+      "step": 799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5602382878588837e-05,
+      "loss": 0.3712,
+      "step": 800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5590908670789713e-05,
+      "loss": 0.3905,
+      "step": 801
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5579423745007132e-05,
+      "loss": 0.3951,
+      "step": 802
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.556792812325814e-05,
+      "loss": 0.3645,
+      "step": 803
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5556421827580273e-05,
+      "loss": 0.3735,
+      "step": 804
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5544904880031533e-05,
+      "loss": 0.3563,
+      "step": 805
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5533377302690354e-05,
+      "loss": 0.3965,
+      "step": 806
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5521839117655527e-05,
+      "loss": 0.3882,
+      "step": 807
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5510290347046192e-05,
+      "loss": 0.413,
+      "step": 808
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5498731013001786e-05,
+      "loss": 0.3799,
+      "step": 809
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.548716113768198e-05,
+      "loss": 0.4008,
+      "step": 810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5475580743266668e-05,
+      "loss": 0.3633,
+      "step": 811
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5463989851955902e-05,
+      "loss": 0.3735,
+      "step": 812
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5452388485969865e-05,
+      "loss": 0.3546,
+      "step": 813
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5440776667548808e-05,
+      "loss": 0.3803,
+      "step": 814
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5429154418953025e-05,
+      "loss": 0.3806,
+      "step": 815
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5417521762462816e-05,
+      "loss": 0.3589,
+      "step": 816
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5405878720378412e-05,
+      "loss": 0.3839,
+      "step": 817
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5394225315019973e-05,
+      "loss": 0.3618,
+      "step": 818
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5382561568727513e-05,
+      "loss": 0.3847,
+      "step": 819
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5370887503860883e-05,
+      "loss": 0.3831,
+      "step": 820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5359203142799697e-05,
+      "loss": 0.3953,
+      "step": 821
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5347508507943324e-05,
+      "loss": 0.3683,
+      "step": 822
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5335803621710817e-05,
+      "loss": 0.3868,
+      "step": 823
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.532408850654089e-05,
+      "loss": 0.384,
+      "step": 824
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5312363184891857e-05,
+      "loss": 0.418,
+      "step": 825
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.530062767924161e-05,
+      "loss": 0.3665,
+      "step": 826
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.528888201208755e-05,
+      "loss": 0.3919,
+      "step": 827
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5277126205946572e-05,
+      "loss": 0.3686,
+      "step": 828
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5265360283354993e-05,
+      "loss": 0.3925,
+      "step": 829
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5253584266868542e-05,
+      "loss": 0.3627,
+      "step": 830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5241798179062279e-05,
+      "loss": 0.3816,
+      "step": 831
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5230002042530587e-05,
+      "loss": 0.3672,
+      "step": 832
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5218195879887103e-05,
+      "loss": 0.3536,
+      "step": 833
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5206379713764688e-05,
+      "loss": 0.4153,
+      "step": 834
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5194553566815375e-05,
+      "loss": 0.3788,
+      "step": 835
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5182717461710347e-05,
+      "loss": 0.3797,
+      "step": 836
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5170871421139853e-05,
+      "loss": 0.3552,
+      "step": 837
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5159015467813203e-05,
+      "loss": 0.3769,
+      "step": 838
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5147149624458713e-05,
+      "loss": 0.3907,
+      "step": 839
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5135273913823653e-05,
+      "loss": 0.392,
+      "step": 840
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.512338835867421e-05,
+      "loss": 0.3885,
+      "step": 841
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5111492981795434e-05,
+      "loss": 0.3924,
+      "step": 842
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5099587805991224e-05,
+      "loss": 0.3579,
+      "step": 843
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5087672854084247e-05,
+      "loss": 0.3576,
+      "step": 844
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5075748148915917e-05,
+      "loss": 0.37,
+      "step": 845
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5063813713346349e-05,
+      "loss": 0.3788,
+      "step": 846
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5051869570254299e-05,
+      "loss": 0.4024,
+      "step": 847
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5039915742537146e-05,
+      "loss": 0.3637,
+      "step": 848
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5027952253110834e-05,
+      "loss": 0.3659,
+      "step": 849
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5015979124909818e-05,
+      "loss": 0.3333,
+      "step": 850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5003996380887045e-05,
+      "loss": 0.3456,
+      "step": 851
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.499200404401388e-05,
+      "loss": 0.4062,
+      "step": 852
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4980002137280091e-05,
+      "loss": 0.4089,
+      "step": 853
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.496799068369379e-05,
+      "loss": 0.3791,
+      "step": 854
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4955969706281383e-05,
+      "loss": 0.3756,
+      "step": 855
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4943939228087542e-05,
+      "loss": 0.365,
+      "step": 856
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4931899272175148e-05,
+      "loss": 0.3915,
+      "step": 857
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4919849861625251e-05,
+      "loss": 0.3498,
+      "step": 858
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4907791019537029e-05,
+      "loss": 0.3761,
+      "step": 859
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4895722769027734e-05,
+      "loss": 0.3578,
+      "step": 860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4883645133232664e-05,
+      "loss": 0.3602,
+      "step": 861
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4871558135305102e-05,
+      "loss": 0.3453,
+      "step": 862
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4859461798416279e-05,
+      "loss": 0.3367,
+      "step": 863
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4847356145755331e-05,
+      "loss": 0.3569,
+      "step": 864
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4835241200529252e-05,
+      "loss": 0.3533,
+      "step": 865
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4823116985962855e-05,
+      "loss": 0.3768,
+      "step": 866
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.481098352529871e-05,
+      "loss": 0.3488,
+      "step": 867
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4798840841797126e-05,
+      "loss": 0.35,
+      "step": 868
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4786688958736085e-05,
+      "loss": 0.366,
+      "step": 869
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4774527899411206e-05,
+      "loss": 0.4093,
+      "step": 870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.47623576871357e-05,
+      "loss": 0.3673,
+      "step": 871
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4750178345240329e-05,
+      "loss": 0.3235,
+      "step": 872
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4737989897073343e-05,
+      "loss": 0.375,
+      "step": 873
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4725792366000462e-05,
+      "loss": 0.3677,
+      "step": 874
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4713585775404817e-05,
+      "loss": 0.4014,
+      "step": 875
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4701370148686906e-05,
+      "loss": 0.3595,
+      "step": 876
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4689145509264544e-05,
+      "loss": 0.3563,
+      "step": 877
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4676911880572826e-05,
+      "loss": 0.37,
+      "step": 878
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4664669286064084e-05,
+      "loss": 0.3704,
+      "step": 879
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.465241774920783e-05,
+      "loss": 0.358,
+      "step": 880
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4640157293490731e-05,
+      "loss": 0.374,
+      "step": 881
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.462788794241654e-05,
+      "loss": 0.3679,
+      "step": 882
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4615609719506068e-05,
+      "loss": 0.3433,
+      "step": 883
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4603322648297132e-05,
+      "loss": 0.3758,
+      "step": 884
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4591026752344512e-05,
+      "loss": 0.394,
+      "step": 885
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4578722055219911e-05,
+      "loss": 0.3901,
+      "step": 886
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4566408580511894e-05,
+      "loss": 0.3921,
+      "step": 887
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4554086351825858e-05,
+      "loss": 0.3722,
+      "step": 888
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4541755392783983e-05,
+      "loss": 0.3894,
+      "step": 889
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.452941572702518e-05,
+      "loss": 0.3626,
+      "step": 890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4517067378205063e-05,
+      "loss": 0.3473,
+      "step": 891
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4504710369995876e-05,
+      "loss": 0.3745,
+      "step": 892
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4492344726086475e-05,
+      "loss": 0.3487,
+      "step": 893
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.447997047018227e-05,
+      "loss": 0.374,
+      "step": 894
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4467587626005173e-05,
+      "loss": 0.3499,
+      "step": 895
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4455196217293566e-05,
+      "loss": 0.3637,
+      "step": 896
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.444279626780225e-05,
+      "loss": 0.3434,
+      "step": 897
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4430387801302397e-05,
+      "loss": 0.3733,
+      "step": 898
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4417970841581505e-05,
+      "loss": 0.3729,
+      "step": 899
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4405545412443355e-05,
+      "loss": 0.3208,
+      "step": 900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4393111537707966e-05,
+      "loss": 0.393,
+      "step": 901
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4380669241211549e-05,
+      "loss": 0.3636,
+      "step": 902
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4368218546806454e-05,
+      "loss": 0.3596,
+      "step": 903
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4355759478361134e-05,
+      "loss": 0.329,
+      "step": 904
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4343292059760092e-05,
+      "loss": 0.3893,
+      "step": 905
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4330816314903847e-05,
+      "loss": 0.3634,
+      "step": 906
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4318332267708868e-05,
+      "loss": 0.369,
+      "step": 907
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4305839942107552e-05,
+      "loss": 0.3573,
+      "step": 908
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4293339362048152e-05,
+      "loss": 0.3783,
+      "step": 909
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4280830551494757e-05,
+      "loss": 0.4214,
+      "step": 910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4268313534427235e-05,
+      "loss": 0.3405,
+      "step": 911
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4255788334841174e-05,
+      "loss": 0.3504,
+      "step": 912
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.424325497674786e-05,
+      "loss": 0.3674,
+      "step": 913
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.423071348417421e-05,
+      "loss": 0.3697,
+      "step": 914
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4218163881162743e-05,
+      "loss": 0.3666,
+      "step": 915
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4205606191771522e-05,
+      "loss": 0.3644,
+      "step": 916
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4193040440074113e-05,
+      "loss": 0.3616,
+      "step": 917
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4180466650159539e-05,
+      "loss": 0.3379,
+      "step": 918
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4167884846132228e-05,
+      "loss": 0.3268,
+      "step": 919
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4155295052111975e-05,
+      "loss": 0.3961,
+      "step": 920
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4142697292233893e-05,
+      "loss": 0.3613,
+      "step": 921
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4130091590648362e-05,
+      "loss": 0.3754,
+      "step": 922
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4117477971520997e-05,
+      "loss": 0.3641,
+      "step": 923
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.410485645903257e-05,
+      "loss": 0.345,
+      "step": 924
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4092227077379007e-05,
+      "loss": 0.3543,
+      "step": 925
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4079589850771305e-05,
+      "loss": 0.3211,
+      "step": 926
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4066944803435512e-05,
+      "loss": 0.3237,
+      "step": 927
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4054291959612656e-05,
+      "loss": 0.3467,
+      "step": 928
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.404163134355872e-05,
+      "loss": 0.3547,
+      "step": 929
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4028962979544584e-05,
+      "loss": 0.3541,
+      "step": 930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4016286891855984e-05,
+      "loss": 0.3808,
+      "step": 931
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4003603104793455e-05,
+      "loss": 0.3819,
+      "step": 932
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.39909116426723e-05,
+      "loss": 0.3418,
+      "step": 933
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3978212529822528e-05,
+      "loss": 0.3484,
+      "step": 934
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3965505790588822e-05,
+      "loss": 0.3727,
+      "step": 935
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.395279144933048e-05,
+      "loss": 0.3824,
+      "step": 936
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3940069530421377e-05,
+      "loss": 0.356,
+      "step": 937
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3927340058249914e-05,
+      "loss": 0.3481,
+      "step": 938
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3914603057218967e-05,
+      "loss": 0.3333,
+      "step": 939
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3901858551745852e-05,
+      "loss": 0.3576,
+      "step": 940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3889106566262268e-05,
+      "loss": 0.3735,
+      "step": 941
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3876347125214251e-05,
+      "loss": 0.3709,
+      "step": 942
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3863580253062137e-05,
+      "loss": 0.3448,
+      "step": 943
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3850805974280496e-05,
+      "loss": 0.3944,
+      "step": 944
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3838024313358112e-05,
+      "loss": 0.383,
+      "step": 945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3825235294797903e-05,
+      "loss": 0.3349,
+      "step": 946
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.381243894311691e-05,
+      "loss": 0.3505,
+      "step": 947
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.379963528284622e-05,
+      "loss": 0.3624,
+      "step": 948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3786824338530934e-05,
+      "loss": 0.3528,
+      "step": 949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3774006134730118e-05,
+      "loss": 0.3545,
+      "step": 950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.376118069601675e-05,
+      "loss": 0.358,
+      "step": 951
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3748348046977687e-05,
+      "loss": 0.3803,
+      "step": 952
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3735508212213598e-05,
+      "loss": 0.3091,
+      "step": 953
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3722661216338937e-05,
+      "loss": 0.3537,
+      "step": 954
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3709807083981871e-05,
+      "loss": 0.3833,
+      "step": 955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3696945839784272e-05,
+      "loss": 0.3458,
+      "step": 956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3684077508401622e-05,
+      "loss": 0.3724,
+      "step": 957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3671202114503004e-05,
+      "loss": 0.351,
+      "step": 958
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.365831968277103e-05,
+      "loss": 0.3832,
+      "step": 959
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3645430237901816e-05,
+      "loss": 0.3327,
+      "step": 960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3632533804604908e-05,
+      "loss": 0.382,
+      "step": 961
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3619630407603262e-05,
+      "loss": 0.3643,
+      "step": 962
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3606720071633178e-05,
+      "loss": 0.3476,
+      "step": 963
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3593802821444254e-05,
+      "loss": 0.3639,
+      "step": 964
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3580878681799349e-05,
+      "loss": 0.346,
+      "step": 965
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3567947677474533e-05,
+      "loss": 0.3525,
+      "step": 966
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3555009833259025e-05,
+      "loss": 0.3342,
+      "step": 967
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3542065173955163e-05,
+      "loss": 0.3259,
+      "step": 968
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3529113724378351e-05,
+      "loss": 0.3261,
+      "step": 969
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3516155509357005e-05,
+      "loss": 0.3957,
+      "step": 970
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3503190553732515e-05,
+      "loss": 0.3552,
+      "step": 971
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3490218882359189e-05,
+      "loss": 0.3265,
+      "step": 972
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3477240520104213e-05,
+      "loss": 0.335,
+      "step": 973
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3464255491847602e-05,
+      "loss": 0.3114,
+      "step": 974
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3451263822482141e-05,
+      "loss": 0.3724,
+      "step": 975
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3438265536913356e-05,
+      "loss": 0.3392,
+      "step": 976
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.342526066005945e-05,
+      "loss": 0.323,
+      "step": 977
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3412249216851265e-05,
+      "loss": 0.3617,
+      "step": 978
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3399231232232231e-05,
+      "loss": 0.3675,
+      "step": 979
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3386206731158314e-05,
+      "loss": 0.3301,
+      "step": 980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3373175738597977e-05,
+      "loss": 0.3304,
+      "step": 981
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3360138279532123e-05,
+      "loss": 0.3491,
+      "step": 982
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3347094378954055e-05,
+      "loss": 0.3807,
+      "step": 983
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3334044061869433e-05,
+      "loss": 0.3274,
+      "step": 984
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3320987353296198e-05,
+      "loss": 0.3523,
+      "step": 985
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3307924278264557e-05,
+      "loss": 0.3483,
+      "step": 986
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.329485486181692e-05,
+      "loss": 0.328,
+      "step": 987
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3281779129007852e-05,
+      "loss": 0.3462,
+      "step": 988
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.326869710490403e-05,
+      "loss": 0.319,
+      "step": 989
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3255608814584188e-05,
+      "loss": 0.3638,
+      "step": 990
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.324251428313907e-05,
+      "loss": 0.3514,
+      "step": 991
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3229413535671391e-05,
+      "loss": 0.3115,
+      "step": 992
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3216306597295783e-05,
+      "loss": 0.3339,
+      "step": 993
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.320319349313874e-05,
+      "loss": 0.3702,
+      "step": 994
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3190074248338576e-05,
+      "loss": 0.3702,
+      "step": 995
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.317694888804538e-05,
+      "loss": 0.3474,
+      "step": 996
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3163817437420969e-05,
