diff --git "a/ReCogDrive_VLM_2B/trainer_state.json" "b/ReCogDrive_VLM_2B/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/ReCogDrive_VLM_2B/trainer_state.json"
@@ -0,0 +1,27336 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.999629248197734,
+  "eval_steps": 500,
+  "global_step": 4551,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.771929824561404e-08,
+      "loss": 1.3343,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7543859649122808e-07,
+      "loss": 1.3382,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6315789473684213e-07,
+      "loss": 1.3555,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5087719298245616e-07,
+      "loss": 1.3234,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.385964912280702e-07,
+      "loss": 1.324,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 1.323,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.140350877192982e-07,
+      "loss": 1.4065,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.017543859649123e-07,
+      "loss": 1.3804,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.894736842105263e-07,
+      "loss": 1.3793,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.771929824561404e-07,
+      "loss": 1.3471,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.649122807017545e-07,
+      "loss": 1.3396,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 1.2943,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1403508771929824e-06,
+      "loss": 1.3156,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2280701754385965e-06,
+      "loss": 1.2983,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3157894736842106e-06,
+      "loss": 1.2835,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4035087719298246e-06,
+      "loss": 1.3129,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4912280701754387e-06,
+      "loss": 1.3097,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 1.2416,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.2439,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 1.1443,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8421052631578948e-06,
+      "loss": 1.1315,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.929824561403509e-06,
+      "loss": 1.165,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.017543859649123e-06,
+      "loss": 1.1549,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 1.1257,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.192982456140351e-06,
+      "loss": 1.1466,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.280701754385965e-06,
+      "loss": 1.1066,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.368421052631579e-06,
+      "loss": 1.1356,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.456140350877193e-06,
+      "loss": 1.0586,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5438596491228075e-06,
+      "loss": 1.062,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 1.0418,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7192982456140356e-06,
+      "loss": 1.0624,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8070175438596493e-06,
+      "loss": 1.0142,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8947368421052634e-06,
+      "loss": 1.0033,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9824561403508774e-06,
+      "loss": 1.0406,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0701754385964915e-06,
+      "loss": 1.0074,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.9907,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2456140350877197e-06,
+      "loss": 1.0032,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.9888,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.421052631578948e-06,
+      "loss": 0.9804,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 0.9807,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.596491228070176e-06,
+      "loss": 0.9643,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 0.9594,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7719298245614037e-06,
+      "loss": 0.9297,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.859649122807018e-06,
+      "loss": 0.9059,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.947368421052632e-06,
+      "loss": 0.8985,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.035087719298246e-06,
+      "loss": 0.899,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.12280701754386e-06,
+      "loss": 0.9205,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.896,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.298245614035088e-06,
+      "loss": 0.8695,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.385964912280702e-06,
+      "loss": 0.9065,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.473684210526316e-06,
+      "loss": 0.847,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.56140350877193e-06,
+      "loss": 0.8706,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.649122807017544e-06,
+      "loss": 0.8565,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 0.884,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.824561403508772e-06,
+      "loss": 0.8574,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.912280701754386e-06,
+      "loss": 0.8514,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-06,
+      "loss": 0.8751,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.087719298245615e-06,
+      "loss": 0.8887,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.175438596491229e-06,
+      "loss": 0.8092,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.8101,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.350877192982457e-06,
+      "loss": 0.8283,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.438596491228071e-06,
+      "loss": 0.8428,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.526315789473685e-06,
+      "loss": 0.8182,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6140350877192985e-06,
+      "loss": 0.8089,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.701754385964913e-06,
+      "loss": 0.8162,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 0.8474,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.877192982456141e-06,
+      "loss": 0.819,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.964912280701755e-06,
+      "loss": 0.8041,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0526315789473685e-06,
+      "loss": 0.8339,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.140350877192983e-06,
+      "loss": 0.7951,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.2280701754385975e-06,
+      "loss": 0.7485,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.8182,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.403508771929825e-06,
+      "loss": 0.7956,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.491228070175439e-06,
+      "loss": 0.7961,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.578947368421054e-06,
+      "loss": 0.8032,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7873,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.754385964912281e-06,
+      "loss": 0.7943,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 0.7868,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.92982456140351e-06,
+      "loss": 0.8205,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 0.805,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.1052631578947375e-06,
+      "loss": 0.8183,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.192982456140352e-06,
+      "loss": 0.7779,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.280701754385966e-06,
+      "loss": 0.7983,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.7796,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.456140350877194e-06,
+      "loss": 0.7999,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.5438596491228074e-06,
+      "loss": 0.8147,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.631578947368423e-06,
+      "loss": 0.8002,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.719298245614036e-06,
+      "loss": 0.789,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.80701754385965e-06,
+      "loss": 0.7837,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 0.803,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.982456140350877e-06,
+      "loss": 0.7918,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.070175438596492e-06,
+      "loss": 0.7817,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.157894736842106e-06,
+      "loss": 0.7882,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.24561403508772e-06,
+      "loss": 0.807,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.8169,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.7829,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.508771929824563e-06,
+      "loss": 0.7762,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.596491228070176e-06,
+      "loss": 0.7829,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.68421052631579e-06,
+      "loss": 0.7372,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.771929824561405e-06,
+      "loss": 0.7624,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.859649122807017e-06,
+      "loss": 0.7483,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 0.7657,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.035087719298246e-06,
+      "loss": 0.7806,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.12280701754386e-06,
+      "loss": 0.7822,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.210526315789474e-06,
+      "loss": 0.7797,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.298245614035088e-06,
+      "loss": 0.7587,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.385964912280703e-06,
+      "loss": 0.7725,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.761,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.56140350877193e-06,
+      "loss": 0.7392,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.649122807017545e-06,
+      "loss": 0.7763,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.736842105263159e-06,
+      "loss": 0.7622,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.824561403508772e-06,
+      "loss": 0.7467,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.912280701754386e-06,
+      "loss": 0.763,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.7543,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0087719298245615e-05,
+      "loss": 0.7162,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.017543859649123e-05,
+      "loss": 0.7528,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0263157894736844e-05,
+      "loss": 0.7233,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0350877192982459e-05,
+      "loss": 0.7612,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.043859649122807e-05,
+      "loss": 0.768,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.7729,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0614035087719299e-05,
+      "loss": 0.7396,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0701754385964913e-05,
+      "loss": 0.7495,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0789473684210528e-05,
+      "loss": 0.7341,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0877192982456142e-05,
+      "loss": 0.7314,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0964912280701755e-05,
+      "loss": 0.7383,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.105263157894737e-05,
+      "loss": 0.7503,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1140350877192983e-05,
+      "loss": 0.7332,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1228070175438597e-05,
+      "loss": 0.7043,
+      "step": 128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1315789473684212e-05,
+      "loss": 0.7489,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1403508771929826e-05,
+      "loss": 0.7726,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1491228070175439e-05,
+      "loss": 0.7493,
+      "step": 131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.769,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.7041,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1754385964912282e-05,
+      "loss": 0.7425,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1842105263157895e-05,
+      "loss": 0.7156,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.192982456140351e-05,
+      "loss": 0.7436,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2017543859649124e-05,
+      "loss": 0.7505,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2105263157894737e-05,
+      "loss": 0.7773,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2192982456140352e-05,
+      "loss": 0.7538,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2280701754385966e-05,
+      "loss": 0.7197,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.236842105263158e-05,
+      "loss": 0.7302,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2456140350877195e-05,
+      "loss": 0.7289,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2543859649122808e-05,
+      "loss": 0.7124,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.7472,
+      "step": 144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2719298245614035e-05,
+      "loss": 0.7526,
+      "step": 145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.280701754385965e-05,
+      "loss": 0.7733,
+      "step": 146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2894736842105264e-05,
+      "loss": 0.761,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2982456140350879e-05,
+      "loss": 0.7349,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3070175438596493e-05,
+      "loss": 0.7234,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3157894736842108e-05,
+      "loss": 0.7518,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3245614035087719e-05,
+      "loss": 0.728,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7341,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3421052631578948e-05,
