diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4762 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 791,
+  "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": 1.5774,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.210526315789474e-07,
+      "loss": 2.0325,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.315789473684211e-07,
+      "loss": 1.6573,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.421052631578948e-07,
+      "loss": 1.5422,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 1.4652,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2631578947368422e-06,
+      "loss": 1.4239,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4736842105263159e-06,
+      "loss": 1.3821,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6842105263157895e-06,
+      "loss": 1.3956,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8947368421052634e-06,
+      "loss": 1.3272,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 1.2911,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3157894736842105e-06,
+      "loss": 1.2653,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5263157894736844e-06,
+      "loss": 1.2739,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7368421052631583e-06,
+      "loss": 1.2295,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9473684210526317e-06,
+      "loss": 1.2498,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 1.2149,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.368421052631579e-06,
+      "loss": 1.207,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.578947368421053e-06,
+      "loss": 1.1964,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.789473684210527e-06,
+      "loss": 1.1957,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.2128,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 1.1877,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4210526315789476e-06,
+      "loss": 1.1839,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.631578947368421e-06,
+      "loss": 1.1865,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.842105263157895e-06,
+      "loss": 1.1608,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.052631578947369e-06,
+      "loss": 1.1334,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 1.1492,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4736842105263165e-06,
+      "loss": 1.1678,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.68421052631579e-06,
+      "loss": 1.1471,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8947368421052634e-06,
+      "loss": 1.1156,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.105263157894738e-06,
+      "loss": 1.1435,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 1.1368,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.526315789473685e-06,
+      "loss": 1.0934,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.736842105263158e-06,
+      "loss": 1.1093,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.947368421052632e-06,
+      "loss": 1.1067,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.157894736842106e-06,
+      "loss": 1.1155,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 1.1333,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.578947368421054e-06,
+      "loss": 1.1088,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.789473684210526e-06,
+      "loss": 1.1035,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.09,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.210526315789475e-06,
+      "loss": 1.0929,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 1.0959,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.631578947368422e-06,
+      "loss": 1.0963,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.842105263157895e-06,
+      "loss": 1.1186,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.05263157894737e-06,
+      "loss": 1.0964,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.263157894736842e-06,
+      "loss": 1.0705,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 1.0615,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.68421052631579e-06,
+      "loss": 1.0857,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.894736842105264e-06,
+      "loss": 1.0757,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0105263157894738e-05,
+      "loss": 1.0795,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0315789473684213e-05,
+      "loss": 1.041,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 1.0754,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.073684210526316e-05,
+      "loss": 1.0703,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0947368421052633e-05,
+      "loss": 1.0828,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1157894736842105e-05,
+      "loss": 1.0721,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.136842105263158e-05,
+      "loss": 1.0521,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 1.0308,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1789473684210527e-05,
+      "loss": 1.0462,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2e-05,
+      "loss": 1.0658,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2210526315789475e-05,
+      "loss": 1.0586,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2421052631578949e-05,
+      "loss": 1.0702,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 1.0468,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2842105263157896e-05,
+      "loss": 1.0407,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.305263157894737e-05,
+      "loss": 1.0314,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3263157894736843e-05,
+      "loss": 1.0773,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3473684210526316e-05,
+      "loss": 1.0317,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 1.049,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3894736842105265e-05,
+      "loss": 1.0123,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4105263157894738e-05,
+      "loss": 1.0257,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4315789473684212e-05,
+      "loss": 1.0412,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4526315789473687e-05,
+      "loss": 1.0419,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 1.0322,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4947368421052632e-05,
+      "loss": 1.0398,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5157894736842107e-05,
+      "loss": 1.0773,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.536842105263158e-05,
+      "loss": 1.0233,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5578947368421052e-05,
+      "loss": 1.025,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 1.026,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.0122,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6210526315789473e-05,
+      "loss": 1.0413,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.642105263157895e-05,
+      "loss": 1.0411,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6631578947368423e-05,
+      "loss": 1.0096,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 1.0063,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.705263157894737e-05,
+      "loss": 1.0144,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7263157894736843e-05,
+      "loss": 1.0264,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7473684210526317e-05,
+      "loss": 1.0291,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.768421052631579e-05,
+      "loss": 1.0117,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 1.0204,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.810526315789474e-05,
+      "loss": 1.0234,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.831578947368421e-05,
+      "loss": 1.0144,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8526315789473684e-05,
+      "loss": 1.0317,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.873684210526316e-05,
+      "loss": 1.0133,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 1.0236,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9157894736842108e-05,
+      "loss": 1.0168,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936842105263158e-05,
+      "loss": 1.0287,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578947368421055e-05,
+      "loss": 1.0184,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9789473684210528e-05,
+      "loss": 1.037,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 1.0157,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999999049040018e-05,
+      "loss": 1.0175,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999961961618814e-05,
+      "loss": 1.0371,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999914413710156e-05,
+      "loss": 1.0348,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999847846764634e-05,
+      "loss": 0.9883,
+      "step": 99
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999762260908862e-05,
+      "loss": 1.0285,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999657656305617e-05,
+      "loss": 0.9958,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999534033153842e-05,
+      "loss": 1.0154,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999391391688664e-05,
+      "loss": 0.9959,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999922973218137e-05,
+      "loss": 0.9833,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999904905493943e-05,
+      "loss": 0.9956,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999884936030647e-05,
+      "loss": 0.9985,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998630648662304e-05,
+      "loss": 1.0135,
+      "step": 107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998392920422897e-05,
+      "loss": 1.0056,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998136176040385e-05,
+      "loss": 0.9693,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9997860416003086e-05,
+      "loss": 0.9946,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9997565640835462e-05,
+      "loss": 1.0038,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999725185109816e-05,
+      "loss": 1.0056,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999691904738798e-05,
+      "loss": 0.9862,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9996567230337888e-05,
+      "loss": 0.9854,
+      "step": 114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996196400617015e-05,
+      "loss": 0.979,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9995806558930647e-05,
+      "loss": 0.9778,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9995397706020227e-05,
+      "loss": 1.0034,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9994969842663368e-05,
+      "loss": 0.995,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999452296967383e-05,
+      "loss": 0.9927,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9994057087901526e-05,
+      "loss": 1.004,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9993572198232528e-05,