+      "loss": 0.3736,
+      "step": 997
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3150679921638827e-05,
+      "loss": 0.3303,
+      "step": 998
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3137536365884072e-05,
+      "loss": 0.3955,
+      "step": 999
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3124386795353393e-05,
+      "loss": 0.3177,
+      "step": 1000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3111231235255018e-05,
+      "loss": 0.342,
+      "step": 1001
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.309806971080865e-05,
+      "loss": 0.3616,
+      "step": 1002
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3084902247245432e-05,
+      "loss": 0.3341,
+      "step": 1003
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3071728869807892e-05,
+      "loss": 0.3304,
+      "step": 1004
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.305854960374989e-05,
+      "loss": 0.3244,
+      "step": 1005
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3045364474336575e-05,
+      "loss": 0.3306,
+      "step": 1006
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3032173506844343e-05,
+      "loss": 0.3274,
+      "step": 1007
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3018976726560777e-05,
+      "loss": 0.3316,
+      "step": 1008
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3005774158784604e-05,
+      "loss": 0.3472,
+      "step": 1009
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2992565828825645e-05,
+      "loss": 0.3464,
+      "step": 1010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2979351762004769e-05,
+      "loss": 0.3343,
+      "step": 1011
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.296613198365384e-05,
+      "loss": 0.3363,
+      "step": 1012
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2952906519115677e-05,
+      "loss": 0.3105,
+      "step": 1013
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2939675393743996e-05,
+      "loss": 0.3428,
+      "step": 1014
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2926438632903361e-05,
+      "loss": 0.3324,
+      "step": 1015
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2913196261969148e-05,
+      "loss": 0.3272,
+      "step": 1016
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2899948306327476e-05,
+      "loss": 0.3272,
+      "step": 1017
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2886694791375183e-05,
+      "loss": 0.3502,
+      "step": 1018
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2873435742519758e-05,
+      "loss": 0.2989,
+      "step": 1019
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.28601711851793e-05,
+      "loss": 0.3391,
+      "step": 1020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2846901144782464e-05,
+      "loss": 0.344,
+      "step": 1021
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2833625646768424e-05,
+      "loss": 0.3145,
+      "step": 1022
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2820344716586806e-05,
+      "loss": 0.3171,
+      "step": 1023
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2807058379697662e-05,
+      "loss": 0.3303,
+      "step": 1024
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2793766661571399e-05,
+      "loss": 0.3404,
+      "step": 1025
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2780469587688744e-05,
+      "loss": 0.3327,
+      "step": 1026
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2767167183540688e-05,
+      "loss": 0.3252,
+      "step": 1027
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2753859474628447e-05,
+      "loss": 0.3332,
+      "step": 1028
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2740546486463404e-05,
+      "loss": 0.3448,
+      "step": 1029
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2727228244567056e-05,
+      "loss": 0.3801,
+      "step": 1030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.271390477447098e-05,
+      "loss": 0.3583,
+      "step": 1031
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.270057610171677e-05,
+      "loss": 0.3617,
+      "step": 1032
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2687242251855998e-05,
+      "loss": 0.3492,
+      "step": 1033
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2673903250450157e-05,
+      "loss": 0.335,
+      "step": 1034
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.266055912307062e-05,
+      "loss": 0.3136,
+      "step": 1035
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.264720989529858e-05,
+      "loss": 0.3569,
+      "step": 1036
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2633855592725015e-05,
+      "loss": 0.3436,
+      "step": 1037
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2620496240950626e-05,
+      "loss": 0.3591,
+      "step": 1038
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2607131865585798e-05,
+      "loss": 0.324,
+      "step": 1039
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2593762492250545e-05,
+      "loss": 0.3314,
+      "step": 1040
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.258038814657446e-05,
+      "loss": 0.3323,
+      "step": 1041
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2567008854196668e-05,
+      "loss": 0.3659,
+      "step": 1042
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.255362464076578e-05,
+      "loss": 0.3527,
+      "step": 1043
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.254023553193984e-05,
+      "loss": 0.3035,
+      "step": 1044
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2526841553386273e-05,
+      "loss": 0.3262,
+      "step": 1045
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2513442730781849e-05,
+      "loss": 0.3303,
+      "step": 1046
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2500039089812615e-05,
+      "loss": 0.3356,
+      "step": 1047
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2486630656173855e-05,
+      "loss": 0.3593,
+      "step": 1048
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2473217455570046e-05,
+      "loss": 0.3357,
+      "step": 1049
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2459799513714801e-05,
+      "loss": 0.3542,
+      "step": 1050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2446376856330824e-05,
+      "loss": 0.3184,
+      "step": 1051
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2432949509149854e-05,
+      "loss": 0.3122,
+      "step": 1052
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2419517497912623e-05,
+      "loss": 0.3511,
+      "step": 1053
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2406080848368806e-05,
+      "loss": 0.3332,
+      "step": 1054
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2392639586276974e-05,
+      "loss": 0.3052,
+      "step": 1055
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2379193737404526e-05,
+      "loss": 0.3264,
+      "step": 1056
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2365743327527668e-05,
+      "loss": 0.3246,
+      "step": 1057
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2352288382431343e-05,
+      "loss": 0.3401,
+      "step": 1058
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2338828927909192e-05,
+      "loss": 0.3492,
+      "step": 1059
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2325364989763495e-05,
+      "loss": 0.3496,
+      "step": 1060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2311896593805135e-05,
+      "loss": 0.3354,
+      "step": 1061
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2298423765853533e-05,
+      "loss": 0.3198,
+      "step": 1062
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.228494653173661e-05,
+      "loss": 0.3313,
+      "step": 1063
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2271464917290736e-05,
+      "loss": 0.3214,
+      "step": 1064
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2257978948360676e-05,
+      "loss": 0.3406,
+      "step": 1065
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2244488650799544e-05,
+      "loss": 0.3021,
+      "step": 1066
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2230994050468746e-05,
+      "loss": 0.3447,
+      "step": 1067
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2217495173237944e-05,
+      "loss": 0.313,
+      "step": 1068
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2203992044984995e-05,
+      "loss": 0.3104,
+      "step": 1069
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2190484691595909e-05,
+      "loss": 0.3155,
+      "step": 1070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2176973138964794e-05,
+      "loss": 0.3369,
+      "step": 1071
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2163457412993804e-05,
+      "loss": 0.329,
+      "step": 1072
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.21499375395931e-05,
+      "loss": 0.3344,
+      "step": 1073
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2136413544680791e-05,
+      "loss": 0.3474,
+      "step": 1074
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2122885454182884e-05,
+      "loss": 0.3198,
+      "step": 1075
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2109353294033248e-05,
+      "loss": 0.3387,
+      "step": 1076
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2095817090173535e-05,
+      "loss": 0.3126,
+      "step": 1077
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2082276868553167e-05,
+      "loss": 0.3131,
+      "step": 1078
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2068732655129259e-05,
+      "loss": 0.3448,
+      "step": 1079
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.205518447586658e-05,
+      "loss": 0.3351,
+      "step": 1080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2041632356737503e-05,
+      "loss": 0.3598,
+      "step": 1081
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2028076323721954e-05,
+      "loss": 0.3236,
+      "step": 1082
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2014516402807362e-05,
+      "loss": 0.3115,
+      "step": 1083
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2000952619988604e-05,
+      "loss": 0.2967,
+      "step": 1084
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.198738500126797e-05,
+      "loss": 0.3396,
+      "step": 1085
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1973813572655095e-05,
+      "loss": 0.3698,
+      "step": 1086
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1960238360166925e-05,
+      "loss": 0.3152,
+      "step": 1087
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.194665938982765e-05,
+      "loss": 0.3693,
+      "step": 1088
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1933076687668674e-05,
+      "loss": 0.311,
+      "step": 1089
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1919490279728549e-05,
+      "loss": 0.3192,
+      "step": 1090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1905900192052938e-05,
+      "loss": 0.338,
+      "step": 1091
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1892306450694547e-05,
+      "loss": 0.3065,
+      "step": 1092
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1878709081713094e-05,
+      "loss": 0.3378,
+      "step": 1093
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1865108111175252e-05,
+      "loss": 0.3393,
+      "step": 1094
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1851503565154594e-05,
+      "loss": 0.338,
+      "step": 1095
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.183789546973155e-05,
+      "loss": 0.3482,
+      "step": 1096
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1824283850993356e-05,
+      "loss": 0.3289,
+      "step": 1097
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1810668735033998e-05,
+      "loss": 0.3205,
+      "step": 1098
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1797050147954166e-05,
+      "loss": 0.3067,
+      "step": 1099
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1783428115861208e-05,
+      "loss": 0.3108,
+      "step": 1100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.176980266486908e-05,
+      "loss": 0.31,
+      "step": 1101
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1756173821098285e-05,
+      "loss": 0.2955,
+      "step": 1102
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1742541610675825e-05,
+      "loss": 0.3315,
+      "step": 1103
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1728906059735173e-05,
+      "loss": 0.327,
+      "step": 1104
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1715267194416188e-05,
+      "loss": 0.3152,
+      "step": 1105
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1701625040865096e-05,
+      "loss": 0.3366,
+      "step": 1106
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1687979625234422e-05,
+      "loss": 0.3258,
+      "step": 1107
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.167433097368294e-05,
+      "loss": 0.3471,
+      "step": 1108
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1660679112375632e-05,
+      "loss": 0.3384,
+      "step": 1109
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1647024067483636e-05,
+      "loss": 0.3179,
+      "step": 1110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1633365865184188e-05,
+      "loss": 0.3297,
+      "step": 1111
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1619704531660581e-05,
+      "loss": 0.3157,
+      "step": 1112
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.16060400931021e-05,
+      "loss": 0.3312,
+      "step": 1113
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1592372575703999e-05,
+      "loss": 0.3162,
+      "step": 1114
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1578702005667423e-05,
+      "loss": 0.2884,
+      "step": 1115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.156502840919937e-05,
+      "loss": 0.3217,
+      "step": 1116
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1551351812512648e-05,
+      "loss": 0.3195,
+      "step": 1117
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1537672241825804e-05,
+      "loss": 0.317,
+      "step": 1118
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1523989723363094e-05,
+      "loss": 0.3492,
+      "step": 1119
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1510304283354425e-05,
+      "loss": 0.323,
+      "step": 1120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1496615948035303e-05,
+      "loss": 0.3093,
+      "step": 1121
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1482924743646784e-05,
+      "loss": 0.3539,
+      "step": 1122
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.146923069643543e-05,
+      "loss": 0.3322,
+      "step": 1123
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1455533832653236e-05,
+      "loss": 0.3233,
+      "step": 1124
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1441834178557617e-05,
+      "loss": 0.2943,
+      "step": 1125
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1428131760411326e-05,
+      "loss": 0.3081,
+      "step": 1126
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.141442660448242e-05,
+      "loss": 0.3443,
+      "step": 1127
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1400718737044196e-05,
+      "loss": 0.3637,
+      "step": 1128
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1387008184375157e-05,
+      "loss": 0.3267,
+      "step": 1129
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1373294972758953e-05,
+      "loss": 0.3015,
+      "step": 1130
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1359579128484327e-05,
+      "loss": 0.3214,
+      "step": 1131
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.134586067784507e-05,
+      "loss": 0.3164,
+      "step": 1132
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1332139647139976e-05,
+      "loss": 0.3304,
+      "step": 1133
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.131841606267278e-05,
+      "loss": 0.3444,
+      "step": 1134
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1304689950752101e-05,
+      "loss": 0.3146,
+      "step": 1135
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1290961337691429e-05,
+      "loss": 0.3242,
+      "step": 1136
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1277230249809028e-05,
+      "loss": 0.3173,
+      "step": 1137
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1263496713427919e-05,
+      "loss": 0.3308,
+      "step": 1138
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1249760754875802e-05,
+      "loss": 0.3153,
+      "step": 1139
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1236022400485034e-05,
+      "loss": 0.3401,
+      "step": 1140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.122228167659256e-05,
+      "loss": 0.3259,
+      "step": 1141
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1208538609539866e-05,
+      "loss": 0.3124,
+      "step": 1142
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1194793225672936e-05,
+      "loss": 0.3253,
+      "step": 1143
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1181045551342182e-05,
+      "loss": 0.3129,
+      "step": 1144
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1167295612902424e-05,
+      "loss": 0.3304,
+      "step": 1145
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1153543436712813e-05,
+      "loss": 0.3536,
+      "step": 1146
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1139789049136787e-05,
+      "loss": 0.3385,
+      "step": 1147
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1126032476542035e-05,
+      "loss": 0.3106,
+      "step": 1148
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1112273745300417e-05,
+      "loss": 0.3091,
+      "step": 1149
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1098512881787951e-05,
+      "loss": 0.318,
+      "step": 1150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1084749912384725e-05,
+      "loss": 0.308,
+      "step": 1151
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1070984863474876e-05,
+      "loss": 0.3331,
+      "step": 1152
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1057217761446524e-05,
+      "loss": 0.3059,
+      "step": 1153
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1043448632691722e-05,
+      "loss": 0.3163,
+      "step": 1154
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1029677503606414e-05,
+      "loss": 0.2938,
+      "step": 1155
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1015904400590377e-05,
+      "loss": 0.3057,
+      "step": 1156
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1002129350047162e-05,
+      "loss": 0.3028,
+      "step": 1157
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0988352378384073e-05,
+      "loss": 0.3241,
+      "step": 1158
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0974573512012077e-05,
+      "loss": 0.3332,
+      "step": 1159
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0960792777345785e-05,
+      "loss": 0.3351,
+      "step": 1160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.094701020080339e-05,
+      "loss": 0.3201,
+      "step": 1161
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0933225808806605e-05,
+      "loss": 0.2938,
+      "step": 1162
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0919439627780642e-05,
+      "loss": 0.3044,
+      "step": 1163
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0905651684154121e-05,
+      "loss": 0.3108,
+      "step": 1164
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.089186200435906e-05,
+      "loss": 0.2773,
+      "step": 1165
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.087807061483079e-05,
+      "loss": 0.3314,
+      "step": 1166
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0864277542007932e-05,
+      "loss": 0.3187,
+      "step": 1167
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0850482812332326e-05,
+      "loss": 0.2818,
+      "step": 1168
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0836686452248992e-05,
+      "loss": 0.3181,
+      "step": 1169
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0822888488206074e-05,
+      "loss": 0.3152,
+      "step": 1170
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0809088946654784e-05,
+      "loss": 0.3393,
+      "step": 1171
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0795287854049376e-05,
+      "loss": 0.3162,
+      "step": 1172
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0781485236847062e-05,
+      "loss": 0.2974,
+      "step": 1173
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0767681121507982e-05,
+      "loss": 0.3033,
+      "step": 1174
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0753875534495146e-05,
+      "loss": 0.2839,
+      "step": 1175
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0740068502274388e-05,
+      "loss": 0.2964,
+      "step": 1176
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.072626005131431e-05,
+      "loss": 0.3387,
+      "step": 1177
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0712450208086238e-05,
+      "loss": 0.3077,
+      "step": 1178
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0698638999064162e-05,
+      "loss": 0.3063,
+      "step": 1179
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0684826450724687e-05,
+      "loss": 0.327,
+      "step": 1180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0671012589547e-05,
+      "loss": 0.2572,
+      "step": 1181
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0657197442012789e-05,
+      "loss": 0.312,
+      "step": 1182
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0643381034606217e-05,
+      "loss": 0.3031,
+      "step": 1183
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0629563393813862e-05,
+      "loss": 0.296,
+      "step": 1184
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0615744546124658e-05,
+      "loss": 0.3188,
+      "step": 1185
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0601924518029865e-05,
+      "loss": 0.3126,
+      "step": 1186
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0588103336022994e-05,
+      "loss": 0.3167,
+      "step": 1187
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.057428102659978e-05,
+      "loss": 0.3122,
+      "step": 1188
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.056045761625811e-05,
+      "loss": 0.3561,
+      "step": 1189
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0546633131497986e-05,
+      "loss": 0.3051,
+      "step": 1190
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0532807598821464e-05,
+      "loss": 0.2877,
+      "step": 1191
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0518981044732616e-05,
+      "loss": 0.294,
+      "step": 1192
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0505153495737473e-05,
+      "loss": 0.3018,