+      "loss": 0.7277,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3508771929824562e-05,
+      "loss": 0.7596,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3596491228070177e-05,
+      "loss": 0.744,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.7392,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3771929824561406e-05,
+      "loss": 0.7247,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.385964912280702e-05,
+      "loss": 0.7207,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3947368421052631e-05,
+      "loss": 0.7169,
+      "step": 159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 0.7076,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.412280701754386e-05,
+      "loss": 0.7375,
+      "step": 161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4210526315789475e-05,
+      "loss": 0.7274,
+      "step": 162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.429824561403509e-05,
+      "loss": 0.7246,
+      "step": 163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4385964912280704e-05,
+      "loss": 0.7277,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4473684210526317e-05,
+      "loss": 0.7233,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4561403508771931e-05,
+      "loss": 0.7319,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4649122807017544e-05,
+      "loss": 0.7025,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.7282,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4824561403508773e-05,
+      "loss": 0.713,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4912280701754388e-05,
+      "loss": 0.7388,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7584,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5087719298245615e-05,
+      "loss": 0.7273,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.517543859649123e-05,
+      "loss": 0.7194,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5263157894736846e-05,
+      "loss": 0.7345,
+      "step": 174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.535087719298246e-05,
+      "loss": 0.7376,
+      "step": 175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.543859649122807e-05,
+      "loss": 0.7195,
+      "step": 176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5526315789473686e-05,
+      "loss": 0.7203,
+      "step": 177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.56140350877193e-05,
+      "loss": 0.7016,
+      "step": 178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5701754385964915e-05,
+      "loss": 0.7322,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.7267,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5877192982456144e-05,
+      "loss": 0.6788,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5964912280701755e-05,
+      "loss": 0.7123,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.605263157894737e-05,
+      "loss": 0.7241,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6140350877192984e-05,
+      "loss": 0.7498,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.62280701754386e-05,
+      "loss": 0.732,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6315789473684213e-05,
+      "loss": 0.7291,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6403508771929827e-05,
+      "loss": 0.7124,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.649122807017544e-05,
+      "loss": 0.694,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6578947368421053e-05,
+      "loss": 0.7344,
+      "step": 189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.7423,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6754385964912282e-05,
+      "loss": 0.7154,
+      "step": 191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.745,
+      "step": 192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.692982456140351e-05,
+      "loss": 0.7166,
+      "step": 193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7017543859649125e-05,
+      "loss": 0.6981,
+      "step": 194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7105263157894737e-05,
+      "loss": 0.7302,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.719298245614035e-05,
+      "loss": 0.6912,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7280701754385966e-05,
+      "loss": 0.7078,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.736842105263158e-05,
+      "loss": 0.7368,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7456140350877195e-05,
+      "loss": 0.7225,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.754385964912281e-05,
+      "loss": 0.6941,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.763157894736842e-05,
+      "loss": 0.7342,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7719298245614035e-05,
+      "loss": 0.704,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.780701754385965e-05,
+      "loss": 0.6839,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.7153,
+      "step": 204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7982456140350878e-05,
+      "loss": 0.6985,
+      "step": 205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8070175438596493e-05,
+      "loss": 0.7064,
+      "step": 206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8157894736842107e-05,
+      "loss": 0.688,
+      "step": 207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.824561403508772e-05,
+      "loss": 0.7213,
+      "step": 208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.7283,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8421052631578947e-05,
+      "loss": 0.6711,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8508771929824562e-05,
+      "loss": 0.7445,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8596491228070176e-05,
+      "loss": 0.72,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.868421052631579e-05,
+      "loss": 0.7426,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8771929824561405e-05,
+      "loss": 0.7245,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.885964912280702e-05,
+      "loss": 0.6965,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.7175,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9035087719298245e-05,
+      "loss": 0.705,
+      "step": 217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912280701754386e-05,
+      "loss": 0.6995,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9210526315789474e-05,
+      "loss": 0.6713,
+      "step": 219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929824561403509e-05,
+      "loss": 0.7051,
+      "step": 220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9385964912280704e-05,
+      "loss": 0.7042,
+      "step": 221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9473684210526318e-05,
+      "loss": 0.6953,
+      "step": 222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9561403508771933e-05,
+      "loss": 0.6988,
+      "step": 223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9649122807017544e-05,
+      "loss": 0.7088,
+      "step": 224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9736842105263158e-05,
+      "loss": 0.7064,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9824561403508773e-05,
+      "loss": 0.6991,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9912280701754387e-05,
+      "loss": 0.7097,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.7186,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0087719298245613e-05,
+      "loss": 0.7306,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.017543859649123e-05,
+      "loss": 0.6877,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0263157894736842e-05,
+      "loss": 0.7185,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.035087719298246e-05,
+      "loss": 0.7016,
+      "step": 232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.043859649122807e-05,
+      "loss": 0.7335,
+      "step": 233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.052631578947369e-05,
+      "loss": 0.6959,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.06140350877193e-05,
+      "loss": 0.6906,
+      "step": 235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0701754385964918e-05,
+      "loss": 0.6857,
+      "step": 236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.078947368421053e-05,
+      "loss": 0.7135,
+      "step": 237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.087719298245614e-05,
+      "loss": 0.7269,
+      "step": 238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0964912280701758e-05,
+      "loss": 0.7037,
+      "step": 239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 0.7259,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1140350877192987e-05,
+      "loss": 0.7158,
+      "step": 241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1228070175438598e-05,
+      "loss": 0.6713,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1315789473684216e-05,
+      "loss": 0.6898,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1403508771929827e-05,
+      "loss": 0.6856,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1491228070175438e-05,
+      "loss": 0.6829,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1578947368421056e-05,
+      "loss": 0.6824,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 0.6658,
+      "step": 247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1754385964912285e-05,
+      "loss": 0.7062,
+      "step": 248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1842105263157896e-05,
+      "loss": 0.6757,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.192982456140351e-05,
+      "loss": 0.6916,
+      "step": 250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2017543859649125e-05,
+      "loss": 0.7029,
+      "step": 251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.210526315789474e-05,
+      "loss": 0.6503,
+      "step": 252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2192982456140354e-05,
+      "loss": 0.6898,
+      "step": 253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2280701754385965e-05,
+      "loss": 0.6988,
+      "step": 254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2368421052631583e-05,
+      "loss": 0.6839,
+      "step": 255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2456140350877194e-05,
+      "loss": 0.7362,
+      "step": 256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.254385964912281e-05,
+      "loss": 0.7136,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2631578947368423e-05,
+      "loss": 0.7404,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2719298245614038e-05,
+      "loss": 0.6993,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2807017543859652e-05,
+      "loss": 0.6877,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2894736842105263e-05,
+      "loss": 0.6766,
+      "step": 261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2982456140350878e-05,
+      "loss": 0.6358,
+      "step": 262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3070175438596492e-05,
+      "loss": 0.6764,
+      "step": 263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3157894736842107e-05,
+      "loss": 0.6811,
+      "step": 264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.324561403508772e-05,
+      "loss": 0.6896,
+      "step": 265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.651,
+      "step": 266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.342105263157895e-05,
+      "loss": 0.6691,
+      "step": 267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3508771929824565e-05,
+      "loss": 0.7248,
+      "step": 268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3596491228070176e-05,
+      "loss": 0.6859,
+      "step": 269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.368421052631579e-05,
+      "loss": 0.7112,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3771929824561405e-05,
+      "loss": 0.6822,
+      "step": 271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.385964912280702e-05,
+      "loss": 0.6965,
+      "step": 272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3947368421052634e-05,
+      "loss": 0.6855,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.403508771929825e-05,
+      "loss": 0.7133,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4122807017543863e-05,
+      "loss": 0.694,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4210526315789474e-05,
+      "loss": 0.6854,
+      "step": 276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.429824561403509e-05,
+      "loss": 0.719,
+      "step": 277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4385964912280703e-05,
+      "loss": 0.6941,
+      "step": 278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4473684210526318e-05,
+      "loss": 0.6902,
+      "step": 279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4561403508771932e-05,
+      "loss": 0.6846,
+      "step": 280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4649122807017543e-05,
+      "loss": 0.6949,
+      "step": 281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.473684210526316e-05,
+      "loss": 0.702,
+      "step": 282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4824561403508772e-05,
+      "loss": 0.6871,
+      "step": 283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.491228070175439e-05,
+      "loss": 0.6734,
+      "step": 284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7095,
+      "step": 285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5087719298245616e-05,
+      "loss": 0.7079,
+      "step": 286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.517543859649123e-05,
+      "loss": 0.6687,
+      "step": 287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.526315789473684e-05,
+      "loss": 0.6984,
+      "step": 288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.535087719298246e-05,
+      "loss": 0.6877,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.543859649122807e-05,