+      "loss": 0.9946,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999306830158906e-05,
+      "loss": 1.0066,
+      "step": 122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9992545398929488e-05,
+      "loss": 0.9626,
+      "step": 123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9992003491248338e-05,
+      "loss": 1.0083,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9991442579576268e-05,
+      "loss": 0.9809,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999086266498009e-05,
+      "loss": 0.9915,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9990263748562754e-05,
+      "loss": 0.9763,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9989645831463352e-05,
+      "loss": 0.9745,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9989008914857115e-05,
+      "loss": 0.9734,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9988352999955407e-05,
+      "loss": 0.9847,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998767808800572e-05,
+      "loss": 1.0072,
+      "step": 131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9986984180291688e-05,
+      "loss": 1.0006,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9986271278133066e-05,
+      "loss": 0.9877,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998553938288574e-05,
+      "loss": 0.9584,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998478849594171e-05,
+      "loss": 0.9812,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9984018618729107e-05,
+      "loss": 1.0122,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9983229752712178e-05,
+      "loss": 0.9799,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998242189939128e-05,
+      "loss": 0.9948,
+      "step": 138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9981595060302883e-05,
+      "loss": 0.9882,
+      "step": 139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9980749237019576e-05,
+      "loss": 1.0044,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9979884431150037e-05,
+      "loss": 0.9617,
+      "step": 141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9979000644339067e-05,
+      "loss": 0.9794,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9978097878267554e-05,
+      "loss": 0.9804,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9977176134652482e-05,
+      "loss": 0.9694,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.997623541524694e-05,
+      "loss": 0.9682,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9975275721840105e-05,
+      "loss": 0.9545,
+      "step": 146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9974297056257223e-05,
+      "loss": 0.9859,
+      "step": 147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.997329942035965e-05,
+      "loss": 0.9864,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.997228281604481e-05,
+      "loss": 0.9902,
+      "step": 149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9971247245246194e-05,
+      "loss": 1.0059,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.997019270993338e-05,
+      "loss": 0.9901,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9969119212112005e-05,
+      "loss": 0.9865,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.996802675382379e-05,
+      "loss": 0.9762,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9966915337146484e-05,
+      "loss": 0.9687,
+      "step": 154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9965784964193925e-05,
+      "loss": 0.9613,
+      "step": 155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9964635637115988e-05,
+      "loss": 0.9625,
+      "step": 156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9963467358098607e-05,
+      "loss": 0.9396,
+      "step": 157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9962280129363746e-05,
+      "loss": 0.9731,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9961073953169422e-05,
+      "loss": 0.9582,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9959848831809688e-05,
+      "loss": 0.9309,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9958604767614626e-05,
+      "loss": 0.988,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9957341762950346e-05,
+      "loss": 0.9342,
+      "step": 162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9956059820218982e-05,
+      "loss": 0.9972,
+      "step": 163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9954758941858685e-05,
+      "loss": 0.9445,
+      "step": 164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9953439130343627e-05,
+      "loss": 0.9711,
+      "step": 165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9952100388183973e-05,
+      "loss": 0.9613,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9950742717925916e-05,
+      "loss": 0.9613,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9949366122151627e-05,
+      "loss": 0.9619,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9947970603479286e-05,
+      "loss": 0.9873,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9946556164563055e-05,
+      "loss": 0.9672,
+      "step": 170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9945122808093088e-05,
+      "loss": 0.9687,
+      "step": 171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9943670536795504e-05,
+      "loss": 0.9848,
+      "step": 172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.994219935343242e-05,
+      "loss": 0.9629,
+      "step": 173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99407092608019e-05,
+      "loss": 0.9584,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9939200261737986e-05,
+      "loss": 0.9649,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.993767235911067e-05,
+      "loss": 0.9527,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99361255558259e-05,
+      "loss": 0.9832,
+      "step": 177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9934559854825574e-05,
+      "loss": 0.9724,
+      "step": 178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9932975259087533e-05,
+      "loss": 0.9466,
+      "step": 179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9931371771625545e-05,
+      "loss": 0.9431,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9929749395489316e-05,
+      "loss": 0.9612,
+      "step": 181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992810813376448e-05,
+      "loss": 0.9508,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992644798957258e-05,
+      "loss": 0.9415,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924768966071087e-05,
+      "loss": 0.9516,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923071066453356e-05,
+      "loss": 0.973,
+      "step": 185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9921354293948666e-05,
+      "loss": 0.9281,
+      "step": 186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9919618651822174e-05,
+      "loss": 0.9668,
+      "step": 187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9917864143374942e-05,
+      "loss": 0.9713,
+      "step": 188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9916090771943893e-05,
+      "loss": 0.9621,
+      "step": 189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9914298540901845e-05,
+      "loss": 0.9716,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9912487453657478e-05,
+      "loss": 0.9318,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991065751365533e-05,
+      "loss": 0.9341,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9908808724375807e-05,
+      "loss": 0.9497,
+      "step": 193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.990694108933515e-05,
+      "loss": 0.9681,
+      "step": 194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.990505461208546e-05,
+      "loss": 0.9289,
+      "step": 195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.990314929621466e-05,
+      "loss": 0.9432,
+      "step": 196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.990122514534651e-05,
+      "loss": 0.9455,
+      "step": 197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.989928216314059e-05,
+      "loss": 0.9737,
+      "step": 198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9897320353292304e-05,
+      "loss": 0.977,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9895339719532848e-05,
+      "loss": 0.9488,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.989334026562923e-05,
+      "loss": 0.9708,
+      "step": 201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9891321995384254e-05,
+      "loss": 0.9761,
+      "step": 202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.988928491263651e-05,
+      "loss": 0.9425,
+      "step": 203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.988722902126036e-05,
+      "loss": 0.9622,
+      "step": 204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.988515432516595e-05,
+      "loss": 0.953,
+      "step": 205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9883060828299187e-05,
+      "loss": 0.9425,
+      "step": 206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.988094853464173e-05,
+      "loss": 0.9659,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9878817448210998e-05,
+      "loss": 0.9257,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.987666757306014e-05,
+      "loss": 0.9447,
+      "step": 209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9874498913278054e-05,
+      "loss": 0.9591,
+      "step": 210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.987231147298935e-05,
+      "loss": 0.9217,
+      "step": 211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9870105256354367e-05,
+      "loss": 0.9525,
+      "step": 212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.986788026756916e-05,
+      "loss": 0.954,
+      "step": 213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9865636510865466e-05,
+      "loss": 0.9624,
+      "step": 214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.986337399051074e-05,
+      "loss": 0.9416,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.986109271080811e-05,
+      "loss": 0.9483,
+      "step": 216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9858792676096395e-05,
+      "loss": 0.9335,
+      "step": 217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9856473890750067e-05,
+      "loss": 0.936,
+      "step": 218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.985413635917928e-05,
+      "loss": 0.9771,
+      "step": 219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.985178008582982e-05,
+      "loss": 0.9358,
+      "step": 220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9849405075183136e-05,
+      "loss": 0.9546,
+      "step": 221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9847011331756315e-05,
+      "loss": 0.9571,
+      "step": 222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9844598860102057e-05,