+      "step": 1193
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0491324978343967e-05,
+      "loss": 0.2956,
+      "step": 1194
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0477495519061884e-05,
+      "loss": 0.2788,
+      "step": 1195
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0463665144402828e-05,
+      "loss": 0.2978,
+      "step": 1196
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0449833880880146e-05,
+      "loss": 0.3345,
+      "step": 1197
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0436001755008893e-05,
+      "loss": 0.2776,
+      "step": 1198
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0422168793305785e-05,
+      "loss": 0.3416,
+      "step": 1199
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0408335022289124e-05,
+      "loss": 0.2893,
+      "step": 1200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0394500468478777e-05,
+      "loss": 0.3262,
+      "step": 1201
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0380665158396106e-05,
+      "loss": 0.3148,
+      "step": 1202
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0366829118563924e-05,
+      "loss": 0.3414,
+      "step": 1203
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0352992375506442e-05,
+      "loss": 0.3146,
+      "step": 1204
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0339154955749223e-05,
+      "loss": 0.3313,
+      "step": 1205
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.032531688581912e-05,
+      "loss": 0.3221,
+      "step": 1206
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0311478192244235e-05,
+      "loss": 0.3183,
+      "step": 1207
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0297638901553875e-05,
+      "loss": 0.3174,
+      "step": 1208
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0283799040278482e-05,
+      "loss": 0.3018,
+      "step": 1209
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0269958634949589e-05,
+      "loss": 0.3189,
+      "step": 1210
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0256117712099779e-05,
+      "loss": 0.291,
+      "step": 1211
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0242276298262621e-05,
+      "loss": 0.3383,
+      "step": 1212
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0228434419972637e-05,
+      "loss": 0.2766,
+      "step": 1213
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0214592103765226e-05,
+      "loss": 0.2896,
+      "step": 1214
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0200749376176635e-05,
+      "loss": 0.3022,
+      "step": 1215
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.018690626374389e-05,
+      "loss": 0.3104,
+      "step": 1216
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0173062793004768e-05,
+      "loss": 0.3131,
+      "step": 1217
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0159218990497723e-05,
+      "loss": 0.3255,
+      "step": 1218
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.014537488276185e-05,
+      "loss": 0.3113,
+      "step": 1219
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.013153049633683e-05,
+      "loss": 0.3112,
+      "step": 1220
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.011768585776287e-05,
+      "loss": 0.2566,
+      "step": 1221
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0103840993580666e-05,
+      "loss": 0.3122,
+      "step": 1222
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0089995930331351e-05,
+      "loss": 0.3178,
+      "step": 1223
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.007615069455643e-05,
+      "loss": 0.3263,
+      "step": 1224
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0062305312797748e-05,
+      "loss": 0.2733,
+      "step": 1225
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0048459811597424e-05,
+      "loss": 0.2858,
+      "step": 1226
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0034614217497807e-05,
+      "loss": 0.3209,
+      "step": 1227
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0020768557041424e-05,
+      "loss": 0.3106,
+      "step": 1228
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0006922856770935e-05,
+      "loss": 0.3052,
+      "step": 1229
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.993077143229068e-06,
+      "loss": 0.3105,
+      "step": 1230
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.979231442958577e-06,
+      "loss": 0.3245,
+      "step": 1231
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.965385782502196e-06,
+      "loss": 0.2869,
+      "step": 1232
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.951540188402578e-06,
+      "loss": 0.3284,
+      "step": 1233
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.937694687202255e-06,
+      "loss": 0.284,
+      "step": 1234
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.923849305443573e-06,
+      "loss": 0.2961,
+      "step": 1235
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.910004069668656e-06,
+      "loss": 0.293,
+      "step": 1236
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.896159006419339e-06,
+      "loss": 0.2904,
+      "step": 1237
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.882314142237135e-06,
+      "loss": 0.2788,
+      "step": 1238
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.868469503663172e-06,
+      "loss": 0.2894,
+      "step": 1239
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.85462511723815e-06,
+      "loss": 0.2887,
+      "step": 1240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.840781009502275e-06,
+      "loss": 0.3267,
+      "step": 1241
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.826937206995234e-06,
+      "loss": 0.2983,
+      "step": 1242
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.81309373625611e-06,
+      "loss": 0.2922,
+      "step": 1243
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.799250623823368e-06,
+      "loss": 0.2934,
+      "step": 1244
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.785407896234777e-06,
+      "loss": 0.3058,
+      "step": 1245
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.771565580027365e-06,
+      "loss": 0.2903,
+      "step": 1246
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.75772370173738e-06,
+      "loss": 0.2692,
+      "step": 1247
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.743882287900226e-06,
+      "loss": 0.2861,
+      "step": 1248
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.730041365050415e-06,
+      "loss": 0.294,
+      "step": 1249
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.716200959721521e-06,
+      "loss": 0.3361,
+      "step": 1250
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.702361098446127e-06,
+      "loss": 0.294,
+      "step": 1251
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.688521807755767e-06,
+      "loss": 0.283,
+      "step": 1252
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.674683114180886e-06,
+      "loss": 0.2852,
+      "step": 1253
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.660845044250782e-06,
+      "loss": 0.2881,
+      "step": 1254
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.647007624493562e-06,
+      "loss": 0.3194,
+      "step": 1255
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.633170881436079e-06,
+      "loss": 0.3033,
+      "step": 1256
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.619334841603894e-06,
+      "loss": 0.2816,
+      "step": 1257
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.605499531521223e-06,
+      "loss": 0.266,
+      "step": 1258
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.591664977710876e-06,
+      "loss": 0.2877,
+      "step": 1259
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.577831206694218e-06,
+      "loss": 0.2799,
+      "step": 1260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.563998244991107e-06,
+      "loss": 0.2724,
+      "step": 1261
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.550166119119855e-06,
+      "loss": 0.3202,
+      "step": 1262
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.536334855597174e-06,
+      "loss": 0.3101,
+      "step": 1263
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.522504480938118e-06,
+      "loss": 0.3033,
+      "step": 1264
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.508675021656036e-06,
+      "loss": 0.2959,
+      "step": 1265
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.49484650426253e-06,
+      "loss": 0.2689,
+      "step": 1266
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.481018955267385e-06,
+      "loss": 0.2992,
+      "step": 1267
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.46719240117854e-06,
+      "loss": 0.3087,
+      "step": 1268
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.453366868502019e-06,
+      "loss": 0.2759,
+      "step": 1269
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.439542383741893e-06,
+      "loss": 0.3352,
+      "step": 1270
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.425718973400222e-06,
+      "loss": 0.2985,
+      "step": 1271
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.41189666397701e-06,
+      "loss": 0.3087,
+      "step": 1272
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.39807548197014e-06,
+      "loss": 0.2899,
+      "step": 1273
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.384255453875347e-06,
+      "loss": 0.2953,
+      "step": 1274
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.370436606186142e-06,
+      "loss": 0.244,
+      "step": 1275
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.356618965393784e-06,
+      "loss": 0.2746,
+      "step": 1276
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.342802557987213e-06,
+      "loss": 0.2728,
+      "step": 1277
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.328987410453002e-06,
+      "loss": 0.2807,
+      "step": 1278
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.315173549275314e-06,
+      "loss": 0.3153,
+      "step": 1279
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.301361000935841e-06,
+      "loss": 0.2829,
+      "step": 1280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.287549791913765e-06,
+      "loss": 0.2759,
+      "step": 1281
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.273739948685692e-06,
+      "loss": 0.3151,
+      "step": 1282
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.259931497725614e-06,
+      "loss": 0.3068,
+      "step": 1283
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.246124465504856e-06,
+      "loss": 0.3266,
+      "step": 1284
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.232318878492021e-06,
+      "loss": 0.3226,
+      "step": 1285
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.21851476315294e-06,
+      "loss": 0.2933,
+      "step": 1286
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.204712145950627e-06,
+      "loss": 0.264,
+      "step": 1287
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.190911053345219e-06,
+      "loss": 0.3179,
+      "step": 1288
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.177111511793933e-06,
+      "loss": 0.283,
+      "step": 1289
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.163313547751011e-06,
+      "loss": 0.2714,
+      "step": 1290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.149517187667675e-06,
+      "loss": 0.2759,
+      "step": 1291
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.135722457992073e-06,
+      "loss": 0.2518,
+      "step": 1292
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.12192938516921e-06,
+      "loss": 0.302,
+      "step": 1293
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.10813799564094e-06,
+      "loss": 0.2389,
+      "step": 1294
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.094348315845877e-06,
+      "loss": 0.2987,
+      "step": 1295
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.080560372219362e-06,
+      "loss": 0.3056,
+      "step": 1296
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.066774191193397e-06,
+      "loss": 0.3008,
+      "step": 1297
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.052989799196614e-06,
+      "loss": 0.3073,
+      "step": 1298
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.039207222654219e-06,
+      "loss": 0.299,
+      "step": 1299
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.025426487987927e-06,
+      "loss": 0.2827,
+      "step": 1300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.01164762161593e-06,
+      "loss": 0.3104,
+      "step": 1301
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.997870649952841e-06,
+      "loss": 0.2895,
+      "step": 1302
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.984095599409628e-06,
+      "loss": 0.2618,
+      "step": 1303
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.970322496393587e-06,
+      "loss": 0.2956,
+      "step": 1304
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.95655136730828e-06,
+      "loss": 0.2901,
+      "step": 1305
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.94278223855348e-06,
+      "loss": 0.2522,
+      "step": 1306
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.92901513652513e-06,
+      "loss": 0.2939,
+      "step": 1307
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.91525008761528e-06,
+      "loss": 0.2932,
+      "step": 1308
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.901487118212056e-06,
+      "loss": 0.3042,
+      "step": 1309
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.887726254699588e-06,
+      "loss": 0.2852,
+      "step": 1310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.873967523457969e-06,
+      "loss": 0.2751,
+      "step": 1311
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.860210950863213e-06,
+      "loss": 0.2809,
+      "step": 1312
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.846456563287188e-06,
+      "loss": 0.3014,
+      "step": 1313
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.832704387097577e-06,
+      "loss": 0.2875,
+      "step": 1314
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.818954448657818e-06,
+      "loss": 0.2886,
+      "step": 1315
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.805206774327068e-06,
+      "loss": 0.268,
+      "step": 1316
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.791461390460136e-06,
+      "loss": 0.2928,
+      "step": 1317
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.777718323407444e-06,
+      "loss": 0.3265,
+      "step": 1318
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.76397759951497e-06,
+      "loss": 0.2691,
+      "step": 1319
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.750239245124201e-06,
+      "loss": 0.3206,
+      "step": 1320
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.736503286572085e-06,
+      "loss": 0.2785,
+      "step": 1321
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.722769750190973e-06,
+      "loss": 0.3086,
+      "step": 1322
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.709038662308573e-06,
+      "loss": 0.2855,
+      "step": 1323
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.695310049247902e-06,
+      "loss": 0.2952,
+      "step": 1324
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.681583937327227e-06,
+      "loss": 0.3055,
+      "step": 1325
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.667860352860027e-06,
+      "loss": 0.2867,
+      "step": 1326
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.654139322154932e-06,
+      "loss": 0.3211,
+      "step": 1327
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.640420871515678e-06,
+      "loss": 0.2798,
+      "step": 1328
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.626705027241047e-06,
+      "loss": 0.2884,
+      "step": 1329
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.612991815624843e-06,
+      "loss": 0.2896,
+      "step": 1330
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.599281262955804e-06,
+      "loss": 0.2909,
+      "step": 1331
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.585573395517583e-06,
+      "loss": 0.2878,
+      "step": 1332
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.571868239588674e-06,
+      "loss": 0.2618,
+      "step": 1333
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.558165821442384e-06,
+      "loss": 0.2772,
+      "step": 1334
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.544466167346766e-06,
+      "loss": 0.3189,
+      "step": 1335
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.530769303564575e-06,
+      "loss": 0.3092,
+      "step": 1336
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.517075256353217e-06,
+      "loss": 0.3052,
+      "step": 1337
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.503384051964699e-06,
+      "loss": 0.2931,
+      "step": 1338
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.489695716645577e-06,
+      "loss": 0.2664,
+      "step": 1339
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.476010276636909e-06,
+      "loss": 0.3036,
+      "step": 1340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.4623277581742e-06,
+      "loss": 0.268,
+      "step": 1341
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.448648187487357e-06,
+      "loss": 0.3136,
+      "step": 1342
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.434971590800634e-06,
+      "loss": 0.2725,
+      "step": 1343
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.421297994332582e-06,
+      "loss": 0.2861,
+      "step": 1344
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.407627424296006e-06,
+      "loss": 0.2874,
+      "step": 1345
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.393959906897905e-06,
+      "loss": 0.2892,
+      "step": 1346
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.380295468339424e-06,
+      "loss": 0.2969,
+      "step": 1347
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.366634134815814e-06,
+      "loss": 0.268,
+      "step": 1348
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.352975932516364e-06,
+      "loss": 0.2437,
+      "step": 1349
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.339320887624367e-06,
+      "loss": 0.3051,
+      "step": 1350
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.32566902631706e-06,
+      "loss": 0.2835,
+      "step": 1351
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.312020374765581e-06,
+      "loss": 0.2896,
+      "step": 1352
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.298374959134906e-06,
+      "loss": 0.3141,
+      "step": 1353
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.284732805583813e-06,
+      "loss": 0.2669,
+      "step": 1354
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.271093940264829e-06,
+      "loss": 0.2699,
+      "step": 1355
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.257458389324176e-06,
+      "loss": 0.2968,
+      "step": 1356
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.243826178901719e-06,
+      "loss": 0.2935,
+      "step": 1357
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.230197335130923e-06,
+      "loss": 0.2914,
+      "step": 1358
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.216571884138793e-06,
+      "loss": 0.2652,
+      "step": 1359
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.202949852045839e-06,
+      "loss": 0.2659,
+      "step": 1360
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.189331264966007e-06,
+      "loss": 0.2728,
+      "step": 1361
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.175716149006649e-06,
+      "loss": 0.2451,
+      "step": 1362
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.162104530268453e-06,
+      "loss": 0.2697,
+      "step": 1363
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.148496434845411e-06,
+      "loss": 0.3031,
+      "step": 1364
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.13489188882475e-06,
+      "loss": 0.2989,
+      "step": 1365
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.121290918286908e-06,
+      "loss": 0.3188,
+      "step": 1366
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.107693549305457e-06,
+      "loss": 0.2822,
+      "step": 1367
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.094099807947066e-06,
+      "loss": 0.2897,
+      "step": 1368
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.080509720271453e-06,
+      "loss": 0.2462,
+      "step": 1369
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.066923312331328e-06,
+      "loss": 0.2627,
+      "step": 1370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.053340610172353e-06,
+      "loss": 0.2731,
+      "step": 1371
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.039761639833078e-06,
+      "loss": 0.2732,
+      "step": 1372
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.026186427344908e-06,
+      "loss": 0.2845,
+      "step": 1373
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.012614998732034e-06,
+      "loss": 0.2993,
+      "step": 1374
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.999047380011399e-06,
+      "loss": 0.2862,
+      "step": 1375
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.985483597192643e-06,
+      "loss": 0.2402,
+      "step": 1376
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.971923676278048e-06,
+      "loss": 0.2471,
+      "step": 1377
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.958367643262498e-06,
+      "loss": 0.2661,
+      "step": 1378
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.944815524133425e-06,
+      "loss": 0.3065,
+      "step": 1379
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.931267344870746e-06,
+      "loss": 0.3161,
+      "step": 1380
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.917723131446838e-06,
+      "loss": 0.2857,
+      "step": 1381
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.90418290982647e-06,
+      "loss": 0.2989,
+      "step": 1382
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.890646705966755e-06,
+      "loss": 0.2914,
+      "step": 1383
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.877114545817116e-06,
+      "loss": 0.277,
+      "step": 1384
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.86358645531921e-06,
+      "loss": 0.2777,
+      "step": 1385
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.8500624604069e-06,
+      "loss": 0.2674,
+      "step": 1386
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.836542587006196e-06,
+      "loss": 0.2834,
+      "step": 1387
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.823026861035209e-06,
+      "loss": 0.2752,
+      "step": 1388
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.809515308404093e-06,
+      "loss": 0.2824,