+      "loss": 0.6858,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5526315789473688e-05,
+      "loss": 0.7044,
+      "step": 291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.56140350877193e-05,
+      "loss": 0.698,
+      "step": 292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5701754385964917e-05,
+      "loss": 0.6926,
+      "step": 293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.578947368421053e-05,
+      "loss": 0.7073,
+      "step": 294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.587719298245614e-05,
+      "loss": 0.6742,
+      "step": 295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5964912280701757e-05,
+      "loss": 0.6781,
+      "step": 296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.605263157894737e-05,
+      "loss": 0.7019,
+      "step": 297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6140350877192986e-05,
+      "loss": 0.675,
+      "step": 298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6228070175438597e-05,
+      "loss": 0.6562,
+      "step": 299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6315789473684215e-05,
+      "loss": 0.6901,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6403508771929826e-05,
+      "loss": 0.6888,
+      "step": 301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6491228070175438e-05,
+      "loss": 0.689,
+      "step": 302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6578947368421055e-05,
+      "loss": 0.7094,
+      "step": 303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.654,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6754385964912284e-05,
+      "loss": 0.7012,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6842105263157896e-05,
+      "loss": 0.698,
+      "step": 306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6929824561403513e-05,
+      "loss": 0.6684,
+      "step": 307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7017543859649125e-05,
+      "loss": 0.7104,
+      "step": 308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7105263157894742e-05,
+      "loss": 0.7153,
+      "step": 309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7192982456140354e-05,
+      "loss": 0.6914,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7280701754385965e-05,
+      "loss": 0.7152,
+      "step": 311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7368421052631583e-05,
+      "loss": 0.6612,
+      "step": 312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7456140350877194e-05,
+      "loss": 0.6941,
+      "step": 313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.754385964912281e-05,
+      "loss": 0.6936,
+      "step": 314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7631578947368423e-05,
+      "loss": 0.6649,
+      "step": 315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.771929824561404e-05,
+      "loss": 0.7155,
+      "step": 316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7807017543859652e-05,
+      "loss": 0.6655,
+      "step": 317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7894736842105263e-05,
+      "loss": 0.6751,
+      "step": 318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.798245614035088e-05,
+      "loss": 0.6386,
+      "step": 319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8070175438596492e-05,
+      "loss": 0.6652,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.815789473684211e-05,
+      "loss": 0.6961,
+      "step": 321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.824561403508772e-05,
+      "loss": 0.7085,
+      "step": 322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.833333333333334e-05,
+      "loss": 0.6577,
+      "step": 323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.842105263157895e-05,
+      "loss": 0.7041,
+      "step": 324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8508771929824564e-05,
+      "loss": 0.6755,
+      "step": 325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.859649122807018e-05,
+      "loss": 0.698,
+      "step": 326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.868421052631579e-05,
+      "loss": 0.7065,
+      "step": 327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8771929824561408e-05,
+      "loss": 0.7163,
+      "step": 328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.885964912280702e-05,
+      "loss": 0.68,
+      "step": 329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8947368421052634e-05,
+      "loss": 0.6825,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9035087719298248e-05,
+      "loss": 0.7068,
+      "step": 331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9122807017543863e-05,
+      "loss": 0.6843,
+      "step": 332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9210526315789477e-05,
+      "loss": 0.6848,
+      "step": 333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9298245614035088e-05,
+      "loss": 0.7073,
+      "step": 334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9385964912280706e-05,
+      "loss": 0.711,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9473684210526317e-05,
+      "loss": 0.6975,
+      "step": 336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.956140350877193e-05,
+      "loss": 0.689,
+      "step": 337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9649122807017546e-05,
+      "loss": 0.6598,
+      "step": 338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.973684210526316e-05,
+      "loss": 0.6672,
+      "step": 339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9824561403508775e-05,
+      "loss": 0.6833,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.991228070175439e-05,
+      "loss": 0.6869,
+      "step": 341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6494,
+      "step": 342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0087719298245615e-05,
+      "loss": 0.6966,
+      "step": 343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.017543859649123e-05,
+      "loss": 0.696,
+      "step": 344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0263157894736844e-05,
+      "loss": 0.6525,
+      "step": 345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.035087719298246e-05,
+      "loss": 0.6798,
+      "step": 346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0438596491228073e-05,
+      "loss": 0.6646,
+      "step": 347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.052631578947369e-05,
+      "loss": 0.6979,
+      "step": 348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.06140350877193e-05,
+      "loss": 0.6714,
+      "step": 349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.070175438596492e-05,
+      "loss": 0.709,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.078947368421053e-05,
+      "loss": 0.7035,
+      "step": 351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.087719298245614e-05,
+      "loss": 0.6806,
+      "step": 352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.096491228070176e-05,
+      "loss": 0.6851,
+      "step": 353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.105263157894737e-05,
+      "loss": 0.6658,
+      "step": 354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.1140350877192986e-05,
+      "loss": 0.6797,
+      "step": 355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.12280701754386e-05,
+      "loss": 0.6722,
+      "step": 356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1315789473684215e-05,
+      "loss": 0.6855,
+      "step": 357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.140350877192983e-05,
+      "loss": 0.6974,
+      "step": 358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.149122807017544e-05,
+      "loss": 0.6858,
+      "step": 359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.157894736842106e-05,
+      "loss": 0.6487,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.669,
+      "step": 361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.175438596491229e-05,
+      "loss": 0.7036,
+      "step": 362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1842105263157895e-05,
+      "loss": 0.6712,
+      "step": 363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.192982456140351e-05,
+      "loss": 0.7092,
+      "step": 364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.2017543859649124e-05,
+      "loss": 0.6652,
+      "step": 365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.210526315789474e-05,
+      "loss": 0.6833,
+      "step": 366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.219298245614035e-05,
+      "loss": 0.6839,
+      "step": 367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.228070175438597e-05,
+      "loss": 0.7206,
+      "step": 368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.236842105263158e-05,
+      "loss": 0.6739,
+      "step": 369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.24561403508772e-05,
+      "loss": 0.6773,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.254385964912281e-05,
+      "loss": 0.6912,
+      "step": 371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.2631578947368426e-05,
+      "loss": 0.6863,
+      "step": 372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.271929824561404e-05,
+      "loss": 0.684,
+      "step": 373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.2807017543859655e-05,
+      "loss": 0.676,
+      "step": 374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.289473684210526e-05,
+      "loss": 0.6695,
+      "step": 375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.298245614035088e-05,
+      "loss": 0.6927,
+      "step": 376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.307017543859649e-05,
+      "loss": 0.6524,
+      "step": 377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3157894736842106e-05,
+      "loss": 0.682,
+      "step": 378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.324561403508772e-05,
+      "loss": 0.714,
+      "step": 379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.7099,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.342105263157895e-05,
+      "loss": 0.6964,
+      "step": 381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3508771929824564e-05,
+      "loss": 0.7141,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.359649122807018e-05,
+      "loss": 0.6585,
+      "step": 383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.368421052631579e-05,
+      "loss": 0.6695,
+      "step": 384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.377192982456141e-05,
+      "loss": 0.6739,
+      "step": 385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.385964912280702e-05,
+      "loss": 0.69,
+      "step": 386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.3947368421052636e-05,
+      "loss": 0.6861,
+      "step": 387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.403508771929825e-05,
+      "loss": 0.7064,
+      "step": 388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4122807017543865e-05,
+      "loss": 0.6836,
+      "step": 389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.421052631578947e-05,
+      "loss": 0.6913,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.429824561403509e-05,
+      "loss": 0.6731,
+      "step": 391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.43859649122807e-05,
+      "loss": 0.6833,
+      "step": 392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.447368421052632e-05,
+      "loss": 0.7126,
+      "step": 393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.456140350877193e-05,
+      "loss": 0.7002,
+      "step": 394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4649122807017546e-05,
+      "loss": 0.6326,
+      "step": 395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.473684210526316e-05,
+      "loss": 0.6821,
+      "step": 396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4824561403508775e-05,
+      "loss": 0.6698,
+      "step": 397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.491228070175439e-05,
+      "loss": 0.6981,
+      "step": 398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 0.6867,
+      "step": 399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.508771929824562e-05,
+      "loss": 0.6742,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.517543859649123e-05,
+      "loss": 0.6696,
+      "step": 401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.526315789473684e-05,
+      "loss": 0.6605,
+      "step": 402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.535087719298246e-05,
+      "loss": 0.7045,
+      "step": 403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.543859649122807e-05,
+      "loss": 0.67,
+      "step": 404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.552631578947369e-05,
+      "loss": 0.6706,
+      "step": 405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.56140350877193e-05,
+      "loss": 0.6951,
+      "step": 406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.570175438596492e-05,
+      "loss": 0.6756,
+      "step": 407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.578947368421053e-05,
+      "loss": 0.6692,
+      "step": 408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.587719298245614e-05,
+      "loss": 0.6679,
+      "step": 409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.5964912280701756e-05,
+      "loss": 0.6935,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.605263157894737e-05,
+      "loss": 0.6865,
+      "step": 411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6140350877192985e-05,
+      "loss": 0.6859,
+      "step": 412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.62280701754386e-05,
+      "loss": 0.6442,
+      "step": 413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6315789473684214e-05,
+      "loss": 0.7061,
+      "step": 414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.640350877192983e-05,