+      "loss": 0.9538,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.984216766480869e-05,
+      "loss": 0.9795,
+      "step": 224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.983971775050016e-05,
+      "loss": 0.9397,
+      "step": 225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9837249121836e-05,
+      "loss": 0.9334,
+      "step": 226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.983476178351135e-05,
+      "loss": 0.9456,
+      "step": 227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9832255740256926e-05,
+      "loss": 0.9642,
+      "step": 228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.982973099683902e-05,
+      "loss": 0.957,
+      "step": 229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9827187558059492e-05,
+      "loss": 0.9557,
+      "step": 230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.982462542875576e-05,
+      "loss": 0.9086,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9822044613800793e-05,
+      "loss": 0.9432,
+      "step": 232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.981944511810309e-05,
+      "loss": 0.9622,
+      "step": 233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9816826946606686e-05,
+      "loss": 0.9462,
+      "step": 234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9814190104291128e-05,
+      "loss": 0.9231,
+      "step": 235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9811534596171483e-05,
+      "loss": 0.9376,
+      "step": 236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9808860427298323e-05,
+      "loss": 0.9512,
+      "step": 237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9806167602757688e-05,
+      "loss": 0.935,
+      "step": 238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9803456127671122e-05,
+      "loss": 0.9497,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.980072600719564e-05,
+      "loss": 0.9221,
+      "step": 240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9797977246523697e-05,
+      "loss": 0.9381,
+      "step": 241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.979520985088323e-05,
+      "loss": 0.9642,
+      "step": 242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9792423825537602e-05,
+      "loss": 0.9308,
+      "step": 243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9789619175785604e-05,
+      "loss": 0.9334,
+      "step": 244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.978679590696146e-05,
+      "loss": 0.9275,
+      "step": 245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97839540244348e-05,
+      "loss": 0.9258,
+      "step": 246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9781093533610655e-05,
+      "loss": 0.9465,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9778214439929453e-05,
+      "loss": 0.927,
+      "step": 248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9775316748867e-05,
+      "loss": 0.9282,
+      "step": 249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9772400465934468e-05,
+      "loss": 0.9568,
+      "step": 250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.97694655966784e-05,
+      "loss": 0.9256,
+      "step": 251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.976651214668068e-05,
+      "loss": 0.9267,
+      "step": 252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9763540121558526e-05,
+      "loss": 0.9174,
+      "step": 253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9760549526964505e-05,
+      "loss": 0.9346,
+      "step": 254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.975754036858648e-05,
+      "loss": 0.9255,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9754512652147632e-05,
+      "loss": 0.942,
+      "step": 256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9751466383406434e-05,
+      "loss": 0.9342,
+      "step": 257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9748401568156645e-05,
+      "loss": 0.9335,
+      "step": 258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.97453182122273e-05,
+      "loss": 0.9326,
+      "step": 259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9742216321482698e-05,
+      "loss": 0.9336,
+      "step": 260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9739095901822384e-05,
+      "loss": 0.922,
+      "step": 261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.973595695918114e-05,
+      "loss": 0.9543,
+      "step": 262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9732799499528993e-05,
+      "loss": 0.9252,
+      "step": 263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9729623528871178e-05,
+      "loss": 0.958,
+      "step": 264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.972642905324813e-05,
+      "loss": 0.9599,
+      "step": 265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9723216078735495e-05,
+      "loss": 0.9147,
+      "step": 266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9719984611444086e-05,
+      "loss": 0.9138,
+      "step": 267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9716734657519896e-05,
+      "loss": 0.9464,
+      "step": 268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.971346622314408e-05,
+      "loss": 0.9438,
+      "step": 269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.971017931453294e-05,
+      "loss": 0.932,
+      "step": 270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9706873937937908e-05,
+      "loss": 0.9205,
+      "step": 271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.970355009964555e-05,
+      "loss": 0.9476,
+      "step": 272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.970020780597754e-05,
+      "loss": 0.9302,
+      "step": 273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.969684706329065e-05,
+      "loss": 0.9119,
+      "step": 274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9693467877976745e-05,
+      "loss": 0.951,
+      "step": 275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9690070256462767e-05,
+      "loss": 0.9773,
+      "step": 276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.968665420521072e-05,
+      "loss": 0.932,
+      "step": 277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9683219730717658e-05,
+      "loss": 0.9197,
+      "step": 278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9679766839515677e-05,
+      "loss": 0.9243,
+      "step": 279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96762955381719e-05,
+      "loss": 0.945,
+      "step": 280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9672805833288465e-05,
+      "loss": 0.9157,
+      "step": 281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.966929773150251e-05,
+      "loss": 0.929,
+      "step": 282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9665771239486163e-05,
+      "loss": 0.9445,
+      "step": 283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9662226363946534e-05,
+      "loss": 0.9293,
+      "step": 284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.965866311162569e-05,
+      "loss": 0.9472,
+      "step": 285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9655081489300646e-05,
+      "loss": 0.9103,
+      "step": 286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.965148150378337e-05,
+      "loss": 0.941,
+      "step": 287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9647863161920747e-05,
+      "loss": 0.9211,
+      "step": 288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9644226470594567e-05,
+      "loss": 0.9194,
+      "step": 289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9640571436721524e-05,
+      "loss": 0.9219,
+      "step": 290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.963689806725321e-05,
+      "loss": 0.9213,
+      "step": 291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.963320636917607e-05,
+      "loss": 0.9437,
+      "step": 292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9629496349511423e-05,
+      "loss": 0.9327,
+      "step": 293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.962576801531543e-05,
+      "loss": 0.93,
+      "step": 294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9622021373679085e-05,
+      "loss": 0.9287,
+      "step": 295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.961825643172819e-05,
+      "loss": 0.9137,
+      "step": 296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9614473196623377e-05,
+      "loss": 0.9097,
+      "step": 297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9610671675560054e-05,
+      "loss": 0.9169,
+      "step": 298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9606851875768404e-05,
+      "loss": 0.9305,
+      "step": 299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9603013804513385e-05,
+      "loss": 0.9336,
+      "step": 300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9599157469094696e-05,
+      "loss": 0.9549,
+      "step": 301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9595282876846785e-05,
+      "loss": 0.9209,
+      "step": 302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9591390035138812e-05,
+      "loss": 0.9132,
+      "step": 303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.958747895137465e-05,
+      "loss": 0.9188,
+      "step": 304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9583549632992872e-05,
+      "loss": 0.9158,
+      "step": 305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9579602087466726e-05,
+      "loss": 0.9424,
+      "step": 306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9575636322304122e-05,
+      "loss": 0.9398,
+      "step": 307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9571652345047632e-05,
+      "loss": 0.9084,
+      "step": 308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.956765016327446e-05,
+      "loss": 0.9229,
+      "step": 309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.956362978459644e-05,
+      "loss": 0.9358,
+      "step": 310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9559591216660007e-05,
+      "loss": 0.9316,
+      "step": 311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9555534467146194e-05,
+      "loss": 0.9104,
+      "step": 312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9551459543770614e-05,
+      "loss": 0.8928,
+      "step": 313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9547366454283447e-05,
+      "loss": 0.9528,
+      "step": 314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9543255206469417e-05,
+      "loss": 0.9328,
+      "step": 315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.953912580814779e-05,
+      "loss": 0.9272,
+      "step": 316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9534978267172357e-05,
+      "loss": 0.914,
+      "step": 317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9530812591431403e-05,
+      "loss": 0.9044,
+      "step": 318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.952662878884771e-05,