+      "step": 1389
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.796007955015007e-06,
+      "loss": 0.265,
+      "step": 1390
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.78250482676206e-06,
+      "loss": 0.2728,
+      "step": 1391
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.769005949531257e-06,
+      "loss": 0.2964,
+      "step": 1392
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.75551134920046e-06,
+      "loss": 0.2338,
+      "step": 1393
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.742021051639327e-06,
+      "loss": 0.2889,
+      "step": 1394
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.728535082709267e-06,
+      "loss": 0.3134,
+      "step": 1395
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.715053468263393e-06,
+      "loss": 0.2679,
+      "step": 1396
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.70157623414647e-06,
+      "loss": 0.2907,
+      "step": 1397
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.688103406194868e-06,
+      "loss": 0.2883,
+      "step": 1398
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.674635010236507e-06,
+      "loss": 0.2932,
+      "step": 1399
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.661171072090814e-06,
+      "loss": 0.2693,
+      "step": 1400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.647711617568657e-06,
+      "loss": 0.2725,
+      "step": 1401
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.634256672472334e-06,
+      "loss": 0.2793,
+      "step": 1402
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.620806262595477e-06,
+      "loss": 0.2879,
+      "step": 1403
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.6073604137230295e-06,
+      "loss": 0.2774,
+      "step": 1404
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.593919151631194e-06,
+      "loss": 0.2558,
+      "step": 1405
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.580482502087378e-06,
+      "loss": 0.2746,
+      "step": 1406
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.56705049085015e-06,
+      "loss": 0.2786,
+      "step": 1407
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.553623143669179e-06,
+      "loss": 0.2386,
+      "step": 1408
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.540200486285202e-06,
+      "loss": 0.2766,
+      "step": 1409
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.526782544429958e-06,
+      "loss": 0.249,
+      "step": 1410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.513369343826149e-06,
+      "loss": 0.2639,
+      "step": 1411
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.499960910187389e-06,
+      "loss": 0.2714,
+      "step": 1412
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.486557269218153e-06,
+      "loss": 0.2607,
+      "step": 1413
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.473158446613729e-06,
+      "loss": 0.2984,
+      "step": 1414
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.459764468060167e-06,
+      "loss": 0.3009,
+      "step": 1415
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.446375359234225e-06,
+      "loss": 0.2288,
+      "step": 1416
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.432991145803338e-06,
+      "loss": 0.2883,
+      "step": 1417
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.419611853425546e-06,
+      "loss": 0.2633,
+      "step": 1418
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.406237507749457e-06,
+      "loss": 0.2733,
+      "step": 1419
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3928681344142014e-06,
+      "loss": 0.3012,
+      "step": 1420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3795037590493736e-06,
+      "loss": 0.2782,
+      "step": 1421
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.366144407274986e-06,
+      "loss": 0.3001,
+      "step": 1422
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.352790104701422e-06,
+      "loss": 0.2699,
+      "step": 1423
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.339440876929385e-06,
+      "loss": 0.237,
+      "step": 1424
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.3260967495498455e-06,
+      "loss": 0.2958,
+      "step": 1425
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.312757748144006e-06,
+      "loss": 0.3019,
+      "step": 1426
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2994238982832334e-06,
+      "loss": 0.2945,
+      "step": 1427
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.286095225529024e-06,
+      "loss": 0.3176,
+      "step": 1428
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.272771755432946e-06,
+      "loss": 0.2845,
+      "step": 1429
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2594535135366e-06,
+      "loss": 0.276,
+      "step": 1430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.246140525371556e-06,
+      "loss": 0.2757,
+      "step": 1431
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.232832816459316e-06,
+      "loss": 0.2551,
+      "step": 1432
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.219530412311262e-06,
+      "loss": 0.2485,
+      "step": 1433
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.206233338428606e-06,
+      "loss": 0.2665,
+      "step": 1434
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.192941620302342e-06,
+      "loss": 0.2832,
+      "step": 1435
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1796552834131985e-06,
+      "loss": 0.2766,
+      "step": 1436
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.166374353231578e-06,
+      "loss": 0.2749,
+      "step": 1437
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.153098855217535e-06,
+      "loss": 0.2819,
+      "step": 1438
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.139828814820702e-06,
+      "loss": 0.2818,
+      "step": 1439
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.126564257480243e-06,
+      "loss": 0.2609,
+      "step": 1440
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.113305208624818e-06,
+      "loss": 0.2598,
+      "step": 1441
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.100051693672525e-06,
+      "loss": 0.311,
+      "step": 1442
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.086803738030856e-06,
+      "loss": 0.2847,
+      "step": 1443
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.073561367096639e-06,
+      "loss": 0.246,
+      "step": 1444
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.060324606256009e-06,
+      "loss": 0.2949,
+      "step": 1445
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.047093480884325e-06,
+      "loss": 0.2702,
+      "step": 1446
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.033868016346162e-06,
+      "loss": 0.2609,
+      "step": 1447
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.020648237995235e-06,
+      "loss": 0.2674,
+      "step": 1448
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.007434171174359e-06,
+      "loss": 0.2849,
+      "step": 1449
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.994225841215399e-06,
+      "loss": 0.2457,
+      "step": 1450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.981023273439227e-06,
+      "loss": 0.2658,
+      "step": 1451
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.967826493155661e-06,
+      "loss": 0.2598,
+      "step": 1452
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.954635525663429e-06,
+      "loss": 0.273,
+      "step": 1453
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.941450396250116e-06,
+      "loss": 0.2994,
+      "step": 1454
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.92827113019211e-06,
+      "loss": 0.2724,
+      "step": 1455
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.915097752754567e-06,
+      "loss": 0.2943,
+      "step": 1456
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.90193028919135e-06,
+      "loss": 0.288,
+      "step": 1457
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.888768764744984e-06,
+      "loss": 0.2632,
+      "step": 1458
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.875613204646607e-06,
+      "loss": 0.2687,
+      "step": 1459
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8624636341159325e-06,
+      "loss": 0.3062,
+      "step": 1460
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.849320078361175e-06,
+      "loss": 0.2798,
+      "step": 1461
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8361825625790325e-06,
+      "loss": 0.2517,
+      "step": 1462
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8230511119546215e-06,
+      "loss": 0.2713,
+      "step": 1463
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.809925751661428e-06,
+      "loss": 0.2747,
+      "step": 1464
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.796806506861264e-06,
+      "loss": 0.2577,
+      "step": 1465
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.783693402704221e-06,
+      "loss": 0.2754,
+      "step": 1466
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7705864643286116e-06,
+      "loss": 0.2477,
+      "step": 1467
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.757485716860934e-06,
+      "loss": 0.2887,
+      "step": 1468
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.744391185415818e-06,
+      "loss": 0.259,
+      "step": 1469
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7313028950959725e-06,
+      "loss": 0.2825,
+      "step": 1470
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.718220870992149e-06,
+      "loss": 0.247,
+      "step": 1471
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.705145138183085e-06,
+      "loss": 0.2669,
+      "step": 1472
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.692075721735443e-06,
+      "loss": 0.2434,
+      "step": 1473
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.679012646703804e-06,
+      "loss": 0.2532,
+      "step": 1474
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.66595593813057e-06,
+      "loss": 0.2454,
+      "step": 1475
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.652905621045944e-06,
+      "loss": 0.2992,
+      "step": 1476
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.63986172046788e-06,
+      "loss": 0.276,
+      "step": 1477
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.626824261402027e-06,
+      "loss": 0.253,
+      "step": 1478
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.61379326884169e-06,
+      "loss": 0.267,
+      "step": 1479
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.600768767767771e-06,
+      "loss": 0.262,
+      "step": 1480
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.587750783148739e-06,
+      "loss": 0.2594,
+      "step": 1481
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.574739339940553e-06,
+      "loss": 0.2629,
+      "step": 1482
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.561734463086647e-06,
+      "loss": 0.2594,
+      "step": 1483
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.548736177517861e-06,
+      "loss": 0.2834,
+      "step": 1484
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.535744508152401e-06,
+      "loss": 0.2651,
+      "step": 1485
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.522759479895788e-06,
+      "loss": 0.2725,
+      "step": 1486
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.509781117640817e-06,
+      "loss": 0.2603,
+      "step": 1487
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.49680944626749e-06,
+      "loss": 0.2671,
+      "step": 1488
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.483844490642999e-06,
+      "loss": 0.2584,
+      "step": 1489
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.470886275621653e-06,
+      "loss": 0.2782,
+      "step": 1490
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.457934826044837e-06,
+      "loss": 0.2708,
+      "step": 1491
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.444990166740976e-06,
+      "loss": 0.2622,
+      "step": 1492
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.432052322525468e-06,
+      "loss": 0.2417,
+      "step": 1493
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.41912131820065e-06,
+      "loss": 0.2783,
+      "step": 1494
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.406197178555747e-06,
+      "loss": 0.2924,
+      "step": 1495
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3932799283668266e-06,
+      "loss": 0.2835,
+      "step": 1496
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.38036959239674e-06,
+      "loss": 0.3057,
+      "step": 1497
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.367466195395094e-06,
+      "loss": 0.2879,
+      "step": 1498
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.354569762098188e-06,
+      "loss": 0.2662,
+      "step": 1499
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.341680317228971e-06,
+      "loss": 0.2519,
+      "step": 1500
+    },
+    {
+      "epoch": 1.9,
+      "eval_loss": 0.28214022517204285,
+      "eval_runtime": 51.2818,
+      "eval_samples_per_second": 40.17,
+      "eval_steps_per_second": 0.176,
+      "step": 1500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3287978854969976e-06,
+      "loss": 0.2691,
+      "step": 1501
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.315922491598381e-06,
+      "loss": 0.2732,
+      "step": 1502
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.303054160215731e-06,
+      "loss": 0.2609,
+      "step": 1503
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2901929160181295e-06,
+      "loss": 0.269,
+      "step": 1504
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2773387836610675e-06,
+      "loss": 0.2871,
+      "step": 1505
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.264491787786404e-06,
+      "loss": 0.2589,
+      "step": 1506
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.251651953022317e-06,
+      "loss": 0.2669,
+      "step": 1507
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.238819303983253e-06,
+      "loss": 0.2586,
+      "step": 1508
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.225993865269884e-06,
+      "loss": 0.2414,
+      "step": 1509
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.213175661469066e-06,
+      "loss": 0.2681,
+      "step": 1510
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.200364717153782e-06,
+      "loss": 0.2415,
+      "step": 1511
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.187561056883091e-06,
+      "loss": 0.2774,
+      "step": 1512
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.174764705202098e-06,
+      "loss": 0.2701,
+      "step": 1513
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1619756866418915e-06,
+      "loss": 0.2405,
+      "step": 1514
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.149194025719505e-06,
+      "loss": 0.2562,
+      "step": 1515
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.136419746937865e-06,
+      "loss": 0.2902,
+      "step": 1516
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.12365287478575e-06,
+      "loss": 0.272,
+      "step": 1517
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.110893433737735e-06,
+      "loss": 0.2693,
+      "step": 1518
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.09814144825415e-06,
+      "loss": 0.2497,
+      "step": 1519
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.085396942781034e-06,
+      "loss": 0.2505,
+      "step": 1520
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.072659941750089e-06,
+      "loss": 0.2485,
+      "step": 1521
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0599304695786265e-06,
+      "loss": 0.2533,
+      "step": 1522
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.047208550669523e-06,
+      "loss": 0.2573,
+      "step": 1523
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.034494209411184e-06,
+      "loss": 0.2318,
+      "step": 1524
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.021787470177477e-06,
+      "loss": 0.2745,
+      "step": 1525
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.009088357327706e-06,
+      "loss": 0.2851,
+      "step": 1526
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.996396895206547e-06,
+      "loss": 0.2556,
+      "step": 1527
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.983713108144018e-06,
+      "loss": 0.2491,
+      "step": 1528
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.971037020455415e-06,
+      "loss": 0.2446,
+      "step": 1529
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.95836865644128e-06,
+      "loss": 0.2132,
+      "step": 1530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.945708040387346e-06,
+      "loss": 0.2376,
+      "step": 1531
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.933055196564492e-06,
+      "loss": 0.251,
+      "step": 1532
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.920410149228697e-06,
+      "loss": 0.2593,
+      "step": 1533
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9077729226209965e-06,
+      "loss": 0.2674,
+      "step": 1534
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.895143540967433e-06,
+      "loss": 0.2716,
+      "step": 1535
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.882522028479007e-06,
+      "loss": 0.257,
+      "step": 1536
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.869908409351639e-06,
+      "loss": 0.2523,
+      "step": 1537
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.85730270776611e-06,
+      "loss": 0.2255,
+      "step": 1538
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.844704947888028e-06,
+      "loss": 0.2636,
+      "step": 1539
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8321151538677775e-06,
+      "loss": 0.2591,
+      "step": 1540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.819533349840465e-06,
+      "loss": 0.2773,
+      "step": 1541
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.806959559925891e-06,
+      "loss": 0.2499,
+      "step": 1542
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.79439380822848e-06,
+      "loss": 0.2685,
+      "step": 1543
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.781836118837263e-06,
+      "loss": 0.2768,
+      "step": 1544
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.769286515825794e-06,
+      "loss": 0.2451,
+      "step": 1545
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.756745023252141e-06,
+      "loss": 0.235,
+      "step": 1546
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.744211665158827e-06,
+      "loss": 0.2474,
+      "step": 1547
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7316864655727646e-06,
+      "loss": 0.2761,
+      "step": 1548
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.719169448505242e-06,
+      "loss": 0.2336,
+      "step": 1549
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.706660637951851e-06,
+      "loss": 0.2657,
+      "step": 1550
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.694160057892451e-06,
+      "loss": 0.2461,
+      "step": 1551
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.681667732291134e-06,
+      "loss": 0.2807,
+      "step": 1552
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.669183685096154e-06,
+      "loss": 0.28,
+      "step": 1553
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.656707940239909e-06,
+      "loss": 0.2742,
+      "step": 1554
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.644240521638872e-06,
+      "loss": 0.2804,
+      "step": 1555
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.63178145319355e-06,
+      "loss": 0.2649,
+      "step": 1556
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.619330758788455e-06,
+      "loss": 0.2534,
+      "step": 1557
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.606888462292035e-06,
+      "loss": 0.2534,
+      "step": 1558
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.594454587556651e-06,
+      "loss": 0.2367,
+      "step": 1559
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.582029158418498e-06,
+      "loss": 0.2675,
+      "step": 1560
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5696121986976074e-06,
+      "loss": 0.2277,
+      "step": 1561
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.557203732197755e-06,
+      "loss": 0.2666,
+      "step": 1562
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5448037827064325e-06,
+      "loss": 0.2648,
+      "step": 1563
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.532412373994828e-06,
+      "loss": 0.247,
+      "step": 1564
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.520029529817733e-06,
+      "loss": 0.2514,
+      "step": 1565
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.507655273913524e-06,
+      "loss": 0.2145,
+      "step": 1566
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4952896300041255e-06,
+      "loss": 0.2249,
+      "step": 1567
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.482932621794939e-06,
+      "loss": 0.2644,
+      "step": 1568
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.470584272974822e-06,
+      "loss": 0.2479,
+      "step": 1569
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.458244607216023e-06,
+      "loss": 0.2694,
+      "step": 1570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.445913648174146e-06,
+      "loss": 0.2779,
+      "step": 1571
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4335914194881115e-06,
+      "loss": 0.2655,
+      "step": 1572
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.42127794478009e-06,
+      "loss": 0.2255,
+      "step": 1573
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.408973247655489e-06,
+      "loss": 0.2421,
+      "step": 1574
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.396677351702869e-06,
+      "loss": 0.2622,
+      "step": 1575
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.384390280493935e-06,
+      "loss": 0.2357,
+      "step": 1576
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.372112057583465e-06,
+      "loss": 0.2273,
+      "step": 1577
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.359842706509271e-06,
+      "loss": 0.1969,
+      "step": 1578
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.347582250792174e-06,
+      "loss": 0.2443,
+      "step": 1579
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.33533071393592e-06,
+      "loss": 0.2291,
+      "step": 1580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.323088119427175e-06,
+      "loss": 0.226,
+      "step": 1581
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.3108544907354585e-06,
+      "loss": 0.2141,
+      "step": 1582
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2986298513130925e-06,
+      "loss": 0.2281,
+      "step": 1583