+      "loss": 0.7237,
+      "step": 415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.649122807017544e-05,
+      "loss": 0.6931,
+      "step": 416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.657894736842106e-05,
+      "loss": 0.6875,
+      "step": 417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.7086,
+      "step": 418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.675438596491229e-05,
+      "loss": 0.6562,
+      "step": 419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.7147,
+      "step": 420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6929824561403516e-05,
+      "loss": 0.6889,
+      "step": 421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7017543859649124e-05,
+      "loss": 0.7002,
+      "step": 422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7105263157894745e-05,
+      "loss": 0.6952,
+      "step": 423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.719298245614035e-05,
+      "loss": 0.6838,
+      "step": 424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.728070175438597e-05,
+      "loss": 0.6836,
+      "step": 425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.736842105263158e-05,
+      "loss": 0.6941,
+      "step": 426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7456140350877196e-05,
+      "loss": 0.6691,
+      "step": 427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.754385964912281e-05,
+      "loss": 0.6954,
+      "step": 428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7631578947368425e-05,
+      "loss": 0.6741,
+      "step": 429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.771929824561404e-05,
+      "loss": 0.6443,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7807017543859654e-05,
+      "loss": 0.6703,
+      "step": 431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.789473684210526e-05,
+      "loss": 0.6836,
+      "step": 432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.798245614035088e-05,
+      "loss": 0.645,
+      "step": 433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.807017543859649e-05,
+      "loss": 0.6918,
+      "step": 434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.815789473684211e-05,
+      "loss": 0.7154,
+      "step": 435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.824561403508772e-05,
+      "loss": 0.6725,
+      "step": 436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.833333333333334e-05,
+      "loss": 0.6848,
+      "step": 437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.842105263157895e-05,
+      "loss": 0.6792,
+      "step": 438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8508771929824563e-05,
+      "loss": 0.6681,
+      "step": 439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.859649122807018e-05,
+      "loss": 0.7236,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.868421052631579e-05,
+      "loss": 0.6812,
+      "step": 441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.877192982456141e-05,
+      "loss": 0.6528,
+      "step": 442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.885964912280702e-05,
+      "loss": 0.6752,
+      "step": 443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8947368421052636e-05,
+      "loss": 0.6511,
+      "step": 444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.903508771929825e-05,
+      "loss": 0.6972,
+      "step": 445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9122807017543865e-05,
+      "loss": 0.6629,
+      "step": 446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.921052631578948e-05,
+      "loss": 0.6784,
+      "step": 447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.929824561403509e-05,
+      "loss": 0.6726,
+      "step": 448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.938596491228071e-05,
+      "loss": 0.6599,
+      "step": 449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.6785,
+      "step": 450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.956140350877193e-05,
+      "loss": 0.6933,
+      "step": 451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9649122807017545e-05,
+      "loss": 0.6928,
+      "step": 452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.973684210526316e-05,
+      "loss": 0.6472,
+      "step": 453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9824561403508774e-05,
+      "loss": 0.7011,
+      "step": 454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.991228070175439e-05,
+      "loss": 0.6872,
+      "step": 455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4e-05,
+      "loss": 0.6726,
+      "step": 456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999999411438416e-05,
+      "loss": 0.6785,
+      "step": 457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999976457540104e-05,
+      "loss": 0.6468,
+      "step": 458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999994702947822e-05,
+      "loss": 0.6689,
+      "step": 459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999905830215837e-05,
+      "loss": 0.6645,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999985285977719e-05,
+      "loss": 0.6797,
+      "step": 461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999978811819347e-05,
+      "loss": 0.6769,
+      "step": 462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999971160550277e-05,
+      "loss": 0.6628,
+      "step": 463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999623321750126e-05,
+      "loss": 0.6865,
+      "step": 464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999952326698751e-05,
+      "loss": 0.7012,
+      "step": 465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999411441273786e-05,
+      "loss": 0.6783,
+      "step": 466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999928784467479e-05,
+      "loss": 0.6881,
+      "step": 467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999915247726326e-05,
+      "loss": 0.6786,
+      "step": 468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999005339118864e-05,
+      "loss": 0.6398,
+      "step": 469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999884643032821e-05,
+      "loss": 0.6849,
+      "step": 470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9998675750984824e-05,
+      "loss": 0.6674,
+      "step": 471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9998493301189155e-05,
+      "loss": 0.6811,
+      "step": 472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9998299081048595e-05,
+      "loss": 0.657,
+      "step": 473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999809309067744e-05,
+      "loss": 0.6641,
+      "step": 474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999787533019695e-05,
+      "loss": 0.6667,
+      "step": 475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9997645799735266e-05,
+      "loss": 0.6891,
+      "step": 476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9997404499427494e-05,
+      "loss": 0.6499,
+      "step": 477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999715142941566e-05,
+      "loss": 0.6995,
+      "step": 478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9996886589848704e-05,
+      "loss": 0.6527,
+      "step": 479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999660998088249e-05,
+      "loss": 0.6697,
+      "step": 480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999632160267984e-05,
+      "loss": 0.6786,
+      "step": 481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9996021455410475e-05,
+      "loss": 0.7133,
+      "step": 482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999570953925104e-05,
+      "loss": 0.7169,
+      "step": 483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9995385854385124e-05,
+      "loss": 0.6609,
+      "step": 484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999505040100324e-05,
+      "loss": 0.6764,
+      "step": 485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999470317930282e-05,
+      "loss": 0.682,
+      "step": 486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999434418948822e-05,
+      "loss": 0.6687,
+      "step": 487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999397343177073e-05,
+      "loss": 0.6672,
+      "step": 488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9993590906368576e-05,
+      "loss": 0.7049,
+      "step": 489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9993196613506876e-05,
+      "loss": 0.6776,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999279055341771e-05,
+      "loss": 0.6383,
+      "step": 491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999237272634007e-05,
+      "loss": 0.6704,
+      "step": 492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9991943132519867e-05,
+      "loss": 0.6633,
+      "step": 493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999150177220995e-05,
+      "loss": 0.6655,
+      "step": 494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999104864567007e-05,
+      "loss": 0.6561,
+      "step": 495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999058375316694e-05,
+      "loss": 0.6863,
+      "step": 496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9990107094974167e-05,
+      "loss": 0.6784,
+      "step": 497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9989618671372304e-05,
+      "loss": 0.6643,
+      "step": 498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998911848264881e-05,
+      "loss": 0.672,
+      "step": 499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998860652909807e-05,
+      "loss": 0.6242,
+      "step": 500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998808281102141e-05,
+      "loss": 0.6933,
+      "step": 501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9987547328727075e-05,
+      "loss": 0.6773,
+      "step": 502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9987000082530223e-05,
+      "loss": 0.6734,
+      "step": 503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998644107275293e-05,
+      "loss": 0.685,
+      "step": 504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998587029972423e-05,
+      "loss": 0.6638,
+      "step": 505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998528776378005e-05,
+      "loss": 0.6573,
+      "step": 506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9984693465263236e-05,
+      "loss": 0.6561,
+      "step": 507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998408740452359e-05,
+      "loss": 0.6835,
+      "step": 508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.99834695819178e-05,
+      "loss": 0.6724,
+      "step": 509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.99828399978095e-05,
+      "loss": 0.659,
+      "step": 510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.998219865256923e-05,
+      "loss": 0.6485,
+      "step": 511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.998154554657448e-05,
+      "loss": 0.6777,
+      "step": 512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.998088068020962e-05,
+      "loss": 0.6681,
+      "step": 513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.998020405386598e-05,
+      "loss": 0.6616,
+      "step": 514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9979515667941796e-05,
+      "loss": 0.6619,
+      "step": 515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997881552284222e-05,
+      "loss": 0.6588,
+      "step": 516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9978103618979325e-05,
+      "loss": 0.6366,
+      "step": 517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997737995677213e-05,
+      "loss": 0.691,
+      "step": 518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997664453664654e-05,
+      "loss": 0.6978,
+      "step": 519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997589735903539e-05,
+      "loss": 0.634,
+      "step": 520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997513842437845e-05,
+      "loss": 0.6458,
+      "step": 521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9974367733122396e-05,
+      "loss": 0.6741,
+      "step": 522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9973585285720834e-05,
+      "loss": 0.6883,
+      "step": 523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.997279108263428e-05,
+      "loss": 0.6593,
+      "step": 524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9971985124330154e-05,
+      "loss": 0.6888,
+      "step": 525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9971167411282835e-05,
+      "loss": 0.6905,
+      "step": 526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.997033794397359e-05,
+      "loss": 0.6706,
+      "step": 527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996949672289061e-05,
+      "loss": 0.679,
+      "step": 528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996864374852901e-05,
+      "loss": 0.6688,
+      "step": 529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996777902139081e-05,
+      "loss": 0.6592,
+      "step": 530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996690254198496e-05,
+      "loss": 0.6776,
+      "step": 531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9966014310827313e-05,
+      "loss": 0.6921,
+      "step": 532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996511432844067e-05,
+      "loss": 0.6524,
+      "step": 533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.99642025953547e-05,
+      "loss": 0.6383,
+      "step": 534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996327911210604e-05,
+      "loss": 0.6816,
+      "step": 535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996234387923819e-05,
+      "loss": 0.6444,
+      "step": 536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996139689730161e-05,
+      "loss": 0.6738,
+      "step": 537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9960438166853654e-05,