+      "loss": 0.8994,
+      "step": 319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.952242686737854e-05,
+      "loss": 0.9459,
+      "step": 320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.95182068350156e-05,
+      "loss": 0.9199,
+      "step": 321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.951396869978507e-05,
+      "loss": 0.9258,
+      "step": 322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9509712469747535e-05,
+      "loss": 0.9449,
+      "step": 323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9505438152998005e-05,
+      "loss": 0.9069,
+      "step": 324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9501145757665884e-05,
+      "loss": 0.9179,
+      "step": 325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9496835291914972e-05,
+      "loss": 0.9168,
+      "step": 326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9492506763943426e-05,
+      "loss": 0.9289,
+      "step": 327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9488160181983763e-05,
+      "loss": 0.9098,
+      "step": 328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9483795554302827e-05,
+      "loss": 0.9042,
+      "step": 329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9479412889201796e-05,
+      "loss": 0.9175,
+      "step": 330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9475012195016148e-05,
+      "loss": 0.9149,
+      "step": 331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.947059348011565e-05,
+      "loss": 0.9149,
+      "step": 332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9466156752904344e-05,
+      "loss": 0.8985,
+      "step": 333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.946170202182053e-05,
+      "loss": 0.9222,
+      "step": 334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.945722929533675e-05,
+      "loss": 0.9138,
+      "step": 335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9452738581959775e-05,
+      "loss": 0.9146,
+      "step": 336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9448229890230574e-05,
+      "loss": 0.9004,
+      "step": 337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9443703228724327e-05,
+      "loss": 0.9188,
+      "step": 338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9439158606050377e-05,
+      "loss": 0.9147,
+      "step": 339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9434596030852234e-05,
+      "loss": 0.9046,
+      "step": 340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9430015511807552e-05,
+      "loss": 0.9348,
+      "step": 341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.942541705762811e-05,
+      "loss": 0.9293,
+      "step": 342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9420800677059797e-05,
+      "loss": 0.9332,
+      "step": 343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9416166378882604e-05,
+      "loss": 0.9082,
+      "step": 344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9411514171910595e-05,
+      "loss": 0.9015,
+      "step": 345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9406844064991892e-05,
+      "loss": 0.9277,
+      "step": 346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.940215606700867e-05,
+      "loss": 0.9195,
+      "step": 347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9397450186877117e-05,
+      "loss": 0.9082,
+      "step": 348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.939272643354745e-05,
+      "loss": 0.9159,
+      "step": 349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9387984816003868e-05,
+      "loss": 0.8845,
+      "step": 350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.938322534326454e-05,
+      "loss": 0.915,
+      "step": 351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.937844802438161e-05,
+      "loss": 0.9067,
+      "step": 352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9373652868441155e-05,
+      "loss": 0.9209,
+      "step": 353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9368839884563175e-05,
+      "loss": 0.9384,
+      "step": 354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.936400908190158e-05,
+      "loss": 0.8906,
+      "step": 355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.935916046964418e-05,
+      "loss": 0.9128,
+      "step": 356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9354294057012635e-05,
+      "loss": 0.908,
+      "step": 357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9349409853262474e-05,
+      "loss": 0.9341,
+      "step": 358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9344507867683065e-05,
+      "loss": 0.9417,
+      "step": 359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9339588109597597e-05,
+      "loss": 0.9092,
+      "step": 360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9334650588363047e-05,
+      "loss": 0.9371,
+      "step": 361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9329695313370188e-05,
+      "loss": 0.8856,
+      "step": 362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.932472229404356e-05,
+      "loss": 0.8909,
+      "step": 363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9319731539841445e-05,
+      "loss": 0.8955,
+      "step": 364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9314723060255857e-05,
+      "loss": 0.9132,
+      "step": 365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9309696864812523e-05,
+      "loss": 0.9164,
+      "step": 366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9304652963070868e-05,
+      "loss": 0.9025,
+      "step": 367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9299591364623987e-05,
+      "loss": 0.9206,
+      "step": 368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9294512079098636e-05,
+      "loss": 0.9451,
+      "step": 369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9289415116155207e-05,
+      "loss": 0.9042,
+      "step": 370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.928430048548772e-05,
+      "loss": 0.9101,
+      "step": 371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.927916819682379e-05,
+      "loss": 0.8794,
+      "step": 372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9274018259924618e-05,
+      "loss": 0.8981,
+      "step": 373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9268850684584975e-05,
+      "loss": 0.8975,
+      "step": 374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9263665480633177e-05,
+      "loss": 0.9116,
+      "step": 375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9258462657931064e-05,
+      "loss": 0.8874,
+      "step": 376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9253242226373986e-05,
+      "loss": 0.872,
+      "step": 377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9248004195890796e-05,
+      "loss": 0.909,
+      "step": 378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9242748576443792e-05,
+      "loss": 0.9167,
+      "step": 379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.923747537802876e-05,
+      "loss": 0.9047,
+      "step": 380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9232184610674886e-05,
+      "loss": 0.8909,
+      "step": 381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9226876284444795e-05,
+      "loss": 0.9321,
+      "step": 382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9221550409434496e-05,
+      "loss": 0.9289,
+      "step": 383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9216206995773373e-05,
+      "loss": 0.8834,
+      "step": 384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.921084605362418e-05,
+      "loss": 0.9221,
+      "step": 385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9205467593182998e-05,
+      "loss": 0.9352,
+      "step": 386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.920007162467922e-05,
+      "loss": 0.8981,
+      "step": 387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9194658158375553e-05,
+      "loss": 0.8973,
+      "step": 388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9189227204567977e-05,
+      "loss": 0.9233,
+      "step": 389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.918377877358573e-05,
+      "loss": 0.8806,
+      "step": 390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.917831287579129e-05,
+      "loss": 0.9191,
+      "step": 391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9172829521580358e-05,
+      "loss": 0.9011,
+      "step": 392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9167328721381835e-05,
+      "loss": 0.9209,
+      "step": 393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9161810485657807e-05,
+      "loss": 0.9088,
+      "step": 394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.915627482490351e-05,
+      "loss": 0.9122,
+      "step": 395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.915072174964733e-05,
+      "loss": 0.9123,
+      "step": 396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9145151270450773e-05,
+      "loss": 0.884,
+      "step": 397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9139563397908443e-05,
+      "loss": 0.9058,
+      "step": 398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9133958142648028e-05,
+      "loss": 0.9292,
+      "step": 399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9128335515330275e-05,
+      "loss": 0.9134,
+      "step": 400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9122695526648968e-05,
+      "loss": 0.9215,
+      "step": 401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.911703818733092e-05,
+      "loss": 0.9035,
+      "step": 402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.911136350813593e-05,
+      "loss": 0.9261,
+      "step": 403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.910567149985679e-05,
+      "loss": 0.914,
+      "step": 404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.909996217331924e-05,
+      "loss": 0.8913,
+      "step": 405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9094235539381965e-05,
+      "loss": 0.9326,
+      "step": 406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9088491608936564e-05,
+      "loss": 0.8968,
+      "step": 407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.908273039290753e-05,
+      "loss": 0.9046,
+      "step": 408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9076951902252238e-05,
+      "loss": 0.9075,
+      "step": 409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9071156147960912e-05,
+      "loss": 0.893,
+      "step": 410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9065343141056616e-05,
+      "loss": 0.9237,
+      "step": 411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.905951289259522e-05,
+      "loss": 0.9066,
+      "step": 412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9053665413665397e-05,
+      "loss": 0.9137,
+      "step": 413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9047800715388578e-05,