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.286414224595182e-06,
+      "loss": 0.231,
+      "step": 1584
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.27420763399954e-06,
+      "loss": 0.2572,
+      "step": 1585
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.26201010292666e-06,
+      "loss": 0.2394,
+      "step": 1586
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.249821654759677e-06,
+      "loss": 0.2624,
+      "step": 1587
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2376423128642995e-06,
+      "loss": 0.2374,
+      "step": 1588
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2254721005887965e-06,
+      "loss": 0.2642,
+      "step": 1589
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.213311041263915e-06,
+      "loss": 0.2415,
+      "step": 1590
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.201159158202877e-06,
+      "loss": 0.2564,
+      "step": 1591
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.189016474701294e-06,
+      "loss": 0.2214,
+      "step": 1592
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.176883014037148e-06,
+      "loss": 0.2434,
+      "step": 1593
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.164758799470751e-06,
+      "loss": 0.2123,
+      "step": 1594
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.152643854244671e-06,
+      "loss": 0.2317,
+      "step": 1595
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1405382015837244e-06,
+      "loss": 0.2336,
+      "step": 1596
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.128441864694904e-06,
+      "loss": 0.2302,
+      "step": 1597
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.116354866767339e-06,
+      "loss": 0.2217,
+      "step": 1598
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.104277230972268e-06,
+      "loss": 0.2256,
+      "step": 1599
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.092208980462976e-06,
+      "loss": 0.2277,
+      "step": 1600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0801501383747505e-06,
+      "loss": 0.2346,
+      "step": 1601
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.068100727824855e-06,
+      "loss": 0.2361,
+      "step": 1602
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0560607719124585e-06,
+      "loss": 0.2358,
+      "step": 1603
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.044030293718618e-06,
+      "loss": 0.2212,
+      "step": 1604
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03200931630621e-06,
+      "loss": 0.2201,
+      "step": 1605
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.01999786271991e-06,
+      "loss": 0.2391,
+      "step": 1606
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.007995955986125e-06,
+      "loss": 0.2383,
+      "step": 1607
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.996003619112959e-06,
+      "loss": 0.2153,
+      "step": 1608
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.984020875090184e-06,
+      "loss": 0.2417,
+      "step": 1609
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.972047746889168e-06,
+      "loss": 0.2457,
+      "step": 1610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.960084257462856e-06,
+      "loss": 0.224,
+      "step": 1611
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.948130429745707e-06,
+      "loss": 0.2308,
+      "step": 1612
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.936186286653656e-06,
+      "loss": 0.2191,
+      "step": 1613
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.924251851084086e-06,
+      "loss": 0.1776,
+      "step": 1614
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.912327145915755e-06,
+      "loss": 0.2151,
+      "step": 1615
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.90041219400878e-06,
+      "loss": 0.1971,
+      "step": 1616
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.888507018204567e-06,
+      "loss": 0.2481,
+      "step": 1617
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.876611641325793e-06,
+      "loss": 0.213,
+      "step": 1618
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.864726086176348e-06,
+      "loss": 0.2237,
+      "step": 1619
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.852850375541285e-06,
+      "loss": 0.2124,
+      "step": 1620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.840984532186798e-06,
+      "loss": 0.2457,
+      "step": 1621
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.829128578860152e-06,
+      "loss": 0.2009,
+      "step": 1622
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.817282538289656e-06,
+      "loss": 0.2108,
+      "step": 1623
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.805446433184625e-06,
+      "loss": 0.2301,
+      "step": 1624
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7936202862353145e-06,
+      "loss": 0.2292,
+      "step": 1625
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7818041201128996e-06,
+      "loss": 0.2395,
+      "step": 1626
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7699979574694165e-06,
+      "loss": 0.1928,
+      "step": 1627
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.758201820937722e-06,
+      "loss": 0.2328,
+      "step": 1628
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.746415733131463e-06,
+      "loss": 0.2274,
+      "step": 1629
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.734639716645007e-06,
+      "loss": 0.2175,
+      "step": 1630
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.722873794053434e-06,
+      "loss": 0.2479,
+      "step": 1631
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.7111179879124525e-06,
+      "loss": 0.2426,
+      "step": 1632
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.699372320758395e-06,
+      "loss": 0.2345,
+      "step": 1633
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6876368151081475e-06,
+      "loss": 0.2498,
+      "step": 1634
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.675911493459111e-06,
+      "loss": 0.217,
+      "step": 1635
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.664196378289184e-06,
+      "loss": 0.2282,
+      "step": 1636
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.65249149205668e-06,
+      "loss": 0.2346,
+      "step": 1637
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.640796857200303e-06,
+      "loss": 0.2318,
+      "step": 1638
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.629112496139121e-06,
+      "loss": 0.2392,
+      "step": 1639
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.617438431272485e-06,
+      "loss": 0.2057,
+      "step": 1640
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.605774684980029e-06,
+      "loss": 0.2363,
+      "step": 1641
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.594121279621591e-06,
+      "loss": 0.2402,
+      "step": 1642
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.582478237537188e-06,
+      "loss": 0.2266,
+      "step": 1643
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.570845581046977e-06,
+      "loss": 0.2176,
+      "step": 1644
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.559223332451195e-06,
+      "loss": 0.2021,
+      "step": 1645
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.5476115140301395e-06,
+      "loss": 0.1984,
+      "step": 1646
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.536010148044098e-06,
+      "loss": 0.24,
+      "step": 1647
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.5244192567333355e-06,
+      "loss": 0.2344,
+      "step": 1648
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.512838862318026e-06,
+      "loss": 0.2561,
+      "step": 1649
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.501268986998219e-06,
+      "loss": 0.2516,
+      "step": 1650
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.489709652953811e-06,
+      "loss": 0.2178,
+      "step": 1651
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.478160882344476e-06,
+      "loss": 0.2451,
+      "step": 1652
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4666226973096475e-06,
+      "loss": 0.2507,
+      "step": 1653
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.455095119968467e-06,
+      "loss": 0.2442,
+      "step": 1654
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.443578172419727e-06,
+      "loss": 0.2261,
+      "step": 1655
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4320718767418615e-06,
+      "loss": 0.2357,
+      "step": 1656
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4205762549928684e-06,
+      "loss": 0.2406,
+      "step": 1657
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.409091329210289e-06,
+      "loss": 0.2077,
+      "step": 1658
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3976171214111665e-06,
+      "loss": 0.2084,
+      "step": 1659
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.386153653591983e-06,
+      "loss": 0.2558,
+      "step": 1660
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.374700947728647e-06,
+      "loss": 0.2353,
+      "step": 1661
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.363259025776425e-06,
+      "loss": 0.2634,
+      "step": 1662
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.351827909669918e-06,
+      "loss": 0.2318,
+      "step": 1663
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.340407621323006e-06,
+      "loss": 0.2472,
+      "step": 1664
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.328998182628811e-06,
+      "loss": 0.222,
+      "step": 1665
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.317599615459664e-06,
+      "loss": 0.2451,
+      "step": 1666
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.306211941667043e-06,
+      "loss": 0.2339,
+      "step": 1667
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2948351830815585e-06,
+      "loss": 0.2115,
+      "step": 1668
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.283469361512881e-06,
+      "loss": 0.2262,
+      "step": 1669
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.272114498749717e-06,
+      "loss": 0.2266,
+      "step": 1670
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.260770616559774e-06,
+      "loss": 0.2276,
+      "step": 1671
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.249437736689704e-06,
+      "loss": 0.2183,
+      "step": 1672
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.238115880865061e-06,
+      "loss": 0.2447,
+      "step": 1673
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.22680507079028e-06,
+      "loss": 0.2774,
+      "step": 1674
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2155053281486005e-06,
+      "loss": 0.2265,
+      "step": 1675
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.204216674602066e-06,
+      "loss": 0.2066,
+      "step": 1676
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1929391317914454e-06,
+      "loss": 0.2159,
+      "step": 1677
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.181672721336223e-06,
+      "loss": 0.2355,
+      "step": 1678
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.170417464834529e-06,
+      "loss": 0.2277,
+      "step": 1679
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.159173383863111e-06,
+      "loss": 0.2451,
+      "step": 1680
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.147940499977308e-06,
+      "loss": 0.2109,
+      "step": 1681
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.136718834710973e-06,
+      "loss": 0.2594,
+      "step": 1682
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.125508409576473e-06,
+      "loss": 0.2357,
+      "step": 1683
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.114309246064613e-06,
+      "loss": 0.241,
+      "step": 1684
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.103121365644608e-06,
+      "loss": 0.2347,
+      "step": 1685
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.091944789764055e-06,
+      "loss": 0.2207,
+      "step": 1686
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.080779539848868e-06,
+      "loss": 0.2142,
+      "step": 1687
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0696256373032595e-06,
+      "loss": 0.2438,
+      "step": 1688
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.058483103509675e-06,
+      "loss": 0.2332,
+      "step": 1689
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.047351959828778e-06,
+      "loss": 0.2289,
+      "step": 1690
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.036232227599395e-06,
+      "loss": 0.2732,
+      "step": 1691
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0251239281384655e-06,
+      "loss": 0.24,
+      "step": 1692
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.014027082741027e-06,
+      "loss": 0.2135,
+      "step": 1693
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0029417126801505e-06,
+      "loss": 0.2248,
+      "step": 1694
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.991867839206905e-06,
+      "loss": 0.2739,
+      "step": 1695
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9808054835503314e-06,
+      "loss": 0.2344,
+      "step": 1696
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.969754666917378e-06,
+      "loss": 0.2269,
+      "step": 1697
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.958715410492886e-06,
+      "loss": 0.1976,
+      "step": 1698
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.947687735439522e-06,
+      "loss": 0.2287,
+      "step": 1699
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.936671662897753e-06,
+      "loss": 0.2292,
+      "step": 1700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.925667213985816e-06,
+      "loss": 0.2532,
+      "step": 1701
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.914674409799647e-06,
+      "loss": 0.2271,
+      "step": 1702
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.903693271412875e-06,
+      "loss": 0.2238,
+      "step": 1703
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.892723819876751e-06,
+      "loss": 0.1811,
+      "step": 1704
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.881766076220137e-06,
+      "loss": 0.208,
+      "step": 1705
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.870820061449437e-06,
+      "loss": 0.2109,
+      "step": 1706
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.859885796548567e-06,
+      "loss": 0.2093,
+      "step": 1707
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.848963302478943e-06,
+      "loss": 0.2212,
+      "step": 1708
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.838052600179392e-06,
+      "loss": 0.2193,
+      "step": 1709
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8271537105661394e-06,
+      "loss": 0.2605,
+      "step": 1710
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.816266654532775e-06,
+      "loss": 0.2626,
+      "step": 1711
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8053914529501913e-06,
+      "loss": 0.2112,
+      "step": 1712
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.794528126666568e-06,
+      "loss": 0.1951,
+      "step": 1713
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.783676696507308e-06,
+      "loss": 0.2283,
+      "step": 1714
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.772837183275012e-06,
+      "loss": 0.2255,
+      "step": 1715
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7620096077494416e-06,
+      "loss": 0.2438,
+      "step": 1716
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.751193990687464e-06,
+      "loss": 0.2708,
+      "step": 1717
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7403903528230322e-06,
+      "loss": 0.2288,
+      "step": 1718
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.729598714867122e-06,
+      "loss": 0.2299,
+      "step": 1719
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7188190975077175e-06,
+      "loss": 0.2251,
+      "step": 1720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7080515214097513e-06,
+      "loss": 0.2504,
+      "step": 1721
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.69729600721507e-06,
+      "loss": 0.2388,
+      "step": 1722
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.686552575542408e-06,
+      "loss": 0.2109,
+      "step": 1723
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6758212469873234e-06,
+      "loss": 0.2308,
+      "step": 1724
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6651020421221827e-06,
+      "loss": 0.2254,
+      "step": 1725
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6543949814961097e-06,
+      "loss": 0.2084,
+      "step": 1726
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6437000856349403e-06,
+      "loss": 0.184,
+      "step": 1727
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.633017375041198e-06,
+      "loss": 0.2512,
+      "step": 1728
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.622346870194042e-06,
+      "loss": 0.2027,
+      "step": 1729
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6116885915492304e-06,
+      "loss": 0.2236,
+      "step": 1730
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.601042559539093e-06,
+      "loss": 0.2081,
+      "step": 1731
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.590408794572469e-06,
+      "loss": 0.2264,
+      "step": 1732
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.579787317034694e-06,
+      "loss": 0.2092,
+      "step": 1733
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5691781472875377e-06,
+      "loss": 0.2346,
+      "step": 1734
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5585813056691844e-06,
+      "loss": 0.2366,
+      "step": 1735
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.547996812494179e-06,
+      "loss": 0.2391,
+      "step": 1736
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5374246880533914e-06,
+      "loss": 0.2192,
+      "step": 1737
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5268649526139896e-06,
+      "loss": 0.2086,
+      "step": 1738
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.516317626419381e-06,
+      "loss": 0.1823,
+      "step": 1739
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5057827296891946e-06,
+      "loss": 0.23,
+      "step": 1740
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.495260282619224e-06,
+      "loss": 0.2276,
+      "step": 1741
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4847503053813924e-06,
+      "loss": 0.2099,
+      "step": 1742
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.47425281812373e-06,
+      "loss": 0.2222,
+      "step": 1743
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.463767840970319e-06,
+      "loss": 0.2218,
+      "step": 1744
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.45329539402125e-06,
+      "loss": 0.2243,
+      "step": 1745
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4428354973526078e-06,
+      "loss": 0.2356,
+      "step": 1746
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4323881710164044e-06,
+      "loss": 0.1879,
+      "step": 1747
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4219534350405638e-06,
+      "loss": 0.2246,
+      "step": 1748
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4115313094288636e-06,
+      "loss": 0.2232,
+      "step": 1749
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4011218141609215e-06,
+      "loss": 0.2279,
+      "step": 1750
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3907249691921285e-06,
+      "loss": 0.2467,
+      "step": 1751
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3803407944536272e-06,
+      "loss": 0.2544,
+      "step": 1752
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.36996930985228e-06,
+      "loss": 0.223,
+      "step": 1753
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3596105352706086e-06,
+      "loss": 0.2067,
+      "step": 1754
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3492644905667828e-06,
+      "loss": 0.2174,
+      "step": 1755
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3389311955745574e-06,
+      "loss": 0.2306,
+      "step": 1756
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.328610670103247e-06,
+      "loss": 0.1963,
+      "step": 1757
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3183029339376972e-06,
+      "loss": 0.2152,
+      "step": 1758
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.308008006838219e-06,
+      "loss": 0.2485,
+      "step": 1759
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2977259085405868e-06,
+      "loss": 0.2186,
+      "step": 1760
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.287456658755964e-06,
+      "loss": 0.2001,
+      "step": 1761
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2772002771708946e-06,
+      "loss": 0.2226,
+      "step": 1762
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2669567834472527e-06,
+      "loss": 0.2206,
+      "step": 1763
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2567261972221964e-06,
+      "loss": 0.2338,
+      "step": 1764
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.246508538108155e-06,
+      "loss": 0.214,
+      "step": 1765
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2363038256927637e-06,
+      "loss": 0.1951,
+      "step": 1766
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2261120795388377e-06,
+      "loss": 0.2211,
+      "step": 1767
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2159333191843477e-06,
+      "loss": 0.2164,
+      "step": 1768
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2057675641423557e-06,
+      "loss": 0.2074,
+      "step": 1769
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1956148339010053e-06,
+      "loss": 0.1976,
+      "step": 1770
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1854751479234603e-06,
+      "loss": 0.2174,
+      "step": 1771
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.17534852564788e-06,
+      "loss": 0.2127,
+      "step": 1772
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.16523498648739e-06,
+      "loss": 0.2099,
+      "step": 1773
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.155134549830019e-06,
+      "loss": 0.208,
+      "step": 1774
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.145047235038694e-06,
+      "loss": 0.2055,
+      "step": 1775
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1349730614511743e-06,
+      "loss": 0.1908,
+      "step": 1776
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.124912048380031e-06,
+      "loss": 0.2038,
+      "step": 1777
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1148642151126117e-06,
+      "loss": 0.2296,
+      "step": 1778
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.104829580910983e-06,
+      "loss": 0.2108,
+      "step": 1779