+      "loss": 0.6792,
+      "step": 538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9959467688458593e-05,
+      "loss": 0.6697,
+      "step": 539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9958485462687606e-05,
+      "loss": 0.6736,
+      "step": 540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.99574914901188e-05,
+      "loss": 0.6995,
+      "step": 541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.99564857713372e-05,
+      "loss": 0.6464,
+      "step": 542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995546830693471e-05,
+      "loss": 0.6512,
+      "step": 543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995443909751019e-05,
+      "loss": 0.6636,
+      "step": 544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995339814366938e-05,
+      "loss": 0.692,
+      "step": 545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995234544602496e-05,
+      "loss": 0.665,
+      "step": 546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9951281005196486e-05,
+      "loss": 0.6644,
+      "step": 547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995020482181047e-05,
+      "loss": 0.6816,
+      "step": 548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.99491168965003e-05,
+      "loss": 0.6638,
+      "step": 549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.994801722990628e-05,
+      "loss": 0.6683,
+      "step": 550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.994690582267566e-05,
+      "loss": 0.6632,
+      "step": 551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.994578267546254e-05,
+      "loss": 0.666,
+      "step": 552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.994464778892796e-05,
+      "loss": 0.6552,
+      "step": 553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.994350116373991e-05,
+      "loss": 0.6492,
+      "step": 554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9942342800573204e-05,
+      "loss": 0.6957,
+      "step": 555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.994117270010964e-05,
+      "loss": 0.6807,
+      "step": 556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993999086303788e-05,
+      "loss": 0.703,
+      "step": 557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9938797290053513e-05,
+      "loss": 0.6384,
+      "step": 558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993759198185903e-05,
+      "loss": 0.6683,
+      "step": 559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993637493916382e-05,
+      "loss": 0.6631,
+      "step": 560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9935146162684206e-05,
+      "loss": 0.6784,
+      "step": 561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993390565314338e-05,
+      "loss": 0.644,
+      "step": 562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993265341127148e-05,
+      "loss": 0.6546,
+      "step": 563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993138943780549e-05,
+      "loss": 0.6467,
+      "step": 564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993011373348938e-05,
+      "loss": 0.6633,
+      "step": 565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.992882629907395e-05,
+      "loss": 0.6679,
+      "step": 566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.992752713531695e-05,
+      "loss": 0.6437,
+      "step": 567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9926216242983017e-05,
+      "loss": 0.6804,
+      "step": 568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9924893622843684e-05,
+      "loss": 0.6748,
+      "step": 569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9923559275677396e-05,
+      "loss": 0.6544,
+      "step": 570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.99222132022695e-05,
+      "loss": 0.6794,
+      "step": 571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.992085540341225e-05,
+      "loss": 0.6906,
+      "step": 572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991948587990479e-05,
+      "loss": 0.6394,
+      "step": 573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991810463255316e-05,
+      "loss": 0.647,
+      "step": 574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991671166217031e-05,
+      "loss": 0.6584,
+      "step": 575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991530696957611e-05,
+      "loss": 0.686,
+      "step": 576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991389055559728e-05,
+      "loss": 0.6719,
+      "step": 577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991246242106749e-05,
+      "loss": 0.6595,
+      "step": 578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9911022566827264e-05,
+      "loss": 0.6542,
+      "step": 579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9909570993724064e-05,
+      "loss": 0.6922,
+      "step": 580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.990810770261222e-05,
+      "loss": 0.6541,
+      "step": 581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.990663269435298e-05,
+      "loss": 0.6422,
+      "step": 582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9905145969814454e-05,
+      "loss": 0.6622,
+      "step": 583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9903647529871696e-05,
+      "loss": 0.641,
+      "step": 584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9902137375406615e-05,
+      "loss": 0.6474,
+      "step": 585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.990061550730803e-05,
+      "loss": 0.6746,
+      "step": 586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.989908192647166e-05,
+      "loss": 0.6067,
+      "step": 587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9897536633800115e-05,
+      "loss": 0.6594,
+      "step": 588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.989597963020289e-05,
+      "loss": 0.6671,
+      "step": 589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.989441091659638e-05,
+      "loss": 0.6831,
+      "step": 590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9892830493903855e-05,
+      "loss": 0.6641,
+      "step": 591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.989123836305552e-05,
+      "loss": 0.6305,
+      "step": 592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.988963452498841e-05,
+      "loss": 0.6678,
+      "step": 593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9888018980646504e-05,
+      "loss": 0.6769,
+      "step": 594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9886391730980645e-05,
+      "loss": 0.6649,
+      "step": 595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9884752776948564e-05,
+      "loss": 0.6335,
+      "step": 596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.988310211951489e-05,
+      "loss": 0.6451,
+      "step": 597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.988143975965113e-05,
+      "loss": 0.6759,
+      "step": 598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9879765698335705e-05,
+      "loss": 0.6416,
+      "step": 599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9878079936553884e-05,
+      "loss": 0.652,
+      "step": 600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.987638247529785e-05,
+      "loss": 0.6504,
+      "step": 601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9874673315566664e-05,
+      "loss": 0.6794,
+      "step": 602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9872952458366267e-05,
+      "loss": 0.6356,
+      "step": 603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9871219904709485e-05,
+      "loss": 0.644,
+      "step": 604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9869475655616044e-05,
+      "loss": 0.6551,
+      "step": 605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986771971211253e-05,
+      "loss": 0.6468,
+      "step": 606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986595207523244e-05,
+      "loss": 0.6767,
+      "step": 607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9864172746016124e-05,
+      "loss": 0.6569,
+      "step": 608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986238172551083e-05,
+      "loss": 0.6456,
+      "step": 609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986057901477069e-05,
+      "loss": 0.6891,
+      "step": 610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.98587646148567e-05,
+      "loss": 0.626,
+      "step": 611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.985693852683675e-05,
+      "loss": 0.6667,
+      "step": 612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.985510075178561e-05,
+      "loss": 0.6487,
+      "step": 613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.985325129078492e-05,
+      "loss": 0.6738,
+      "step": 614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.985139014492319e-05,
+      "loss": 0.6271,
+      "step": 615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.984951731529584e-05,
+      "loss": 0.6559,
+      "step": 616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.984763280300514e-05,
+      "loss": 0.6674,
+      "step": 617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.984573660916024e-05,
+      "loss": 0.6412,
+      "step": 618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9843828734877155e-05,
+      "loss": 0.6616,
+      "step": 619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.98419091812788e-05,
+      "loss": 0.6664,
+      "step": 620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.983997794949495e-05,
+      "loss": 0.6355,
+      "step": 621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.983803504066224e-05,
+      "loss": 0.6672,
+      "step": 622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.983608045592421e-05,
+      "loss": 0.6732,
+      "step": 623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.983411419643125e-05,
+      "loss": 0.6586,
+      "step": 624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9832136263340605e-05,
+      "loss": 0.682,
+      "step": 625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9830146657816435e-05,
+      "loss": 0.6416,
+      "step": 626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.982814538102974e-05,
+      "loss": 0.6265,
+      "step": 627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.982613243415839e-05,
+      "loss": 0.6481,
+      "step": 628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.982410781838712e-05,
+      "loss": 0.644,
+      "step": 629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.982207153490756e-05,
+      "loss": 0.6479,
+      "step": 630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.982002358491817e-05,
+      "loss": 0.6596,
+      "step": 631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9817963969624306e-05,
+      "loss": 0.6735,
+      "step": 632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.981589269023818e-05,
+      "loss": 0.6902,
+      "step": 633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.981380974797886e-05,
+      "loss": 0.6888,
+      "step": 634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9811715144072286e-05,
+      "loss": 0.6648,
+      "step": 635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.980960887975127e-05,
+      "loss": 0.6844,
+      "step": 636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.980749095625546e-05,
+      "loss": 0.6588,
+      "step": 637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.980536137483141e-05,
+      "loss": 0.6456,
+      "step": 638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.980322013673249e-05,
+      "loss": 0.702,
+      "step": 639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9801067243218955e-05,
+      "loss": 0.6349,
+      "step": 640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9798902695557925e-05,
+      "loss": 0.6782,
+      "step": 641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.979672649502336e-05,
+      "loss": 0.6778,
+      "step": 642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9794538642896086e-05,
+      "loss": 0.64,
+      "step": 643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9792339140463794e-05,
+      "loss": 0.6555,
+      "step": 644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9790127989021024e-05,
+      "loss": 0.6487,
+      "step": 645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.978790518986918e-05,
+      "loss": 0.6302,
+      "step": 646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9785670744316506e-05,
+      "loss": 0.6475,
+      "step": 647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.978342465367813e-05,
+      "loss": 0.665,
+      "step": 648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.978116691927599e-05,
+      "loss": 0.6324,
+      "step": 649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9778897542438915e-05,
+      "loss": 0.6585,
+      "step": 650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.977661652450257e-05,
+      "loss": 0.6532,
+      "step": 651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9774323866809485e-05,
+      "loss": 0.659,
+      "step": 652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.977201957070902e-05,
+      "loss": 0.6505,
+      "step": 653
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.976970363755739e-05,
+      "loss": 0.6369,
+      "step": 654
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.976737606871767e-05,
+      "loss": 0.6527,
+      "step": 655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.976503686555978e-05,
+      "loss": 0.6858,
+      "step": 656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.976268602946049e-05,
+      "loss": 0.6503,
+      "step": 657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.97603235618034e-05,
+      "loss": 0.669,
+      "step": 658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9757949463978975e-05,