+      "loss": 0.8853,
+      "step": 414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.904191880891895e-05,
+      "loss": 0.9168,
+      "step": 415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.903601970544343e-05,
+      "loss": 0.9233,
+      "step": 416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9030103416181637e-05,
+      "loss": 0.8915,
+      "step": 417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9024169952385887e-05,
+      "loss": 0.9097,
+      "step": 418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9018219325341146e-05,
+      "loss": 0.9165,
+      "step": 419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9012251546365033e-05,
+      "loss": 0.9207,
+      "step": 420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9006266626807788e-05,
+      "loss": 0.8928,
+      "step": 421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9000264578052244e-05,
+      "loss": 0.9091,
+      "step": 422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8994245411513825e-05,
+      "loss": 0.8946,
+      "step": 423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8988209138640496e-05,
+      "loss": 0.8954,
+      "step": 424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8982155770912772e-05,
+      "loss": 0.9082,
+      "step": 425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8976085319843668e-05,
+      "loss": 0.8897,
+      "step": 426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.89699977969787e-05,
+      "loss": 0.9085,
+      "step": 427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8963893213895844e-05,
+      "loss": 0.89,
+      "step": 428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8957771582205536e-05,
+      "loss": 0.8951,
+      "step": 429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8951632913550625e-05,
+      "loss": 0.89,
+      "step": 430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8945477219606367e-05,
+      "loss": 0.8862,
+      "step": 431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.89393045120804e-05,
+      "loss": 0.9058,
+      "step": 432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8933114802712726e-05,
+      "loss": 0.8961,
+      "step": 433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8926908103275663e-05,
+      "loss": 0.9204,
+      "step": 434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8920684425573865e-05,
+      "loss": 0.9098,
+      "step": 435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8914443781444273e-05,
+      "loss": 0.8867,
+      "step": 436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8908186182756084e-05,
+      "loss": 0.8762,
+      "step": 437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.890191164141076e-05,
+      "loss": 0.9006,
+      "step": 438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.889562016934196e-05,
+      "loss": 0.8832,
+      "step": 439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8889311778515577e-05,
+      "loss": 0.9193,
+      "step": 440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8882986480929658e-05,
+      "loss": 0.9148,
+      "step": 441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8876644288614417e-05,
+      "loss": 0.8919,
+      "step": 442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.887028521363219e-05,
+      "loss": 0.9054,
+      "step": 443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.886390926807743e-05,
+      "loss": 0.8911,
+      "step": 444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.885751646407668e-05,
+      "loss": 0.9038,
+      "step": 445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8851106813788537e-05,
+      "loss": 0.8997,
+      "step": 446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.884468032940364e-05,
+      "loss": 0.918,
+      "step": 447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.883823702314466e-05,
+      "loss": 0.8851,
+      "step": 448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8831776907266235e-05,
+      "loss": 0.8954,
+      "step": 449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8825299994055e-05,
+      "loss": 0.8852,
+      "step": 450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8818806295829516e-05,
+      "loss": 0.896,
+      "step": 451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8812295824940284e-05,
+      "loss": 0.9127,
+      "step": 452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.88057685937697e-05,
+      "loss": 0.8899,
+      "step": 453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.879922461473203e-05,
+      "loss": 0.8797,
+      "step": 454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.87926639002734e-05,
+      "loss": 0.8876,
+      "step": 455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8786086462871764e-05,
+      "loss": 0.9196,
+      "step": 456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.877949231503688e-05,
+      "loss": 0.9034,
+      "step": 457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8772881469310293e-05,
+      "loss": 0.9233,
+      "step": 458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.87662539382653e-05,
+      "loss": 0.8956,
+      "step": 459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8759609734506935e-05,
+      "loss": 0.9225,
+      "step": 460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8752948870671942e-05,
+      "loss": 0.9015,
+      "step": 461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8746271359428752e-05,
+      "loss": 0.9209,
+      "step": 462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8739577213477456e-05,
+      "loss": 0.8963,
+      "step": 463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.873286644554978e-05,
+      "loss": 0.8775,
+      "step": 464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8726139068409075e-05,
+      "loss": 0.9067,
+      "step": 465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8719395094850267e-05,
+      "loss": 0.8907,
+      "step": 466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.871263453769986e-05,
+      "loss": 0.8898,
+      "step": 467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8705857409815887e-05,
+      "loss": 0.9054,
+      "step": 468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8699063724087905e-05,
+      "loss": 0.8808,
+      "step": 469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8692253493436962e-05,
+      "loss": 0.8916,
+      "step": 470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.868542673081557e-05,
+      "loss": 0.9066,
+      "step": 471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8678583449207685e-05,
+      "loss": 0.8933,
+      "step": 472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8671723661628683e-05,
+      "loss": 0.8896,
+      "step": 473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8664847381125327e-05,
+      "loss": 0.9074,
+      "step": 474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8657954620775757e-05,
+      "loss": 0.911,
+      "step": 475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.865104539368945e-05,
+      "loss": 0.8993,
+      "step": 476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8644119713007197e-05,
+      "loss": 0.8925,
+      "step": 477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8637177591901096e-05,
+      "loss": 0.8738,
+      "step": 478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8630219043574504e-05,
+      "loss": 0.8911,
+      "step": 479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8623244081262022e-05,
+      "loss": 0.8848,
+      "step": 480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.861625271822947e-05,
+      "loss": 0.8986,
+      "step": 481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.860924496777386e-05,
+      "loss": 0.9051,
+      "step": 482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8602220843223377e-05,
+      "loss": 0.9231,
+      "step": 483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8595180357937336e-05,
+      "loss": 0.8871,
+      "step": 484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.858812352530618e-05,
+      "loss": 0.9075,
+      "step": 485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8581050358751444e-05,
+      "loss": 0.9086,
+      "step": 486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.857396087172572e-05,
+      "loss": 0.9005,
+      "step": 487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8566855077712647e-05,
+      "loss": 0.9124,
+      "step": 488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8559732990226874e-05,
+      "loss": 0.8944,
+      "step": 489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8552594622814045e-05,
+      "loss": 0.9002,
+      "step": 490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.854543998905076e-05,
+      "loss": 0.8933,
+      "step": 491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8538269102544563e-05,
+      "loss": 0.9103,
+      "step": 492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8531081976933904e-05,
+      "loss": 0.881,
+      "step": 493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8523878625888122e-05,
+      "loss": 0.8961,
+      "step": 494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8516659063107413e-05,
+      "loss": 0.8955,
+      "step": 495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8509423302322805e-05,
+      "loss": 0.8837,
+      "step": 496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8502171357296144e-05,
+      "loss": 0.8956,
+      "step": 497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.849490324182004e-05,
+      "loss": 0.8842,
+      "step": 498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8487618969717872e-05,
+      "loss": 0.8848,
+      "step": 499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8480318554843744e-05,
+      "loss": 0.8903,
+      "step": 500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8473002011082453e-05,
+      "loss": 0.8818,
+      "step": 501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8465669352349487e-05,
+      "loss": 0.851,
+      "step": 502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8458320592590976e-05,
+      "loss": 0.9054,
+      "step": 503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8450955745783667e-05,
+      "loss": 0.8893,
+      "step": 504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8443574825934915e-05,
+      "loss": 0.9287,
+      "step": 505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8436177847082636e-05,
+      "loss": 0.8935,
+      "step": 506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8428764823295293e-05,
+      "loss": 0.9067,
+      "step": 507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8421335768671868e-05,
+      "loss": 0.8949,