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.094808165011932e-06,
+      "loss": 0.2385,
+      "step": 1780
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0847999866268864e-06,
+      "loss": 0.2384,
+      "step": 1781
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.074805064941898e-06,
+      "loss": 0.2159,
+      "step": 1782
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0648234191176162e-06,
+      "loss": 0.2259,
+      "step": 1783
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0548550682892285e-06,
+      "loss": 0.2043,
+      "step": 1784
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0449000315664466e-06,
+      "loss": 0.22,
+      "step": 1785
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0349583280334504e-06,
+      "loss": 0.2282,
+      "step": 1786
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0250299767488556e-06,
+      "loss": 0.2089,
+      "step": 1787
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.015114996745696e-06,
+      "loss": 0.2293,
+      "step": 1788
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0052134070313564e-06,
+      "loss": 0.2164,
+      "step": 1789
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9953252265875666e-06,
+      "loss": 0.1965,
+      "step": 1790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.985450474370336e-06,
+      "loss": 0.2291,
+      "step": 1791
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.975589169309938e-06,
+      "loss": 0.2165,
+      "step": 1792
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.965741330310874e-06,
+      "loss": 0.2024,
+      "step": 1793
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9559069762518155e-06,
+      "loss": 0.2281,
+      "step": 1794
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9460861259855987e-06,
+      "loss": 0.1998,
+      "step": 1795
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9362787983391593e-06,
+      "loss": 0.2341,
+      "step": 1796
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.926485012113519e-06,
+      "loss": 0.198,
+      "step": 1797
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9167047860837383e-06,
+      "loss": 0.2043,
+      "step": 1798
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.906938138998874e-06,
+      "loss": 0.2396,
+      "step": 1799
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8971850895819663e-06,
+      "loss": 0.2227,
+      "step": 1800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8874456565299746e-06,
+      "loss": 0.2003,
+      "step": 1801
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.87771985851376e-06,
+      "loss": 0.1926,
+      "step": 1802
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.868007714178048e-06,
+      "loss": 0.2088,
+      "step": 1803
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.858309242141385e-06,
+      "loss": 0.2409,
+      "step": 1804
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.848624460996111e-06,
+      "loss": 0.2264,
+      "step": 1805
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8389533893083167e-06,
+      "loss": 0.191,
+      "step": 1806
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8292960456178086e-06,
+      "loss": 0.2295,
+      "step": 1807
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8196524484380873e-06,
+      "loss": 0.2245,
+      "step": 1808
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.810022616256286e-06,
+      "loss": 0.209,
+      "step": 1809
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.800406567533165e-06,
+      "loss": 0.2167,
+      "step": 1810
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7908043207030454e-06,
+      "loss": 0.2274,
+      "step": 1811
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7812158941738052e-06,
+      "loss": 0.2551,
+      "step": 1812
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.771641306326818e-06,
+      "loss": 0.1941,
+      "step": 1813
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7620805755169257e-06,
+      "loss": 0.2335,
+      "step": 1814
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.752533720072417e-06,
+      "loss": 0.2238,
+      "step": 1815
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7430007582949757e-06,
+      "loss": 0.1669,
+      "step": 1816
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7334817084596453e-06,
+      "loss": 0.2304,
+      "step": 1817
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.723976588814811e-06,
+      "loss": 0.1939,
+      "step": 1818
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.714485417582139e-06,
+      "loss": 0.2135,
+      "step": 1819
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7050082129565713e-06,
+      "loss": 0.2065,
+      "step": 1820
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.695544993106265e-06,
+      "loss": 0.2472,
+      "step": 1821
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.686095776172566e-06,
+      "loss": 0.2108,
+      "step": 1822
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6766605802699874e-06,
+      "loss": 0.2444,
+      "step": 1823
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.66723942348615e-06,
+      "loss": 0.1939,
+      "step": 1824
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.657832323881776e-06,
+      "loss": 0.207,
+      "step": 1825
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.648439299490623e-06,
+      "loss": 0.2298,
+      "step": 1826
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6390603683194814e-06,
+      "loss": 0.2301,
+      "step": 1827
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6296955483481135e-06,
+      "loss": 0.2272,
+      "step": 1828
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.62034485752923e-06,
+      "loss": 0.2109,
+      "step": 1829
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6110083137884666e-06,
+      "loss": 0.2181,
+      "step": 1830
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.601685935024323e-06,
+      "loss": 0.2192,
+      "step": 1831
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5923777391081595e-06,
+      "loss": 0.2157,
+      "step": 1832
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5830837438841337e-06,
+      "loss": 0.1968,
+      "step": 1833
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5738039671691884e-06,
+      "loss": 0.2277,
+      "step": 1834
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5645384267530093e-06,
+      "loss": 0.2129,
+      "step": 1835
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5552871403979863e-06,
+      "loss": 0.2174,
+      "step": 1836
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5460501258391813e-06,
+      "loss": 0.1934,
+      "step": 1837
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5368274007843075e-06,
+      "loss": 0.2584,
+      "step": 1838
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.527618982913671e-06,
+      "loss": 0.2207,
+      "step": 1839
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5184248898801632e-06,
+      "loss": 0.22,
+      "step": 1840
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.509245139309203e-06,
+      "loss": 0.242,
+      "step": 1841
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5000797487987248e-06,
+      "loss": 0.2047,
+      "step": 1842
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4909287359191268e-06,
+      "loss": 0.2127,
+      "step": 1843
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4817921182132445e-06,
+      "loss": 0.2155,
+      "step": 1844
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.472669913196323e-06,
+      "loss": 0.2314,
+      "step": 1845
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4635621383559716e-06,
+      "loss": 0.2367,
+      "step": 1846
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4544688111521432e-06,
+      "loss": 0.1741,
+      "step": 1847
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4453899490170874e-06,
+      "loss": 0.1801,
+      "step": 1848
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4363255693553246e-06,
+      "loss": 0.2251,
+      "step": 1849
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.427275689543618e-06,
+      "loss": 0.2296,
+      "step": 1850
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.418240326930924e-06,
+      "loss": 0.1998,
+      "step": 1851
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4092194988383788e-06,
+      "loss": 0.2007,
+      "step": 1852
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4002132225592533e-06,
+      "loss": 0.2389,
+      "step": 1853
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3912215153589136e-06,
+      "loss": 0.1869,
+      "step": 1854
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3822443944748087e-06,
+      "loss": 0.2153,
+      "step": 1855
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3732818771164124e-06,
+      "loss": 0.2482,
+      "step": 1856
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3643339804652167e-06,
+      "loss": 0.2217,
+      "step": 1857
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3554007216746756e-06,
+      "loss": 0.2112,
+      "step": 1858
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.346482117870179e-06,
+      "loss": 0.2192,
+      "step": 1859
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3375781861490356e-06,
+      "loss": 0.1966,
+      "step": 1860
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.328688943580413e-06,
+      "loss": 0.2217,
+      "step": 1861
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3198144072053306e-06,
+      "loss": 0.2211,
+      "step": 1862
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3109545940366086e-06,
+      "loss": 0.1993,
+      "step": 1863
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3021095210588405e-06,
+      "loss": 0.2268,
+      "step": 1864
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2932792052283723e-06,
+      "loss": 0.1786,
+      "step": 1865
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2844636634732474e-06,
+      "loss": 0.2009,
+      "step": 1866
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.275662912693197e-06,
+      "loss": 0.2309,
+      "step": 1867
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2668769697595917e-06,
+      "loss": 0.2004,
+      "step": 1868
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.258105851515415e-06,
+      "loss": 0.189,
+      "step": 1869
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.249349574775237e-06,
+      "loss": 0.1841,
+      "step": 1870
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2406081563251668e-06,
+      "loss": 0.2018,
+      "step": 1871
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.231881612922838e-06,
+      "loss": 0.1864,
+      "step": 1872
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2231699612973636e-06,
+      "loss": 0.2341,
+      "step": 1873
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2144732181493054e-06,
+      "loss": 0.2104,
+      "step": 1874
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2057914001506552e-06,
+      "loss": 0.1891,
+      "step": 1875
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1971245239447804e-06,
+      "loss": 0.2179,
+      "step": 1876
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1884726061464167e-06,
+      "loss": 0.1953,
+      "step": 1877
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.179835663341614e-06,
+      "loss": 0.2215,
+      "step": 1878
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1712137120877173e-06,
+      "loss": 0.2232,
+      "step": 1879
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.162606768913339e-06,
+      "loss": 0.2179,
+      "step": 1880
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.154014850318307e-06,
+      "loss": 0.1943,
+      "step": 1881
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.145437972773662e-06,
+      "loss": 0.2186,
+      "step": 1882
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1368761527215964e-06,
+      "loss": 0.1898,
+      "step": 1883
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1283294065754477e-06,
+      "loss": 0.203,
+      "step": 1884
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.119797750719651e-06,
+      "loss": 0.2156,
+      "step": 1885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1112812015097096e-06,
+      "loss": 0.2206,
+      "step": 1886
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1027797752721733e-06,
+      "loss": 0.222,
+      "step": 1887
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0942934883046008e-06,
+      "loss": 0.2182,
+      "step": 1888
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.085822356875519e-06,
+      "loss": 0.1992,
+      "step": 1889
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.077366397224413e-06,
+      "loss": 0.2366,
+      "step": 1890
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.068925625561673e-06,
+      "loss": 0.1845,
+      "step": 1891
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0605000580685817e-06,
+      "loss": 0.1981,
+      "step": 1892
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0520897108972704e-06,
+      "loss": 0.194,
+      "step": 1893
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.043694600170687e-06,
+      "loss": 0.2145,
+      "step": 1894
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0353147419825834e-06,
+      "loss": 0.2094,
+      "step": 1895
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.02695015239746e-06,
+      "loss": 0.1886,
+      "step": 1896
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0186008474505557e-06,
+      "loss": 0.1935,
+      "step": 1897
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.010266843147799e-06,
+      "loss": 0.1902,
+      "step": 1898
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.001948155465797e-06,
+      "loss": 0.1963,
+      "step": 1899
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.993644800351786e-06,
+      "loss": 0.2013,
+      "step": 1900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9853567937236096e-06,
+      "loss": 0.192,
+      "step": 1901
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9770841514696936e-06,
+      "loss": 0.2247,
+      "step": 1902
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.968826889449004e-06,
+      "loss": 0.2269,
+      "step": 1903
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9605850234910274e-06,
+      "loss": 0.2027,
+      "step": 1904
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.952358569395727e-06,
+      "loss": 0.2032,
+      "step": 1905
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.944147542933532e-06,
+      "loss": 0.1999,
+      "step": 1906
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9359519598452903e-06,
+      "loss": 0.2148,
+      "step": 1907
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9277718358422447e-06,
+      "loss": 0.2095,
+      "step": 1908
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9196071866059973e-06,
+      "loss": 0.1762,
+      "step": 1909
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9114580277884976e-06,
+      "loss": 0.1777,
+      "step": 1910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9033243750119835e-06,
+      "loss": 0.2354,
+      "step": 1911
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8952062438689823e-06,
+      "loss": 0.2329,
+      "step": 1912
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8871036499222517e-06,
+      "loss": 0.1931,
+      "step": 1913
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8790166087047778e-06,
+      "loss": 0.2093,
+      "step": 1914
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8709451357197206e-06,
+      "loss": 0.2094,
+      "step": 1915
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.862889246440397e-06,
+      "loss": 0.1833,
+      "step": 1916
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8548489563102567e-06,
+      "loss": 0.2394,
+      "step": 1917
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.846824280742835e-06,
+      "loss": 0.1962,
+      "step": 1918
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8388152351217436e-06,
+      "loss": 0.189,
+      "step": 1919
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8308218348006246e-06,
+      "loss": 0.2205,
+      "step": 1920
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8228440951031245e-06,
+      "loss": 0.2092,
+      "step": 1921
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8148820313228776e-06,
+      "loss": 0.2199,
+      "step": 1922
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8069356587234565e-06,
+      "loss": 0.1923,
+      "step": 1923
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7990049925383624e-06,
+      "loss": 0.2004,
+      "step": 1924
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7910900479709826e-06,
+      "loss": 0.1848,
+      "step": 1925
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7831908401945609e-06,
+      "loss": 0.2311,
+      "step": 1926
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7753073843521818e-06,
+      "loss": 0.1999,
+      "step": 1927
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7674396955567252e-06,
+      "loss": 0.2268,
+      "step": 1928
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.759587788890853e-06,
+      "loss": 0.1755,
+      "step": 1929
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7517516794069634e-06,
+      "loss": 0.2006,
+      "step": 1930
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.743931382127173e-06,
+      "loss": 0.2174,
+      "step": 1931
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7361269120432923e-06,
+      "loss": 0.1785,
+      "step": 1932
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7283382841167817e-06,
+      "loss": 0.2491,
+      "step": 1933
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.720565513278739e-06,
+      "loss": 0.191,
+      "step": 1934
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7128086144298595e-06,
+      "loss": 0.2003,
+      "step": 1935
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7050676024404067e-06,
+      "loss": 0.2002,
+      "step": 1936
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6973424921502014e-06,
+      "loss": 0.2088,
+      "step": 1937
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6896332983685648e-06,
+      "loss": 0.2103,
+      "step": 1938
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.68194003587432e-06,
+      "loss": 0.1913,
+      "step": 1939
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.674262719415738e-06,
+      "loss": 0.2113,
+      "step": 1940
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6666013637105205e-06,
+      "loss": 0.2038,
+      "step": 1941
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6589559834457869e-06,
+      "loss": 0.2136,
+      "step": 1942
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6513265932780131e-06,
+      "loss": 0.2104,
+      "step": 1943
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6437132078330332e-06,
+      "loss": 0.202,
+      "step": 1944
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6361158417059919e-06,
+      "loss": 0.2273,
+      "step": 1945
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6285345094613258e-06,
+      "loss": 0.1891,
+      "step": 1946
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6209692256327413e-06,
+      "loss": 0.2239,
+      "step": 1947
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6134200047231673e-06,
+      "loss": 0.1851,
+      "step": 1948
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6058868612047518e-06,
+      "loss": 0.229,
+      "step": 1949
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.598369809518814e-06,
+      "loss": 0.1938,
+      "step": 1950
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.590868864075822e-06,
+      "loss": 0.2231,
+      "step": 1951
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.583384039255379e-06,
+      "loss": 0.2147,
+      "step": 1952
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5759153494061708e-06,
+      "loss": 0.2051,
+      "step": 1953
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.568462808845964e-06,
+      "loss": 0.1954,
+      "step": 1954
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5610264318615598e-06,
+      "loss": 0.1832,
+      "step": 1955
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5536062327087686e-06,
+      "loss": 0.1685,
+      "step": 1956
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.546202225612401e-06,
+      "loss": 0.1873,
+      "step": 1957
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5388144247662118e-06,
+      "loss": 0.1861,
+      "step": 1958
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5314428443328988e-06,
+      "loss": 0.2226,
+      "step": 1959
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5240874984440612e-06,
+      "loss": 0.1871,
+      "step": 1960
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.516748401200172e-06,
+      "loss": 0.2319,
+      "step": 1961
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5094255666705626e-06,
+      "loss": 0.2178,
+      "step": 1962
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5021190088933778e-06,
+      "loss": 0.2212,
+      "step": 1963
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4948287418755725e-06,
+      "loss": 0.2014,
+      "step": 1964
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4875547795928603e-06,
+      "loss": 0.204,
+      "step": 1965
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4802971359897001e-06,
+      "loss": 0.2215,
+      "step": 1966
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4730558249792749e-06,
+      "loss": 0.1949,
+      "step": 1967
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4658308604434456e-06,
+      "loss": 0.1967,
+      "step": 1968
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4586222562327479e-06,
+      "loss": 0.2126,
+      "step": 1969
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4514300261663472e-06,
+      "loss": 0.2236,
+      "step": 1970
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.444254184032018e-06,
+      "loss": 0.1851,
+      "step": 1971
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4370947435861248e-06,
+      "loss": 0.2111,
+      "step": 1972
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4299517185535816e-06,
+      "loss": 0.1802,
+      "step": 1973
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4228251226278433e-06,
+      "loss": 0.226,
+      "step": 1974
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4157149694708593e-06,
+      "loss": 0.2115,
+      "step": 1975