+      "loss": 0.6671,
+      "step": 659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.975556373738451e-05,
+      "loss": 0.6938,
+      "step": 660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9753166383424166e-05,
+      "loss": 0.6653,
+      "step": 661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9750757403508927e-05,
+      "loss": 0.6663,
+      "step": 662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9748336799056614e-05,
+      "loss": 0.6686,
+      "step": 663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9745904571491916e-05,
+      "loss": 0.6452,
+      "step": 664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9743460722246346e-05,
+      "loss": 0.6483,
+      "step": 665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9741005252758255e-05,
+      "loss": 0.645,
+      "step": 666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9738538164472856e-05,
+      "loss": 0.6689,
+      "step": 667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9736059458842154e-05,
+      "loss": 0.6386,
+      "step": 668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.973356913732504e-05,
+      "loss": 0.6774,
+      "step": 669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.973106720138721e-05,
+      "loss": 0.6337,
+      "step": 670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9728553652501215e-05,
+      "loss": 0.6527,
+      "step": 671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.972602849214643e-05,
+      "loss": 0.6368,
+      "step": 672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9723491721809076e-05,
+      "loss": 0.6351,
+      "step": 673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.972094334298218e-05,
+      "loss": 0.6597,
+      "step": 674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.971838335716564e-05,
+      "loss": 0.6595,
+      "step": 675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.971581176586616e-05,
+      "loss": 0.6641,
+      "step": 676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.971322857059726e-05,
+      "loss": 0.6627,
+      "step": 677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9710633772879335e-05,
+      "loss": 0.6703,
+      "step": 678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.970802737423958e-05,
+      "loss": 0.6588,
+      "step": 679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.970540937621201e-05,
+      "loss": 0.6367,
+      "step": 680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9702779780337476e-05,
+      "loss": 0.6728,
+      "step": 681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.970013858816367e-05,
+      "loss": 0.6636,
+      "step": 682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.96974858012451e-05,
+      "loss": 0.6691,
+      "step": 683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.969482142114307e-05,
+      "loss": 0.636,
+      "step": 684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.969214544942576e-05,
+      "loss": 0.6709,
+      "step": 685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.968945788766812e-05,
+      "loss": 0.6504,
+      "step": 686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9686758737451955e-05,
+      "loss": 0.6628,
+      "step": 687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.968404800036589e-05,
+      "loss": 0.6968,
+      "step": 688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.968132567800536e-05,
+      "loss": 0.661,
+      "step": 689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.96785917719726e-05,
+      "loss": 0.6258,
+      "step": 690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.96758462838767e-05,
+      "loss": 0.6739,
+      "step": 691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.967308921533354e-05,
+      "loss": 0.646,
+      "step": 692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.967032056796583e-05,
+      "loss": 0.6634,
+      "step": 693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.966754034340308e-05,
+      "loss": 0.6384,
+      "step": 694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.966474854328165e-05,
+      "loss": 0.6557,
+      "step": 695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9661945169244654e-05,
+      "loss": 0.6717,
+      "step": 696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.965913022294206e-05,
+      "loss": 0.6624,
+      "step": 697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.965630370603065e-05,
+      "loss": 0.6355,
+      "step": 698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9653465620174e-05,
+      "loss": 0.6886,
+      "step": 699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9650615967042476e-05,
+      "loss": 0.6705,
+      "step": 700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9647754748313294e-05,
+      "loss": 0.6758,
+      "step": 701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.964488196567046e-05,
+      "loss": 0.6619,
+      "step": 702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.964199762080478e-05,
+      "loss": 0.6863,
+      "step": 703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.963910171541386e-05,
+      "loss": 0.6507,
+      "step": 704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.963619425120213e-05,
+      "loss": 0.6837,
+      "step": 705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.96332752298808e-05,
+      "loss": 0.6532,
+      "step": 706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.963034465316791e-05,
+      "loss": 0.6405,
+      "step": 707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.962740252278827e-05,
+      "loss": 0.6613,
+      "step": 708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9624448840473515e-05,
+      "loss": 0.6434,
+      "step": 709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.962148360796206e-05,
+      "loss": 0.6573,
+      "step": 710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.961850682699913e-05,
+      "loss": 0.6539,
+      "step": 711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.961551849933675e-05,
+      "loss": 0.6522,
+      "step": 712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.961251862673372e-05,
+      "loss": 0.652,
+      "step": 713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.960950721095567e-05,
+      "loss": 0.6431,
+      "step": 714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.960648425377499e-05,
+      "loss": 0.6481,
+      "step": 715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9603449756970877e-05,
+      "loss": 0.6529,
+      "step": 716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.960040372232933e-05,
+      "loss": 0.6544,
+      "step": 717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.959734615164311e-05,
+      "loss": 0.6805,
+      "step": 718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.95942770467118e-05,
+      "loss": 0.6552,
+      "step": 719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9591196409341754e-05,
+      "loss": 0.6656,
+      "step": 720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.958810424134612e-05,
+      "loss": 0.6731,
+      "step": 721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.958500054454482e-05,
+      "loss": 0.6552,
+      "step": 722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9581885320764574e-05,
+      "loss": 0.6507,
+      "step": 723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.957875857183889e-05,
+      "loss": 0.6677,
+      "step": 724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9575620299608045e-05,
+      "loss": 0.6386,
+      "step": 725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.957247050591911e-05,
+      "loss": 0.6412,
+      "step": 726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.956930919262592e-05,
+      "loss": 0.6588,
+      "step": 727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.956613636158913e-05,
+      "loss": 0.6361,
+      "step": 728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9562952014676116e-05,
+      "loss": 0.6678,
+      "step": 729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9559756153761084e-05,
+      "loss": 0.6321,
+      "step": 730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9556548780724976e-05,
+      "loss": 0.6618,
+      "step": 731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9553329897455544e-05,
+      "loss": 0.6594,
+      "step": 732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.95500995058473e-05,
+      "loss": 0.6531,
+      "step": 733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.954685760780152e-05,
+      "loss": 0.6868,
+      "step": 734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9543604205226254e-05,
+      "loss": 0.6538,
+      "step": 735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.954033930003634e-05,
+      "loss": 0.6572,
+      "step": 736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9537062894153384e-05,
+      "loss": 0.6724,
+      "step": 737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9533774989505745e-05,
+      "loss": 0.6791,
+      "step": 738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.953047558802854e-05,
+      "loss": 0.6602,
+      "step": 739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.95271646916637e-05,
+      "loss": 0.6628,
+      "step": 740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9523842302359873e-05,
+      "loss": 0.6552,
+      "step": 741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.952050842207249e-05,
+      "loss": 0.6413,
+      "step": 742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9517163052763756e-05,
+      "loss": 0.6788,
+      "step": 743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.951380619640262e-05,
+      "loss": 0.6539,
+      "step": 744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.951043785496478e-05,
+      "loss": 0.6455,
+      "step": 745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.950705803043275e-05,
+      "loss": 0.6572,
+      "step": 746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.950366672479574e-05,
+      "loss": 0.6567,
+      "step": 747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9500263940049746e-05,
+      "loss": 0.6704,
+      "step": 748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.949684967819752e-05,
+      "loss": 0.6253,
+      "step": 749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9493423941248564e-05,
+      "loss": 0.647,
+      "step": 750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9489986731219127e-05,
+      "loss": 0.6413,
+      "step": 751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9486538050132234e-05,
+      "loss": 0.6499,
+      "step": 752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9483077900017635e-05,
+      "loss": 0.6567,
+      "step": 753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.947960628291185e-05,
+      "loss": 0.6777,
+      "step": 754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9476123200858125e-05,
+      "loss": 0.6538,
+      "step": 755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9472628655906484e-05,
+      "loss": 0.652,
+      "step": 756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.946912265011368e-05,
+      "loss": 0.6799,
+      "step": 757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.946560518554321e-05,
+      "loss": 0.6556,
+      "step": 758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9462076264265305e-05,
+      "loss": 0.6486,
+      "step": 759
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.945853588835697e-05,
+      "loss": 0.6701,
+      "step": 760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.945498405990193e-05,
+      "loss": 0.6514,
+      "step": 761
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.945142078099066e-05,
+      "loss": 0.6607,
+      "step": 762
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.944784605372035e-05,
+      "loss": 0.6538,
+      "step": 763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.944425988019498e-05,
+      "loss": 0.6511,
+      "step": 764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9440662262525197e-05,
+      "loss": 0.6452,
+      "step": 765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.943705320282844e-05,
+      "loss": 0.6534,
+      "step": 766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9433432703228866e-05,
+      "loss": 0.6567,
+      "step": 767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.942980076585735e-05,
+      "loss": 0.6772,
+      "step": 768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9426157392851524e-05,
+      "loss": 0.6758,
+      "step": 769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.942250258635573e-05,
+      "loss": 0.646,
+      "step": 770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9418836348521045e-05,
+      "loss": 0.6417,
+      "step": 771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9415158681505274e-05,
+      "loss": 0.7025,
+      "step": 772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9411469587472964e-05,
+      "loss": 0.6295,
+      "step": 773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.940776906859536e-05,
+      "loss": 0.6421,
+      "step": 774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.940405712705045e-05,
+      "loss": 0.6286,
+      "step": 775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.940033376502294e-05,
+      "loss": 0.637,
+      "step": 776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9396598984704264e-05,
+      "loss": 0.6389,
+      "step": 777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9392852788292556e-05,
+      "loss": 0.6616,
+      "step": 778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9389095177992696e-05,
+      "loss": 0.6458,