+      "step": 508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8413890697341815e-05,
+      "loss": 0.8699,
+      "step": 509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.840642962346508e-05,
+      "loss": 0.8966,
+      "step": 510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8398952561232016e-05,
+      "loss": 0.904,
+      "step": 511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8391459524863403e-05,
+      "loss": 0.8795,
+      "step": 512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8383950528610395e-05,
+      "loss": 0.8705,
+      "step": 513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8376425586754502e-05,
+      "loss": 0.8809,
+      "step": 514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.836888471360756e-05,
+      "loss": 0.8939,
+      "step": 515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8361327923511712e-05,
+      "loss": 0.9027,
+      "step": 516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.835375523083936e-05,
+      "loss": 0.8844,
+      "step": 517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8346166649993162e-05,
+      "loss": 0.887,
+      "step": 518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8338562195405995e-05,
+      "loss": 0.897,
+      "step": 519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8330941881540917e-05,
+      "loss": 0.899,
+      "step": 520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.832330572289116e-05,
+      "loss": 0.8848,
+      "step": 521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8315653733980086e-05,
+      "loss": 0.8973,
+      "step": 522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8307985929361162e-05,
+      "loss": 0.8942,
+      "step": 523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8300302323617944e-05,
+      "loss": 0.8972,
+      "step": 524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8292602931364026e-05,
+      "loss": 0.8888,
+      "step": 525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8284887767243046e-05,
+      "loss": 0.9092,
+      "step": 526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8277156845928624e-05,
+      "loss": 0.8699,
+      "step": 527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8269410182124354e-05,
+      "loss": 0.8733,
+      "step": 528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.826164779056377e-05,
+      "loss": 0.923,
+      "step": 529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8253869686010324e-05,
+      "loss": 0.8816,
+      "step": 530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.824607588325734e-05,
+      "loss": 0.8766,
+      "step": 531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8238266397128014e-05,
+      "loss": 0.9013,
+      "step": 532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8230441242475365e-05,
+      "loss": 0.879,
+      "step": 533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.82226004341822e-05,
+      "loss": 0.8988,
+      "step": 534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.821474398716112e-05,
+      "loss": 0.874,
+      "step": 535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8206871916354453e-05,
+      "loss": 0.8795,
+      "step": 536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8198984236734246e-05,
+      "loss": 0.919,
+      "step": 537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.819108096330224e-05,
+      "loss": 0.8943,
+      "step": 538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8183162111089826e-05,
+      "loss": 0.8748,
+      "step": 539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8175227695158028e-05,
+      "loss": 0.8988,
+      "step": 540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.816727773059747e-05,
+      "loss": 0.876,
+      "step": 541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8159312232528342e-05,
+      "loss": 0.8783,
+      "step": 542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.815133121610039e-05,
+      "loss": 0.8847,
+      "step": 543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.814333469649287e-05,
+      "loss": 0.9016,
+      "step": 544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.813532268891452e-05,
+      "loss": 0.8971,
+      "step": 545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.812729520860354e-05,
+      "loss": 0.8913,
+      "step": 546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8119252270827546e-05,
+      "loss": 0.8705,
+      "step": 547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.811119389088357e-05,
+      "loss": 0.8989,
+      "step": 548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8103120084098003e-05,
+      "loss": 0.8941,
+      "step": 549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.809503086582658e-05,
+      "loss": 0.8896,
+      "step": 550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8086926251454347e-05,
+      "loss": 0.8901,
+      "step": 551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.807880625639563e-05,
+      "loss": 0.8821,
+      "step": 552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8070670896094015e-05,
+      "loss": 0.8936,
+      "step": 553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.80625201860223e-05,
+      "loss": 0.9096,
+      "step": 554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8054354141682485e-05,
+      "loss": 0.9003,
+      "step": 555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8046172778605736e-05,
+      "loss": 0.8815,
+      "step": 556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8037976112352348e-05,
+      "loss": 0.8955,
+      "step": 557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.802976415851172e-05,
+      "loss": 0.865,
+      "step": 558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.802153693270234e-05,
+      "loss": 0.8834,
+      "step": 559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8013294450571725e-05,
+      "loss": 0.896,
+      "step": 560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.800503672779642e-05,
+      "loss": 0.8951,
+      "step": 561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7996763780081954e-05,
+      "loss": 0.9067,
+      "step": 562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7988475623162807e-05,
+      "loss": 0.8994,
+      "step": 563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7980172272802398e-05,
+      "loss": 0.8974,
+      "step": 564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7971853744793026e-05,
+      "loss": 0.8836,
+      "step": 565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.796352005495587e-05,
+      "loss": 0.8656,
+      "step": 566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.795517121914094e-05,
+      "loss": 0.8945,
+      "step": 567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7946807253227054e-05,
+      "loss": 0.8527,
+      "step": 568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7938428173121806e-05,
+      "loss": 0.876,
+      "step": 569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7930033994761534e-05,
+      "loss": 0.8942,
+      "step": 570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7921624734111292e-05,
+      "loss": 0.8992,
+      "step": 571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.791320040716483e-05,
+      "loss": 0.8686,
+      "step": 572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7904761029944534e-05,
+      "loss": 0.8778,
+      "step": 573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7896306618501425e-05,
+      "loss": 0.8721,
+      "step": 574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7887837188915123e-05,
+      "loss": 0.8698,
+      "step": 575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.78793527572938e-05,
+      "loss": 0.8748,
+      "step": 576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7870853339774165e-05,
+      "loss": 0.8875,
+      "step": 577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7862338952521432e-05,
+      "loss": 0.8777,
+      "step": 578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7853809611729285e-05,
+      "loss": 0.8703,
+      "step": 579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7845265333619845e-05,
+      "loss": 0.8627,
+      "step": 580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.783670613444365e-05,
+      "loss": 0.8811,
+      "step": 581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7828132030479604e-05,
+      "loss": 0.9072,
+      "step": 582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7819543038034973e-05,
+      "loss": 0.8675,
+      "step": 583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7810939173445333e-05,
+      "loss": 0.8889,
+      "step": 584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.780232045307454e-05,
+      "loss": 0.8738,
+      "step": 585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7793686893314716e-05,
+      "loss": 0.8937,
+      "step": 586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.77850385105862e-05,
+      "loss": 0.8725,
+      "step": 587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7776375321337523e-05,
+      "loss": 0.8709,
+      "step": 588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7767697342045376e-05,
+      "loss": 0.9043,
+      "step": 589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.775900458921458e-05,
+      "loss": 0.8823,
+      "step": 590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.775029707937806e-05,
+      "loss": 0.9034,
+      "step": 591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7741574829096803e-05,
+      "loss": 0.8626,
+      "step": 592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7732837854959822e-05,
+      "loss": 0.8928,
+      "step": 593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7724086173584153e-05,
+      "loss": 0.8833,
+      "step": 594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7715319801614788e-05,
+      "loss": 0.8755,
+      "step": 595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7706538755724667e-05,
+      "loss": 0.8746,
+      "step": 596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7697743052614637e-05,
+      "loss": 0.8628,
+      "step": 597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7688932709013418e-05,
+      "loss": 0.8742,
+      "step": 598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.768010774167758e-05,
+      "loss": 0.8613,
+      "step": 599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7671268167391505e-05,
+      "loss": 0.8693,
+      "step": 600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7662414002967353e-05,
+      "loss": 0.885,
+      "step": 601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7653545265245043e-05,
+      "loss": 0.8836,
+      "step": 602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.76446619710922e-05,
+      "loss": 0.8528,