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4086212727130654e-06,
+      "loss": 0.2018,
+      "step": 1976
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4015440459533424e-06,
+      "loss": 0.1998,
+      "step": 1977
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3944833027590044e-06,
+      "loss": 0.2062,
+      "step": 1978
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.387439056665767e-06,
+      "loss": 0.1569,
+      "step": 1979
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3804113211777116e-06,
+      "loss": 0.2032,
+      "step": 1980
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3734001097672722e-06,
+      "loss": 0.1833,
+      "step": 1981
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.36640543587521e-06,
+      "loss": 0.1962,
+      "step": 1982
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3594273129105761e-06,
+      "loss": 0.1855,
+      "step": 1983
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3524657542507002e-06,
+      "loss": 0.2221,
+      "step": 1984
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3455207732411523e-06,
+      "loss": 0.2002,
+      "step": 1985
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3385923831957192e-06,
+      "loss": 0.1936,
+      "step": 1986
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3316805973963942e-06,
+      "loss": 0.1944,
+      "step": 1987
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3247854290933248e-06,
+      "loss": 0.2232,
+      "step": 1988
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3179068915048166e-06,
+      "loss": 0.2027,
+      "step": 1989
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3110449978172802e-06,
+      "loss": 0.2284,
+      "step": 1990
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3041997611852298e-06,
+      "loss": 0.1768,
+      "step": 1991
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2973711947312418e-06,
+      "loss": 0.2219,
+      "step": 1992
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2905593115459336e-06,
+      "loss": 0.1841,
+      "step": 1993
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2837641246879485e-06,
+      "loss": 0.1952,
+      "step": 1994
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2769856471839105e-06,
+      "loss": 0.1947,
+      "step": 1995
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2702238920284215e-06,
+      "loss": 0.1893,
+      "step": 1996
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.263478872184023e-06,
+      "loss": 0.1832,
+      "step": 1997
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2567506005811691e-06,
+      "loss": 0.1993,
+      "step": 1998
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2500390901182147e-06,
+      "loss": 0.1867,
+      "step": 1999
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2433443536613787e-06,
+      "loss": 0.1937,
+      "step": 2000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.236666404044722e-06,
+      "loss": 0.1864,
+      "step": 2001
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2300052540701291e-06,
+      "loss": 0.1955,
+      "step": 2002
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2233609165072725e-06,
+      "loss": 0.2109,
+      "step": 2003
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2167334040936052e-06,
+      "loss": 0.1902,
+      "step": 2004
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2101227295343133e-06,
+      "loss": 0.2161,
+      "step": 2005
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2035289055023135e-06,
+      "loss": 0.2331,
+      "step": 2006
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1969519446382161e-06,
+      "loss": 0.2302,
+      "step": 2007
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1903918595503007e-06,
+      "loss": 0.1974,
+      "step": 2008
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.183848662814502e-06,
+      "loss": 0.1975,
+      "step": 2009
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1773223669743727e-06,
+      "loss": 0.2121,
+      "step": 2010
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1708129845410721e-06,
+      "loss": 0.1911,
+      "step": 2011
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1643205279933313e-06,
+      "loss": 0.2097,
+      "step": 2012
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1578450097774308e-06,
+      "loss": 0.2065,
+      "step": 2013
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1513864423071918e-06,
+      "loss": 0.1995,
+      "step": 2014
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1449448379639283e-06,
+      "loss": 0.2172,
+      "step": 2015
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1385202090964386e-06,
+      "loss": 0.2223,
+      "step": 2016
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1321125680209833e-06,
+      "loss": 0.19,
+      "step": 2017
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.125721927021247e-06,
+      "loss": 0.2046,
+      "step": 2018
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.119348298348336e-06,
+      "loss": 0.1955,
+      "step": 2019
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.112991694220733e-06,
+      "loss": 0.2557,
+      "step": 2020
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1066521268242926e-06,
+      "loss": 0.1886,
+      "step": 2021
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.100329608312203e-06,
+      "loss": 0.183,
+      "step": 2022
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.094024150804971e-06,
+      "loss": 0.183,
+      "step": 2023
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0877357663903987e-06,
+      "loss": 0.1912,
+      "step": 2024
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0814644671235553e-06,
+      "loss": 0.2188,
+      "step": 2025
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0752102650267615e-06,
+      "loss": 0.202,
+      "step": 2026
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.068973172089557e-06,
+      "loss": 0.2006,
+      "step": 2027
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0627532002686836e-06,
+      "loss": 0.2115,
+      "step": 2028
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0565503614880667e-06,
+      "loss": 0.1862,
+      "step": 2029
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.050364667638779e-06,
+      "loss": 0.214,
+      "step": 2030
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0441961305790315e-06,
+      "loss": 0.2497,
+      "step": 2031
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.038044762134145e-06,
+      "loss": 0.1903,
+      "step": 2032
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0319105740965218e-06,
+      "loss": 0.2115,
+      "step": 2033
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0257935782256356e-06,
+      "loss": 0.1985,
+      "step": 2034
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.019693786247996e-06,
+      "loss": 0.2132,
+      "step": 2035
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0136112098571383e-06,
+      "loss": 0.2194,
+      "step": 2036
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0075458607135901e-06,
+      "loss": 0.1773,
+      "step": 2037
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0014977504448532e-06,
+      "loss": 0.2182,
+      "step": 2038
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.954668906453868e-07,
+      "loss": 0.1843,
+      "step": 2039
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.894532928765743e-07,
+      "loss": 0.1816,
+      "step": 2040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.83456968666715e-07,
+      "loss": 0.221,
+      "step": 2041
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.77477929510985e-07,
+      "loss": 0.2044,
+      "step": 2042
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.715161868714273e-07,
+      "loss": 0.2207,
+      "step": 2043
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.655717521769337e-07,
+      "loss": 0.1672,
+      "step": 2044
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.596446368232049e-07,
+      "loss": 0.2459,
+      "step": 2045
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.537348521727497e-07,
+      "loss": 0.181,
+      "step": 2046
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.478424095548444e-07,
+      "loss": 0.2028,
+      "step": 2047
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.419673202655288e-07,
+      "loss": 0.2059,
+      "step": 2048
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.361095955675681e-07,
+      "loss": 0.2182,
+      "step": 2049
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.302692466904439e-07,
+      "loss": 0.1897,
+      "step": 2050
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.244462848303259e-07,
+      "loss": 0.1927,
+      "step": 2051
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.186407211500537e-07,
+      "loss": 0.1949,
+      "step": 2052
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.128525667791077e-07,
+      "loss": 0.2016,
+      "step": 2053
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.070818328136044e-07,
+      "loss": 0.2184,
+      "step": 2054
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.013285303162545e-07,
+      "loss": 0.2006,
+      "step": 2055
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.955926703163609e-07,
+      "loss": 0.1984,
+      "step": 2056
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.898742638097813e-07,
+      "loss": 0.2104,
+      "step": 2057
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.841733217589166e-07,
+      "loss": 0.2164,
+      "step": 2058
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.784898550926924e-07,
+      "loss": 0.1868,
+      "step": 2059
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.728238747065243e-07,
+      "loss": 0.1932,
+      "step": 2060
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.671753914623148e-07,
+      "loss": 0.1964,
+      "step": 2061
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.615444161884179e-07,
+      "loss": 0.1802,
+      "step": 2062
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.559309596796295e-07,
+      "loss": 0.2199,
+      "step": 2063
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.503350326971549e-07,
+      "loss": 0.2131,
+      "step": 2064
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.447566459685963e-07,
+      "loss": 0.1958,
+      "step": 2065
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.391958101879371e-07,
+      "loss": 0.1888,
+      "step": 2066
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.336525360155056e-07,
+      "loss": 0.2085,
+      "step": 2067
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.281268340779692e-07,
+      "loss": 0.2185,
+      "step": 2068
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.226187149683107e-07,
+      "loss": 0.1823,
+      "step": 2069
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.171281892457983e-07,
+      "loss": 0.2125,
+      "step": 2070
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.116552674359812e-07,
+      "loss": 0.1841,
+      "step": 2071
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.061999600306558e-07,
+      "loss": 0.1914,
+      "step": 2072
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.007622774878509e-07,
+      "loss": 0.2018,
+      "step": 2073
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.953422302318137e-07,
+      "loss": 0.1881,
+      "step": 2074
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.89939828652977e-07,
+      "loss": 0.196,
+      "step": 2075
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.845550831079507e-07,
+      "loss": 0.1939,
+      "step": 2076
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.791880039194932e-07,
+      "loss": 0.2029,
+      "step": 2077
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.73838601376502e-07,
+      "loss": 0.2283,
+      "step": 2078
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.685068857339817e-07,
+      "loss": 0.2208,
+      "step": 2079
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.631928672130328e-07,
+      "loss": 0.2249,
+      "step": 2080
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.578965560008311e-07,
+      "loss": 0.1813,
+      "step": 2081
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.526179622506047e-07,
+      "loss": 0.1879,
+      "step": 2082
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.473570960816179e-07,
+      "loss": 0.211,
+      "step": 2083
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.421139675791511e-07,
+      "loss": 0.2293,
+      "step": 2084
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.368885867944753e-07,
+      "loss": 0.246,
+      "step": 2085
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.316809637448508e-07,
+      "loss": 0.2171,
+      "step": 2086
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.26491108413484e-07,
+      "loss": 0.2068,
+      "step": 2087
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.213190307495222e-07,
+      "loss": 0.2001,
+      "step": 2088
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.161647406680372e-07,
+      "loss": 0.2061,
+      "step": 2089
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.110282480499953e-07,
+      "loss": 0.1962,
+      "step": 2090
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.059095627422508e-07,
+      "loss": 0.2194,
+      "step": 2091
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.008086945575121e-07,
+      "loss": 0.1915,
+      "step": 2092
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.957256532743417e-07,
+      "loss": 0.2042,
+      "step": 2093
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.906604486371205e-07,
+      "loss": 0.212,
+      "step": 2094
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.856130903560331e-07,
+      "loss": 0.2132,
+      "step": 2095
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.805835881070622e-07,
+      "loss": 0.208,
+      "step": 2096
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.755719515319503e-07,
+      "loss": 0.2349,
+      "step": 2097
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.70578190238197e-07,
+      "loss": 0.2067,
+      "step": 2098
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.656023137990297e-07,
+      "loss": 0.1944,
+      "step": 2099
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.606443317533905e-07,
+      "loss": 0.1792,
+      "step": 2100
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.557042536059232e-07,
+      "loss": 0.2012,
+      "step": 2101
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.507820888269389e-07,
+      "loss": 0.2286,
+      "step": 2102
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.458778468524163e-07,
+      "loss": 0.2125,
+      "step": 2103
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.409915370839747e-07,
+      "loss": 0.1877,
+      "step": 2104
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.361231688888525e-07,
+      "loss": 0.1829,
+      "step": 2105
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.312727515998962e-07,
+      "loss": 0.2173,
+      "step": 2106
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.26440294515539e-07,
+      "loss": 0.1858,
+      "step": 2107
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.216258068997849e-07,
+      "loss": 0.1845,
+      "step": 2108
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.168292979821888e-07,
+      "loss": 0.1916,
+      "step": 2109
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.120507769578377e-07,
+      "loss": 0.2403,
+      "step": 2110
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.072902529873403e-07,
+      "loss": 0.2116,
+      "step": 2111
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.025477351967978e-07,
+      "loss": 0.2023,
+      "step": 2112
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.978232326777999e-07,
+      "loss": 0.2187,
+      "step": 2113
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.93116754487394e-07,
+      "loss": 0.2048,
+      "step": 2114
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.884283096480769e-07,
+      "loss": 0.2024,
+      "step": 2115
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.837579071477761e-07,
+      "loss": 0.2448,
+      "step": 2116
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.791055559398284e-07,
+      "loss": 0.2348,
+      "step": 2117
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.744712649429696e-07,
+      "loss": 0.2326,
+      "step": 2118
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.698550430413075e-07,
+      "loss": 0.2222,
+      "step": 2119
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.652568990843166e-07,
+      "loss": 0.163,
+      "step": 2120
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.606768418868114e-07,
+      "loss": 0.2065,
+      "step": 2121
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.561148802289352e-07,
+      "loss": 0.1923,
+      "step": 2122
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.515710228561433e-07,
+      "loss": 0.1917,
+      "step": 2123
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.470452784791802e-07,
+      "loss": 0.1926,
+      "step": 2124
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.42537655774068e-07,
+      "loss": 0.1869,
+      "step": 2125
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.380481633820922e-07,
+      "loss": 0.2042,
+      "step": 2126
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.335768099097771e-07,
+      "loss": 0.223,
+      "step": 2127
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.291236039288805e-07,
+      "loss": 0.2039,
+      "step": 2128
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.246885539763624e-07,
+      "loss": 0.2162,
+      "step": 2129
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.202716685543829e-07,
+      "loss": 0.2044,
+      "step": 2130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.1587295613028e-07,
+      "loss": 0.1761,
+      "step": 2131
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.114924251365494e-07,
+      "loss": 0.217,
+      "step": 2132
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.07130083970837e-07,
+      "loss": 0.2047,
+      "step": 2133
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.027859409959135e-07,
+      "loss": 0.2044,
+      "step": 2134
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.984600045396681e-07,
+      "loss": 0.1943,
+      "step": 2135
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.941522828950829e-07,
+      "loss": 0.1865,
+      "step": 2136
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.898627843202231e-07,
+      "loss": 0.1881,
+      "step": 2137
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.855915170382209e-07,
+      "loss": 0.214,
+      "step": 2138
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.813384892372542e-07,
+      "loss": 0.2353,
+      "step": 2139
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.771037090705421e-07,
+      "loss": 0.2134,
+      "step": 2140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7288718465631765e-07,
+      "loss": 0.2257,
+      "step": 2141
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6868892407781587e-07,
+      "loss": 0.2203,
+      "step": 2142
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6450893538326327e-07,
+      "loss": 0.172,
+      "step": 2143
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.603472265858555e-07,
+      "loss": 0.2128,
+      "step": 2144
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5620380566374413e-07,
+      "loss": 0.1777,
+      "step": 2145
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5207868056002746e-07,
+      "loss": 0.2066,
+      "step": 2146
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4797185918272203e-07,
+      "loss": 0.2136,
+      "step": 2147
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.438833494047623e-07,
+      "loss": 0.152,
+      "step": 2148
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.398131590639765e-07,
+      "loss": 0.197,
+      "step": 2149
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3576129596307304e-07,
+      "loss": 0.1959,
+      "step": 2150
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3172776786962635e-07,
+      "loss": 0.2171,
+      "step": 2151
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.277125825160622e-07,
+      "loss": 0.1947,
+      "step": 2152
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.237157475996434e-07,
+      "loss": 0.2114,
+      "step": 2153
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1973727078245317e-07,
+      "loss": 0.1821,
+      "step": 2154
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1577715969138397e-07,
+      "loss": 0.1852,
+      "step": 2155
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1183542191811646e-07,
+      "loss": 0.1984,
+      "step": 2156
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.079120650191104e-07,
+      "loss": 0.173,
+      "step": 2157
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.040070965155918e-07,
+      "loss": 0.2189,
+      "step": 2158
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.001205238935335e-07,
+      "loss": 0.2085,
+      "step": 2159
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9625235460363785e-07,
+      "loss": 0.1842,
+      "step": 2160
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9240259606133756e-07,
+      "loss": 0.2081,
+      "step": 2161
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8857125564676026e-07,
+      "loss": 0.2029,
+      "step": 2162
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.847583407047339e-07,
+      "loss": 0.195,
+      "step": 2163
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8096385854475834e-07,
+      "loss": 0.2121,
+      "step": 2164
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7718781644100143e-07,
+      "loss": 0.2444,
+      "step": 2165
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7343022163227826e-07,
+      "loss": 0.1908,
+      "step": 2166
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.696910813220389e-07,
+      "loss": 0.2116,
+      "step": 2167
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.659704026783595e-07,
+      "loss": 0.2247,
+      "step": 2168
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.622681928339178e-07,
+      "loss": 0.2083,
+      "step": 2169
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5858445888599434e-07,
+      "loss": 0.1951,
+      "step": 2170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5491920789644476e-07,
+      "loss": 0.1894,
+      "step": 2171