+      "step": 779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.938532615601626e-05,
+      "loss": 0.6637,
+      "step": 780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.938154572458156e-05,
+      "loss": 0.6731,
+      "step": 781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.937775388591361e-05,
+      "loss": 0.6647,
+      "step": 782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9373950642244134e-05,
+      "loss": 0.68,
+      "step": 783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9370135995811574e-05,
+      "loss": 0.6395,
+      "step": 784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.936630994886109e-05,
+      "loss": 0.6638,
+      "step": 785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.936247250364455e-05,
+      "loss": 0.6594,
+      "step": 786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.935862366242052e-05,
+      "loss": 0.6448,
+      "step": 787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.935476342745428e-05,
+      "loss": 0.6742,
+      "step": 788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.935089180101782e-05,
+      "loss": 0.6417,
+      "step": 789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.934700878538983e-05,
+      "loss": 0.6682,
+      "step": 790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.93431143828557e-05,
+      "loss": 0.6361,
+      "step": 791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.933920859570753e-05,
+      "loss": 0.6417,
+      "step": 792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9335291426244114e-05,
+      "loss": 0.621,
+      "step": 793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.933136287677095e-05,
+      "loss": 0.6307,
+      "step": 794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.932742294960022e-05,
+      "loss": 0.6828,
+      "step": 795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.932347164705084e-05,
+      "loss": 0.6167,
+      "step": 796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.931950897144837e-05,
+      "loss": 0.6549,
+      "step": 797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.931553492512509e-05,
+      "loss": 0.6339,
+      "step": 798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.931154951041998e-05,
+      "loss": 0.6341,
+      "step": 799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9307552729678706e-05,
+      "loss": 0.6621,
+      "step": 800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.930354458525361e-05,
+      "loss": 0.6633,
+      "step": 801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.929952507950373e-05,
+      "loss": 0.6339,
+      "step": 802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.929549421479481e-05,
+      "loss": 0.6638,
+      "step": 803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9291451993499236e-05,
+      "loss": 0.6215,
+      "step": 804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.928739841799613e-05,
+      "loss": 0.6291,
+      "step": 805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.928333349067125e-05,
+      "loss": 0.652,
+      "step": 806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.927925721391707e-05,
+      "loss": 0.668,
+      "step": 807
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.927516959013272e-05,
+      "loss": 0.6673,
+      "step": 808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.927107062172403e-05,
+      "loss": 0.678,
+      "step": 809
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9266960311103485e-05,
+      "loss": 0.6422,
+      "step": 810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9262838660690256e-05,
+      "loss": 0.6241,
+      "step": 811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.92587056729102e-05,
+      "loss": 0.675,
+      "step": 812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.925456135019582e-05,
+      "loss": 0.6251,
+      "step": 813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.925040569498631e-05,
+      "loss": 0.6595,
+      "step": 814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.924623870972753e-05,
+      "loss": 0.6374,
+      "step": 815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9242060396872014e-05,
+      "loss": 0.6437,
+      "step": 816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9237870758878946e-05,
+      "loss": 0.6764,
+      "step": 817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9233669798214195e-05,
+      "loss": 0.6392,
+      "step": 818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.922945751735028e-05,
+      "loss": 0.6545,
+      "step": 819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.922523391876638e-05,
+      "loss": 0.6413,
+      "step": 820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.922099900494836e-05,
+      "loss": 0.6132,
+      "step": 821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.921675277838871e-05,
+      "loss": 0.6294,
+      "step": 822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9212495241586615e-05,
+      "loss": 0.6731,
+      "step": 823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.920822639704787e-05,
+      "loss": 0.607,
+      "step": 824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9203946247284986e-05,
+      "loss": 0.6722,
+      "step": 825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.919965479481708e-05,
+      "loss": 0.6439,
+      "step": 826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9195352042169924e-05,
+      "loss": 0.6327,
+      "step": 827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9191037991875973e-05,
+      "loss": 0.666,
+      "step": 828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.91867126464743e-05,
+      "loss": 0.6673,
+      "step": 829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.918237600851063e-05,
+      "loss": 0.6715,
+      "step": 830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9178028080537356e-05,
+      "loss": 0.6545,
+      "step": 831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9173668865113496e-05,
+      "loss": 0.6148,
+      "step": 832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9169298364804716e-05,
+      "loss": 0.6005,
+      "step": 833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.916491658218333e-05,
+      "loss": 0.6452,
+      "step": 834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9160523519828275e-05,
+      "loss": 0.6398,
+      "step": 835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.915611918032515e-05,
+      "loss": 0.6457,
+      "step": 836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.915170356626617e-05,
+      "loss": 0.6572,
+      "step": 837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.914727668025021e-05,
+      "loss": 0.6542,
+      "step": 838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9142838524882754e-05,
+      "loss": 0.6317,
+      "step": 839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9138389102775936e-05,
+      "loss": 0.6583,
+      "step": 840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.913392841654851e-05,
+      "loss": 0.6477,
+      "step": 841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9129456468825863e-05,
+      "loss": 0.6441,
+      "step": 842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.912497326224001e-05,
+      "loss": 0.6767,
+      "step": 843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9120478799429607e-05,
+      "loss": 0.649,
+      "step": 844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9115973083039914e-05,
+      "loss": 0.6453,
+      "step": 845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.911145611572282e-05,
+      "loss": 0.6496,
+      "step": 846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.910692790013684e-05,
+      "loss": 0.636,
+      "step": 847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9102388438947104e-05,
+      "loss": 0.6129,
+      "step": 848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.909783773482538e-05,
+      "loss": 0.6364,
+      "step": 849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.909327579045002e-05,
+      "loss": 0.6445,
+      "step": 850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.908870260850602e-05,
+      "loss": 0.6553,
+      "step": 851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.908411819168497e-05,
+      "loss": 0.6219,
+      "step": 852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.907952254268509e-05,
+      "loss": 0.6404,
+      "step": 853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.90749156642112e-05,
+      "loss": 0.6716,
+      "step": 854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.907029755897473e-05,
+      "loss": 0.6405,
+      "step": 855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.906566822969372e-05,
+      "loss": 0.6529,
+      "step": 856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9061027679092814e-05,
+      "loss": 0.6787,
+      "step": 857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.905637590990327e-05,
+      "loss": 0.6518,
+      "step": 858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9051712924862926e-05,
+      "loss": 0.6359,
+      "step": 859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9047038726716245e-05,
+      "loss": 0.6616,
+      "step": 860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.904235331821428e-05,
+      "loss": 0.6315,
+      "step": 861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.903765670211469e-05,
+      "loss": 0.6489,
+      "step": 862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.903294888118171e-05,
+      "loss": 0.6226,
+      "step": 863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.902822985818619e-05,
+      "loss": 0.6693,
+      "step": 864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9023499635905554e-05,
+      "loss": 0.6575,
+      "step": 865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.901875821712384e-05,
+      "loss": 0.6153,
+      "step": 866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.901400560463167e-05,
+      "loss": 0.6803,
+      "step": 867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.900924180122624e-05,
+      "loss": 0.6435,
+      "step": 868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9004466809711343e-05,
+      "loss": 0.6469,
+      "step": 869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.899968063289735e-05,
+      "loss": 0.6523,
+      "step": 870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.899488327360124e-05,
+      "loss": 0.6598,
+      "step": 871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.899007473464653e-05,
+      "loss": 0.6653,
+      "step": 872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.898525501886336e-05,
+      "loss": 0.6396,
+      "step": 873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.898042412908842e-05,
+      "loss": 0.6031,
+      "step": 874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.897558206816499e-05,
+      "loss": 0.6652,
+      "step": 875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.897072883894291e-05,
+      "loss": 0.6291,
+      "step": 876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.896586444427863e-05,
+      "loss": 0.6369,
+      "step": 877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.896098888703512e-05,
+      "loss": 0.6414,
+      "step": 878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8956102170081956e-05,
+      "loss": 0.6655,
+      "step": 879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8951204296295274e-05,
+      "loss": 0.643,
+      "step": 880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.894629526855777e-05,
+      "loss": 0.6658,
+      "step": 881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.894137508975871e-05,
+      "loss": 0.6509,
+      "step": 882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.893644376279392e-05,
+      "loss": 0.6415,
+      "step": 883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89315012905658e-05,
+      "loss": 0.6259,
+      "step": 884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8926547675983286e-05,
+      "loss": 0.6384,
+      "step": 885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8921582921961896e-05,
+      "loss": 0.6182,
+      "step": 886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.891660703142369e-05,
+      "loss": 0.6356,
+      "step": 887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.891162000729728e-05,
+      "loss": 0.6949,
+      "step": 888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.890662185251784e-05,
+      "loss": 0.6296,
+      "step": 889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.89016125700271e-05,
+      "loss": 0.6299,
+      "step": 890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.889659216277331e-05,
+      "loss": 0.6441,
+      "step": 891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.889156063371132e-05,
+      "loss": 0.6456,
+      "step": 892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.888651798580247e-05,
+      "loss": 0.6445,
+      "step": 893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.888146422201468e-05,
+      "loss": 0.6734,
+      "step": 894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.88763993453224e-05,
+      "loss": 0.6474,
+      "step": 895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.887132335870662e-05,
+      "loss": 0.6563,
+      "step": 896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8866236265154864e-05,
+      "loss": 0.6542,
+      "step": 897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.886113806766121e-05,
+      "loss": 0.6731,
+      "step": 898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.885602876922626e-05,
+      "loss": 0.6086,
+      "step": 899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.885090837285715e-05,
+      "loss": 0.6505,