+      "step": 603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7635764137404136e-05,
+      "loss": 0.8895,
+      "step": 604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.762685178110382e-05,
+      "loss": 0.8816,
+      "step": 605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7617924919141843e-05,
+      "loss": 0.8765,
+      "step": 606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.760898356849638e-05,
+      "loss": 0.8849,
+      "step": 607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7600027746173168e-05,
+      "loss": 0.8925,
+      "step": 608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.759105746920546e-05,
+      "loss": 0.8843,
+      "step": 609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7582072754654005e-05,
+      "loss": 0.8528,
+      "step": 610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7573073619607012e-05,
+      "loss": 0.8762,
+      "step": 611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7564060081180113e-05,
+      "loss": 0.8497,
+      "step": 612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7555032156516342e-05,
+      "loss": 0.8872,
+      "step": 613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7545989862786087e-05,
+      "loss": 0.8856,
+      "step": 614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7536933217187062e-05,
+      "loss": 0.8848,
+      "step": 615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.752786223694429e-05,
+      "loss": 0.8871,
+      "step": 616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7518776939310045e-05,
+      "loss": 0.8799,
+      "step": 617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7509677341563837e-05,
+      "loss": 0.8694,
+      "step": 618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.750056346101237e-05,
+      "loss": 0.8579,
+      "step": 619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.749143531498952e-05,
+      "loss": 0.8879,
+      "step": 620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7482292920856285e-05,
+      "loss": 0.8786,
+      "step": 621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.747313629600077e-05,
+      "loss": 0.8711,
+      "step": 622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7463965457838145e-05,
+      "loss": 0.8841,
+      "step": 623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.745478042381061e-05,
+      "loss": 0.8571,
+      "step": 624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7445581211387358e-05,
+      "loss": 0.8811,
+      "step": 625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.743636783806456e-05,
+      "loss": 0.857,
+      "step": 626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.742714032136531e-05,
+      "loss": 0.9036,
+      "step": 627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.741789867883961e-05,
+      "loss": 0.8687,
+      "step": 628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7408642928064327e-05,
+      "loss": 0.88,
+      "step": 629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.739937308664316e-05,
+      "loss": 0.8685,
+      "step": 630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7390089172206594e-05,
+      "loss": 0.8835,
+      "step": 631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.73807912024119e-05,
+      "loss": 0.9031,
+      "step": 632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7371479194943068e-05,
+      "loss": 0.8672,
+      "step": 633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7362153167510797e-05,
+      "loss": 0.8657,
+      "step": 634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7352813137852437e-05,
+      "loss": 0.8915,
+      "step": 635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7343459123731983e-05,
+      "loss": 0.8775,
+      "step": 636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7334091142940018e-05,
+      "loss": 0.8956,
+      "step": 637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7324709213293696e-05,
+      "loss": 0.8972,
+      "step": 638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.731531335263669e-05,
+      "loss": 0.8736,
+      "step": 639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.730590357883918e-05,
+      "loss": 0.8883,
+      "step": 640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.72964799097978e-05,
+      "loss": 0.8563,
+      "step": 641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7287042363435614e-05,
+      "loss": 0.8843,
+      "step": 642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7277590957702082e-05,
+      "loss": 0.8553,
+      "step": 643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.726812571057302e-05,
+      "loss": 0.8679,
+      "step": 644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.725864664005057e-05,
+      "loss": 0.882,
+      "step": 645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7249153764163166e-05,
+      "loss": 0.8697,
+      "step": 646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.72396471009655e-05,
+      "loss": 0.8874,
+      "step": 647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.723012666853848e-05,
+      "loss": 0.8775,
+      "step": 648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.722059248498921e-05,
+      "loss": 0.8998,
+      "step": 649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.721104456845094e-05,
+      "loss": 0.899,
+      "step": 650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7201482937083048e-05,
+      "loss": 0.8743,
+      "step": 651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7191907609070994e-05,
+      "loss": 0.8789,
+      "step": 652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.718231860262628e-05,
+      "loss": 0.8732,
+      "step": 653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.717271593598643e-05,
+      "loss": 0.8757,
+      "step": 654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.716309962741494e-05,
+      "loss": 0.8729,
+      "step": 655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7153469695201278e-05,
+      "loss": 0.8643,
+      "step": 656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7143826157660788e-05,
+      "loss": 0.8786,
+      "step": 657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.713416903313471e-05,
+      "loss": 0.8917,
+      "step": 658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7124498339990124e-05,
+      "loss": 0.8658,
+      "step": 659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7114814096619916e-05,
+      "loss": 0.8795,
+      "step": 660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.710511632144274e-05,
+      "loss": 0.8652,
+      "step": 661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7095405032902987e-05,
+      "loss": 0.8847,
+      "step": 662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.708568024947075e-05,
+      "loss": 0.8334,
+      "step": 663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7075941989641792e-05,
+      "loss": 0.8682,
+      "step": 664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7066190271937504e-05,
+      "loss": 0.8787,
+      "step": 665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7056425114904868e-05,
+      "loss": 0.8678,
+      "step": 666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7046646537116432e-05,
+      "loss": 0.876,
+      "step": 667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7036854557170274e-05,
+      "loss": 0.8952,
+      "step": 668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.702704919368995e-05,
+      "loss": 0.8634,
+      "step": 669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7017230465324477e-05,
+      "loss": 0.907,
+      "step": 670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7007398390748296e-05,
+      "loss": 0.8827,
+      "step": 671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.699755298866122e-05,
+      "loss": 0.8856,
+      "step": 672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.698769427778842e-05,
+      "loss": 0.8959,
+      "step": 673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6977822276880368e-05,
+      "loss": 0.872,
+      "step": 674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.696793700471283e-05,
+      "loss": 0.848,
+      "step": 675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6958038480086792e-05,
+      "loss": 0.8633,
+      "step": 676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6948126721828466e-05,
+      "loss": 0.8743,
+      "step": 677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6938201748789214e-05,
+      "loss": 0.8662,
+      "step": 678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6928263579845546e-05,
+      "loss": 0.8743,
+      "step": 679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6918312233899058e-05,
+      "loss": 0.8841,
+      "step": 680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6908347729876423e-05,
+      "loss": 0.8639,
+      "step": 681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.689837008672932e-05,
+      "loss": 0.8824,
+      "step": 682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6888379323434433e-05,
+      "loss": 0.8687,
+      "step": 683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6878375458993397e-05,
+      "loss": 0.8531,
+      "step": 684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6868358512432755e-05,
+      "loss": 0.8865,
+      "step": 685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.685832850280394e-05,
+      "loss": 0.89,
+      "step": 686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6848285449183227e-05,
+      "loss": 0.8859,
+      "step": 687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6838229370671704e-05,
+      "loss": 0.8611,
+      "step": 688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6828160286395222e-05,
+      "loss": 0.8547,
+      "step": 689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.681807821550438e-05,
+      "loss": 0.8834,
+      "step": 690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.680798317717446e-05,
+      "loss": 0.8706,
+      "step": 691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6797875190605428e-05,
+      "loss": 0.8837,
+      "step": 692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.678775427502186e-05,
+      "loss": 0.8706,
+      "step": 693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6777620449672927e-05,
+      "loss": 0.8807,
+      "step": 694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6767473733832355e-05,
+      "loss": 0.8592,
+      "step": 695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.675731414679838e-05,
+      "loss": 0.8483,
+      "step": 696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.674714170789373e-05,
+      "loss": 0.856,
+      "step": 697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6736956436465573e-05,