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5127244689169173e-07,
+      "loss": 0.1956,
+      "step": 2172
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.476441828627175e-07,
+      "loss": 0.2021,
+      "step": 2173
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.440344227650394e-07,
+      "loss": 0.1779,
+      "step": 2174
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4044317351870415e-07,
+      "loss": 0.1635,
+      "step": 2175
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.368704420082736e-07,
+      "loss": 0.2266,
+      "step": 2176
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.33316235082809e-07,
+      "loss": 0.1924,
+      "step": 2177
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2978055955585896e-07,
+      "loss": 0.2498,
+      "step": 2178
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2626342220544613e-07,
+      "loss": 0.2104,
+      "step": 2179
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.227648297740593e-07,
+      "loss": 0.1822,
+      "step": 2180
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1928478896863013e-07,
+      "loss": 0.207,
+      "step": 2181
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1582330646052764e-07,
+      "loss": 0.2307,
+      "step": 2182
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.12380388885547e-07,
+      "loss": 0.1934,
+      "step": 2183
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.089560428438887e-07,
+      "loss": 0.2024,
+      "step": 2184
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.055502749001571e-07,
+      "loss": 0.1805,
+      "step": 2185
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0216309158333623e-07,
+      "loss": 0.1788,
+      "step": 2186
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9879449938678304e-07,
+      "loss": 0.1963,
+      "step": 2187
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9544450476821974e-07,
+      "loss": 0.1925,
+      "step": 2188
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.921131141497102e-07,
+      "loss": 0.2011,
+      "step": 2189
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8880033391765817e-07,
+      "loss": 0.2161,
+      "step": 2190
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8550617042278686e-07,
+      "loss": 0.2032,
+      "step": 2191
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.822306299801325e-07,
+      "loss": 0.2202,
+      "step": 2192
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7897371886903114e-07,
+      "loss": 0.2372,
+      "step": 2193
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7573544333310385e-07,
+      "loss": 0.1962,
+      "step": 2194
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.725158095802494e-07,
+      "loss": 0.2278,
+      "step": 2195
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6931482378262487e-07,
+      "loss": 0.211,
+      "step": 2196
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.661324920766406e-07,
+      "loss": 0.1849,
+      "step": 2197
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6296882056294994e-07,
+      "loss": 0.2018,
+      "step": 2198
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5982381530642807e-07,
+      "loss": 0.2136,
+      "step": 2199
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.566974823361712e-07,
+      "loss": 0.21,
+      "step": 2200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5358982764547737e-07,
+      "loss": 0.1807,
+      "step": 2201
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5050085719183656e-07,
+      "loss": 0.2225,
+      "step": 2202
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.474305768969232e-07,
+      "loss": 0.1905,
+      "step": 2203
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4437899264658114e-07,
+      "loss": 0.1827,
+      "step": 2204
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4134611029081325e-07,
+      "loss": 0.2189,
+      "step": 2205
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3833193564376854e-07,
+      "loss": 0.2368,
+      "step": 2206
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3533647448373277e-07,
+      "loss": 0.2037,
+      "step": 2207
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.323597325531224e-07,
+      "loss": 0.2312,
+      "step": 2208
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2940171555845936e-07,
+      "loss": 0.2057,
+      "step": 2209
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2646242917037763e-07,
+      "loss": 0.222,
+      "step": 2210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2354187902359991e-07,
+      "loss": 0.2152,
+      "step": 2211
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2064007071692984e-07,
+      "loss": 0.2122,
+      "step": 2212
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1775700981324532e-07,
+      "loss": 0.2001,
+      "step": 2213
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1489270183948196e-07,
+      "loss": 0.2036,
+      "step": 2214
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1204715228662741e-07,
+      "loss": 0.1845,
+      "step": 2215
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0922036660970813e-07,
+      "loss": 0.2175,
+      "step": 2216
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0641235022777707e-07,
+      "loss": 0.1889,
+      "step": 2217
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0362310852391155e-07,
+      "loss": 0.1868,
+      "step": 2218
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.008526468451888e-07,
+      "loss": 0.219,
+      "step": 2219
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9810097050269373e-07,
+      "loss": 0.2072,
+      "step": 2220
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9536808477149004e-07,
+      "loss": 0.21,
+      "step": 2221
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9265399489062365e-07,
+      "loss": 0.2056,
+      "step": 2222
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.899587060631103e-07,
+      "loss": 0.2037,
+      "step": 2223
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8728222345591684e-07,
+      "loss": 0.188,
+      "step": 2224
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8462455219996568e-07,
+      "loss": 0.1892,
+      "step": 2225
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8198569739010908e-07,
+      "loss": 0.2357,
+      "step": 2226
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.793656640851349e-07,
+      "loss": 0.1902,
+      "step": 2227
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7676445730774427e-07,
+      "loss": 0.2041,
+      "step": 2228
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7418208204454835e-07,
+      "loss": 0.1905,
+      "step": 2229
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7161854324606043e-07,
+      "loss": 0.1893,
+      "step": 2230
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.690738458266805e-07,
+      "loss": 0.1984,
+      "step": 2231
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6654799466468864e-07,
+      "loss": 0.1957,
+      "step": 2232
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6404099460223922e-07,
+      "loss": 0.1721,
+      "step": 2233
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6155285044534452e-07,
+      "loss": 0.2161,
+      "step": 2234
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5908356696387128e-07,
+      "loss": 0.1986,
+      "step": 2235
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.566331488915307e-07,
+      "loss": 0.2265,
+      "step": 2236
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.542016009258651e-07,
+      "loss": 0.2273,
+      "step": 2237
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.517889277282447e-07,
+      "loss": 0.2038,
+      "step": 2238
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4939513392385418e-07,
+      "loss": 0.1959,
+      "step": 2239
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4702022410168937e-07,
+      "loss": 0.2111,
+      "step": 2240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4466420281454063e-07,
+      "loss": 0.2085,
+      "step": 2241
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4232707457898953e-07,
+      "loss": 0.1749,
+      "step": 2242
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4000884387540104e-07,
+      "loss": 0.213,
+      "step": 2243
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.377095151479091e-07,
+      "loss": 0.2216,
+      "step": 2244
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.354290928044144e-07,
+      "loss": 0.169,
+      "step": 2245
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3316758121657336e-07,
+      "loss": 0.2343,
+      "step": 2246
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.309249847197891e-07,
+      "loss": 0.2082,
+      "step": 2247
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2870130761320488e-07,
+      "loss": 0.1872,
+      "step": 2248
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2649655415969186e-07,
+      "loss": 0.1978,
+      "step": 2249
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2431072858584803e-07,
+      "loss": 0.2262,
+      "step": 2250
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2214383508198146e-07,
+      "loss": 0.2386,
+      "step": 2251
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.199958778021104e-07,
+      "loss": 0.182,
+      "step": 2252
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1786686086395105e-07,
+      "loss": 0.1968,
+      "step": 2253
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1575678834890635e-07,
+      "loss": 0.2114,
+      "step": 2254
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1366566430206838e-07,
+      "loss": 0.1643,
+      "step": 2255
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1159349273219932e-07,
+      "loss": 0.1653,
+      "step": 2256
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0954027761173159e-07,
+      "loss": 0.1905,
+      "step": 2257
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0750602287675438e-07,
+      "loss": 0.208,
+      "step": 2258
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0549073242701269e-07,
+      "loss": 0.2003,
+      "step": 2259
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.03494410125895e-07,
+      "loss": 0.2049,
+      "step": 2260
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0151705980042448e-07,
+      "loss": 0.1972,
+      "step": 2261
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.955868524125889e-08,
+      "loss": 0.1989,
+      "step": 2262
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.761929020267402e-08,
+      "loss": 0.2094,
+      "step": 2263
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.569887840256586e-08,
+      "loss": 0.2041,
+      "step": 2264
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.379745352243397e-08,
+      "loss": 0.2047,
+      "step": 2265
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.191501920738477e-08,
+      "loss": 0.2087,
+      "step": 2266
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.005157906611383e-08,
+      "loss": 0.2092,
+      "step": 2267
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.820713667090697e-08,
+      "loss": 0.1819,
+      "step": 2268
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.638169555763021e-08,
+      "loss": 0.2274,
+      "step": 2269
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.457525922572318e-08,
+      "loss": 0.2008,
+      "step": 2270
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.278783113819244e-08,
+      "loss": 0.1918,
+      "step": 2271
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.101941472160702e-08,
+      "loss": 0.1839,
+      "step": 2272
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.927001336608509e-08,
+      "loss": 0.2039,
+      "step": 2273
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.753963042529733e-08,
+      "loss": 0.1914,
+      "step": 2274
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.582826921645247e-08,
+      "loss": 0.178,
+      "step": 2275
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.413593302029399e-08,
+      "loss": 0.1946,
+      "step": 2276
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.246262508109669e-08,
+      "loss": 0.1947,
+      "step": 2277
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.080834860665353e-08,
+      "loss": 0.2056,
+      "step": 2278
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.917310676827549e-08,
+      "loss": 0.2033,
+      "step": 2279
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.755690270078275e-08,
+      "loss": 0.2072,
+      "step": 2280
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.595973950249912e-08,
+      "loss": 0.2063,
+      "step": 2281
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.438162023524875e-08,
+      "loss": 0.2066,
+      "step": 2282
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.282254792434494e-08,
+      "loss": 0.2144,
+      "step": 2283
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.128252555858693e-08,
+      "loss": 0.1855,
+      "step": 2284
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.976155609025869e-08,
+      "loss": 0.1961,
+      "step": 2285
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.825964243511339e-08,
+      "loss": 0.1912,
+      "step": 2286
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6776787472379026e-08,
+      "loss": 0.2208,
+      "step": 2287
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5312994044742774e-08,
+      "loss": 0.2103,
+      "step": 2288
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.386826495835107e-08,
+      "loss": 0.2168,
+      "step": 2289
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.2442602982807346e-08,
+      "loss": 0.1978,
+      "step": 2290
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.103601085115761e-08,
+      "loss": 0.1944,
+      "step": 2291
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.964849125989268e-08,
+      "loss": 0.179,
+      "step": 2292
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.828004686893928e-08,
+      "loss": 0.2188,
+      "step": 2293
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6930680301657816e-08,
+      "loss": 0.2337,
+      "step": 2294
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.560039414483575e-08,
+      "loss": 0.1991,
+      "step": 2295
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.428919094867978e-08,
+      "loss": 0.1667,
+      "step": 2296
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.299707322681923e-08,
+      "loss": 0.2113,
+      "step": 2297
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.172404345628933e-08,
+      "loss": 0.1904,
+      "step": 2298
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.047010407753904e-08,
+      "loss": 0.1922,
+      "step": 2299
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.923525749441548e-08,
+      "loss": 0.2197,
+      "step": 2300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8019506074167264e-08,
+      "loss": 0.2221,
+      "step": 2301
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.682285214743564e-08,
+      "loss": 0.1893,
+      "step": 2302
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5645298008252227e-08,
+      "loss": 0.1774,
+      "step": 2303
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4486845914032396e-08,
+      "loss": 0.2097,
+      "step": 2304
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3347498085571914e-08,
+      "loss": 0.1978,
+      "step": 2305
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.222725670704363e-08,
+      "loss": 0.1965,
+      "step": 2306
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.112612392599523e-08,
+      "loss": 0.229,
+      "step": 2307
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.004410185333706e-08,
+      "loss": 0.2175,
+      "step": 2308
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8981192563346527e-08,
+      "loss": 0.2231,
+      "step": 2309
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7937398093662582e-08,
+      "loss": 0.1994,
+      "step": 2310
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6912720445276817e-08,
+      "loss": 0.2069,
+      "step": 2311
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.590716158253681e-08,
+      "loss": 0.2326,
+      "step": 2312
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4920723433136117e-08,
+      "loss": 0.1945,
+      "step": 2313
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3953407888114287e-08,
+      "loss": 0.1987,
+      "step": 2314
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3005216801851306e-08,
+      "loss": 0.2063,
+      "step": 2315
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2076151992065363e-08,
+      "loss": 0.17,
+      "step": 2316
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.116621523980955e-08,
+      "loss": 0.1997,
+      "step": 2317
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.027540828946739e-08,
+      "loss": 0.2068,
+      "step": 2318
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9403732848747302e-08,
+      "loss": 0.2001,
+      "step": 2319
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.855119058868704e-08,
+      "loss": 0.209,
+      "step": 2320
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7717783143639255e-08,
+      "loss": 0.1971,
+      "step": 2321
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6903512111280383e-08,
+      "loss": 0.1859,
+      "step": 2322
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6108379052597324e-08,
+      "loss": 0.1786,
+      "step": 2323
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5332385491890755e-08,
+      "loss": 0.1804,
+      "step": 2324
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4575532916768497e-08,
+      "loss": 0.1981,
+      "step": 2325
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3837822778147714e-08,
+      "loss": 0.181,
+      "step": 2326
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3119256490244926e-08,
+      "loss": 0.1779,
+      "step": 2327
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.241983543057934e-08,
+      "loss": 0.2077,
+      "step": 2328
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.173956093996731e-08,
+      "loss": 0.1739,
+      "step": 2329
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1078434322522313e-08,
+      "loss": 0.1866,
+      "step": 2330
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0436456845647203e-08,
+      "loss": 0.2166,
+      "step": 2331
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.813629740037522e-09,
+      "loss": 0.1955,
+      "step": 2332
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.209954199677073e-09,
+      "loss": 0.2274,
+      "step": 2333
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.62543138183458e-09,
+      "loss": 0.2082,
+      "step": 2334
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.060062407061475e-09,
+      "loss": 0.2174,
+      "step": 2335
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.513848359191888e-09,
+      "loss": 0.1971,
+      "step": 2336
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.986790285340439e-09,
+      "loss": 0.231,
+      "step": 2337
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.478889195894455e-09,
+      "loss": 0.1736,
+      "step": 2338
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.990146064521752e-09,
+      "loss": 0.2482,
+      "step": 2339
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.5205618281617415e-09,
+      "loss": 0.183,
+      "step": 2340
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.070137387023222e-09,
+      "loss": 0.2166,
+      "step": 2341
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.638873604587701e-09,
+      "loss": 0.2079,
+      "step": 2342
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.22677130760385e-09,
+      "loss": 0.2117,
+      "step": 2343
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.833831286088607e-09,
+      "loss": 0.1836,
+      "step": 2344
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4600542933227454e-09,
+      "loss": 0.1809,
+      "step": 2345
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.105441045849755e-09,
+      "loss": 0.213,
+      "step": 2346
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7699922234769584e-09,
+      "loss": 0.1786,
+      "step": 2347
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4537084692721756e-09,
+      "loss": 0.194,
+      "step": 2348
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1565903895637286e-09,
+      "loss": 0.1867,
+      "step": 2349
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.878638553939327e-09,
+      "loss": 0.1866,
+      "step": 2350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6198534952405198e-09,
+      "loss": 0.2055,
+      "step": 2351
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3802357095715758e-09,
+      "loss": 0.2117,
+      "step": 2352
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1597856562850507e-09,
+      "loss": 0.2121,
+      "step": 2353
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.585037579951106e-10,
+      "loss": 0.2072,
+      "step": 2354
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.763904005653189e-10,
+      "loss": 0.1995,
+      "step": 2355
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.134459331130771e-10,
+      "loss": 0.2357,
+      "step": 2356
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.696706680118457e-10,
+      "loss": 0.2314,
+      "step": 2357
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.4506488088115184e-10,
+      "loss": 0.2059,
+      "step": 2358
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.396288105965816e-10,
+      "loss": 0.2028,
+      "step": 2359
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5336265928422856e-10,
+      "loss": 0.1975,
+      "step": 2360
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.626659231625312e-11,
+      "loss": 0.2003,
+      "step": 2361
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.8340738321984706e-11,
+      "loss": 0.2496,
+      "step": 2362
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.585189174599052e-12,
+      "loss": 0.2109,
+      "step": 2363
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.1878,
+      "step": 2364
+    },
+    {
+      "epoch": 3.0,
+      "step": 2364,
+      "total_flos": 3.131392840932786e+18,
+      "train_loss": 0.34906878738385166,
+      "train_runtime": 80525.9722,
+      "train_samples_per_second": 3.76,
+      "train_steps_per_second": 0.029
+    }
+  ],
+  "max_steps": 2364,
+  "num_train_epochs": 3,
+  "total_flos": 3.131392840932786e+18,
+  "trial_name": null,
+  "trial_params": null
+}