+      "loss": 0.8903,
+      "step": 698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.672675835188547e-05,
+      "loss": 0.8966,
+      "step": 699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6716547473549372e-05,
+      "loss": 0.8456,
+      "step": 700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6706323820877545e-05,
+      "loss": 0.8882,
+      "step": 701
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6696087413314557e-05,
+      "loss": 0.8789,
+      "step": 702
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6685838270329243e-05,
+      "loss": 0.8772,
+      "step": 703
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6675576411414644e-05,
+      "loss": 0.8908,
+      "step": 704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6665301856088004e-05,
+      "loss": 0.8826,
+      "step": 705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6655014623890697e-05,
+      "loss": 0.8726,
+      "step": 706
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.664471473438822e-05,
+      "loss": 0.8612,
+      "step": 707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6634402207170134e-05,
+      "loss": 0.8757,
+      "step": 708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6624077061850047e-05,
+      "loss": 0.8547,
+      "step": 709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.661373931806555e-05,
+      "loss": 0.8637,
+      "step": 710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6603388995478214e-05,
+      "loss": 0.8599,
+      "step": 711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6593026113773516e-05,
+      "loss": 0.892,
+      "step": 712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6582650692660833e-05,
+      "loss": 0.8571,
+      "step": 713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6572262751873383e-05,
+      "loss": 0.8716,
+      "step": 714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.65618623111682e-05,
+      "loss": 0.8634,
+      "step": 715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6551449390326087e-05,
+      "loss": 0.8638,
+      "step": 716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.654102400915159e-05,
+      "loss": 0.855,
+      "step": 717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6530586187472944e-05,
+      "loss": 0.8796,
+      "step": 718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6520135945142057e-05,
+      "loss": 0.8685,
+      "step": 719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6509673302034444e-05,
+      "loss": 0.8623,
+      "step": 720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6499198278049228e-05,
+      "loss": 0.863,
+      "step": 721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6488710893109054e-05,
+      "loss": 0.8977,
+      "step": 722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6478211167160093e-05,
+      "loss": 0.8513,
+      "step": 723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.646769912017199e-05,
+      "loss": 0.8825,
+      "step": 724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6457174772137804e-05,
+      "loss": 0.8584,
+      "step": 725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6446638143074016e-05,
+      "loss": 0.8474,
+      "step": 726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6436089253020444e-05,
+      "loss": 0.861,
+      "step": 727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.642552812204023e-05,
+      "loss": 0.8652,
+      "step": 728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6414954770219807e-05,
+      "loss": 0.8599,
+      "step": 729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.640436921766884e-05,
+      "loss": 0.8628,
+      "step": 730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6393771484520203e-05,
+      "loss": 0.8593,
+      "step": 731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6383161590929933e-05,
+      "loss": 0.8662,
+      "step": 732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6372539557077204e-05,
+      "loss": 0.8584,
+      "step": 733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6361905403164275e-05,
+      "loss": 0.8661,
+      "step": 734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.635125914941645e-05,
+      "loss": 0.8651,
+      "step": 735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6340600816082056e-05,
+      "loss": 0.8746,
+      "step": 736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6329930423432383e-05,
+      "loss": 0.8446,
+      "step": 737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6319247991761672e-05,
+      "loss": 0.8458,
+      "step": 738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.630855354138705e-05,
+      "loss": 0.8928,
+      "step": 739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6297847092648508e-05,
+      "loss": 0.8491,
+      "step": 740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.628712866590885e-05,
+      "loss": 0.8788,
+      "step": 741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.627639828155367e-05,
+      "loss": 0.8532,
+      "step": 742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6265655959991295e-05,
+      "loss": 0.8952,
+      "step": 743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6254901721652764e-05,
+      "loss": 0.8634,
+      "step": 744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6244135586991784e-05,
+      "loss": 0.8624,
+      "step": 745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6233357576484666e-05,
+      "loss": 0.8491,
+      "step": 746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.622256771063034e-05,
+      "loss": 0.8691,
+      "step": 747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6211766009950255e-05,
+      "loss": 0.8462,
+      "step": 748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.620095249498839e-05,
+      "loss": 0.869,
+      "step": 749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6190127186311178e-05,
+      "loss": 0.8698,
+      "step": 750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6179290104507493e-05,
+      "loss": 0.8611,
+      "step": 751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.61684412701886e-05,
+      "loss": 0.8777,
+      "step": 752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6157580703988105e-05,
+      "loss": 0.8752,
+      "step": 753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6146708426561944e-05,
+      "loss": 0.8703,
+      "step": 754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6135824458588318e-05,
+      "loss": 0.8613,
+      "step": 755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6124928820767657e-05,
+      "loss": 0.883,
+      "step": 756
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6114021533822595e-05,
+      "loss": 0.8651,
+      "step": 757
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6103102618497922e-05,
+      "loss": 0.8556,
+      "step": 758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6092172095560538e-05,
+      "loss": 0.8611,
+      "step": 759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.608122998579942e-05,
+      "loss": 0.8623,
+      "step": 760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.607027631002559e-05,
+      "loss": 0.8862,
+      "step": 761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6059311089072064e-05,
+      "loss": 0.8791,
+      "step": 762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6048334343793814e-05,
+      "loss": 0.8602,
+      "step": 763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.603734609506772e-05,
+      "loss": 0.8531,
+      "step": 764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6026346363792565e-05,
+      "loss": 0.8605,
+      "step": 765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6015335170888953e-05,
+      "loss": 0.8793,
+      "step": 766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.600431253729929e-05,
+      "loss": 0.8857,
+      "step": 767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5993278483987743e-05,
+      "loss": 0.8849,
+      "step": 768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.59822330319402e-05,
+      "loss": 0.8619,
+      "step": 769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5971176202164228e-05,
+      "loss": 0.8597,
+      "step": 770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5960108015689027e-05,
+      "loss": 0.8746,
+      "step": 771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.594902849356541e-05,
+      "loss": 0.8734,
+      "step": 772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5937937656865733e-05,
+      "loss": 0.8774,
+      "step": 773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5926835526683884e-05,
+      "loss": 0.8712,
+      "step": 774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5915722124135227e-05,
+      "loss": 0.8711,
+      "step": 775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5904597470356557e-05,
+      "loss": 0.8763,
+      "step": 776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5893461586506087e-05,
+      "loss": 0.8625,
+      "step": 777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.588231449376337e-05,
+      "loss": 0.8665,
+      "step": 778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.587115621332928e-05,
+      "loss": 0.8574,
+      "step": 779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.585998676642598e-05,
+      "loss": 0.8585,
+      "step": 780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5848806174296863e-05,
+      "loss": 0.861,
+      "step": 781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5837614458206522e-05,
+      "loss": 0.8801,
+      "step": 782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5826411639440698e-05,
+      "loss": 0.8839,
+      "step": 783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.581519773930626e-05,
+      "loss": 0.8756,
+      "step": 784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5803972779131154e-05,
+      "loss": 0.8738,
+      "step": 785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5792736780264346e-05,
+      "loss": 0.8683,
+      "step": 786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5781489764075816e-05,
+      "loss": 0.8403,
+      "step": 787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.577023175195648e-05,
+      "loss": 0.8483,
+      "step": 788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5758962765318177e-05,
+      "loss": 0.8787,
+      "step": 789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5747682825593624e-05,
+      "loss": 0.8522,
+      "step": 790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.573639195423636e-05,
+      "loss": 0.9269,
+      "step": 791
+    }
+  ],
+  "max_steps": 2373,
+  "num_train_epochs": 3,
+  "total_flos": 1.0466956714212065e+18,
+  "trial_name": null,
+  "trial_params": null
+}