+      "step": 900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8845776881567545e-05,
+      "loss": 0.6707,
+      "step": 901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8840634298377644e-05,
+      "loss": 0.665,
+      "step": 902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.883548062631418e-05,
+      "loss": 0.6371,
+      "step": 903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.88303158684104e-05,
+      "loss": 0.6466,
+      "step": 904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.882514002770609e-05,
+      "loss": 0.6532,
+      "step": 905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.881995310724753e-05,
+      "loss": 0.6264,
+      "step": 906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.881475511008757e-05,
+      "loss": 0.6254,
+      "step": 907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.880954603928554e-05,
+      "loss": 0.663,
+      "step": 908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.880432589790729e-05,
+      "loss": 0.6627,
+      "step": 909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.87990946890252e-05,
+      "loss": 0.6504,
+      "step": 910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.6602,
+      "step": 911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.878859908107159e-05,
+      "loss": 0.634,
+      "step": 912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.878333468817737e-05,
+      "loss": 0.6391,
+      "step": 913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.877805924013394e-05,
+      "loss": 0.6469,
+      "step": 914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.877277274004621e-05,
+      "loss": 0.6483,
+      "step": 915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.876747519102562e-05,
+      "loss": 0.6249,
+      "step": 916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8762166596190096e-05,
+      "loss": 0.5991,
+      "step": 917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.875684695866409e-05,
+      "loss": 0.6161,
+      "step": 918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.875151628157853e-05,
+      "loss": 0.6678,
+      "step": 919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.874617456807084e-05,
+      "loss": 0.6396,
+      "step": 920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8740821821284946e-05,
+      "loss": 0.6828,
+      "step": 921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.873545804437127e-05,
+      "loss": 0.643,
+      "step": 922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.873008324048674e-05,
+      "loss": 0.6122,
+      "step": 923
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.872469741279475e-05,
+      "loss": 0.6253,
+      "step": 924
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.871930056446518e-05,
+      "loss": 0.6577,
+      "step": 925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.871389269867442e-05,
+      "loss": 0.6351,
+      "step": 926
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.870847381860534e-05,
+      "loss": 0.6459,
+      "step": 927
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.870304392744726e-05,
+      "loss": 0.6888,
+      "step": 928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.869760302839603e-05,
+      "loss": 0.6401,
+      "step": 929
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.869215112465394e-05,
+      "loss": 0.6229,
+      "step": 930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.868668821942978e-05,
+      "loss": 0.6433,
+      "step": 931
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8681214315938786e-05,
+      "loss": 0.6463,
+      "step": 932
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.867572941740272e-05,
+      "loss": 0.637,
+      "step": 933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.867023352704975e-05,
+      "loss": 0.6219,
+      "step": 934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8664726648114565e-05,
+      "loss": 0.6487,
+      "step": 935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.86592087838383e-05,
+      "loss": 0.6475,
+      "step": 936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8653679937468556e-05,
+      "loss": 0.6432,
+      "step": 937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.86481401122594e-05,
+      "loss": 0.662,
+      "step": 938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.864258931147136e-05,
+      "loss": 0.6747,
+      "step": 939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.863702753837142e-05,
+      "loss": 0.662,
+      "step": 940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.863145479623303e-05,
+      "loss": 0.6267,
+      "step": 941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.86258710883361e-05,
+      "loss": 0.6292,
+      "step": 942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.862027641796697e-05,
+      "loss": 0.6343,
+      "step": 943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.861467078841846e-05,
+      "loss": 0.6107,
+      "step": 944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.860905420298982e-05,
+      "loss": 0.6365,
+      "step": 945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.860342666498677e-05,
+      "loss": 0.6081,
+      "step": 946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.859778817772144e-05,
+      "loss": 0.6234,
+      "step": 947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.859213874451245e-05,
+      "loss": 0.6691,
+      "step": 948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8586478368684826e-05,
+      "loss": 0.6484,
+      "step": 949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.858080705357005e-05,
+      "loss": 0.6327,
+      "step": 950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8575124802506046e-05,
+      "loss": 0.6516,
+      "step": 951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.856943161883717e-05,
+      "loss": 0.6559,
+      "step": 952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.856372750591419e-05,
+      "loss": 0.639,
+      "step": 953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.855801246709435e-05,
+      "loss": 0.6509,
+      "step": 954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8552286505741304e-05,
+      "loss": 0.6457,
+      "step": 955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8546549625225113e-05,
+      "loss": 0.6493,
+      "step": 956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.85408018289223e-05,
+      "loss": 0.6149,
+      "step": 957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.853504312021579e-05,
+      "loss": 0.6311,
+      "step": 958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.852927350249495e-05,
+      "loss": 0.6487,
+      "step": 959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8523492979155534e-05,
+      "loss": 0.6206,
+      "step": 960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.851770155359975e-05,
+      "loss": 0.638,
+      "step": 961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.851189922923621e-05,
+      "loss": 0.6318,
+      "step": 962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8506086009479934e-05,
+      "loss": 0.6418,
+      "step": 963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.850026189775237e-05,
+      "loss": 0.6555,
+      "step": 964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.849442689748135e-05,
+      "loss": 0.6739,
+      "step": 965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.848858101210114e-05,
+      "loss": 0.6421,
+      "step": 966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.84827242450524e-05,
+      "loss": 0.6395,
+      "step": 967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8476856599782204e-05,
+      "loss": 0.6364,
+      "step": 968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.847097807974402e-05,
+      "loss": 0.6389,
+      "step": 969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.846508868839772e-05,
+      "loss": 0.6072,
+      "step": 970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8459188429209565e-05,
+      "loss": 0.5989,
+      "step": 971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.845327730565223e-05,
+      "loss": 0.6336,
+      "step": 972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.844735532120478e-05,
+      "loss": 0.6468,
+      "step": 973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.844142247935265e-05,
+      "loss": 0.6342,
+      "step": 974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.84354787835877e-05,
+      "loss": 0.6239,
+      "step": 975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.842952423740816e-05,
+      "loss": 0.6307,
+      "step": 976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8423558844318635e-05,
+      "loss": 0.6346,
+      "step": 977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8417582607830125e-05,
+      "loss": 0.6302,
+      "step": 978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.841159553146003e-05,
+      "loss": 0.6307,
+      "step": 979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8405597618732105e-05,
+      "loss": 0.6191,
+      "step": 980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.839958887317649e-05,
+      "loss": 0.647,
+      "step": 981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.839356929832969e-05,
+      "loss": 0.6714,
+      "step": 982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.838753889773463e-05,
+      "loss": 0.6322,
+      "step": 983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.838149767494053e-05,
+      "loss": 0.6244,
+      "step": 984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.837544563350305e-05,
+      "loss": 0.6051,
+      "step": 985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.836938277698417e-05,
+      "loss": 0.634,
+      "step": 986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.836330910895228e-05,
+      "loss": 0.6091,
+      "step": 987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.835722463298208e-05,
+      "loss": 0.6311,
+      "step": 988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.835112935265468e-05,
+      "loss": 0.6461,
+      "step": 989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.834502327155751e-05,
+      "loss": 0.6579,
+      "step": 990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8338906393284393e-05,
+      "loss": 0.6508,
+      "step": 991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.833277872143548e-05,
+      "loss": 0.6683,
+      "step": 992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8326640259617275e-05,
+      "loss": 0.657,
+      "step": 993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.832049101144265e-05,
+      "loss": 0.6252,
+      "step": 994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.831433098053082e-05,
+      "loss": 0.6579,
+      "step": 995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.830816017050733e-05,
+      "loss": 0.6357,
+      "step": 996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.830197858500409e-05,
+      "loss": 0.653,
+      "step": 997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.829578622765935e-05,
+      "loss": 0.623,
+      "step": 998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8289583102117676e-05,
+      "loss": 0.6418,
+      "step": 999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.828336921203001e-05,
+      "loss": 0.6045,
+      "step": 1000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.827714456105359e-05,
+      "loss": 0.6489,
+      "step": 1001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.827090915285202e-05,
+      "loss": 0.6325,
+      "step": 1002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8264662991095216e-05,
+      "loss": 0.647,
+      "step": 1003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.825840607945943e-05,
+      "loss": 0.6049,
+      "step": 1004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.825213842162724e-05,
+      "loss": 0.6233,
+      "step": 1005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.824586002128755e-05,
+      "loss": 0.6337,
+      "step": 1006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8239570882135575e-05,
+      "loss": 0.6322,
+      "step": 1007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.823327100787288e-05,
+      "loss": 0.6412,
+      "step": 1008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8226960402207316e-05,
+      "loss": 0.649,
+      "step": 1009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8220639068853056e-05,
+      "loss": 0.6378,
+      "step": 1010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8214307011530616e-05,
+      "loss": 0.6415,
+      "step": 1011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.820796423396678e-05,
+      "loss": 0.5894,
+      "step": 1012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8201610739894675e-05,
+      "loss": 0.6517,
+      "step": 1013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.819524653305372e-05,
+      "loss": 0.6322,
+      "step": 1014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8188871617189646e-05,
+      "loss": 0.6433,
+      "step": 1015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.818248599605448e-05,
+      "loss": 0.6078,
+      "step": 1016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8176089673406547e-05,
+      "loss": 0.6582,
+      "step": 1017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.816968265301049e-05,
+      "loss": 0.615,
+      "step": 1018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8163264938637225e-05,
+      "loss": 0.6326,
+      "step": 1019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.815683653406398e-05,
+      "loss": 0.6508,
+      "step": 1020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8150397443074266e-05,
+      "loss": 0.6477,
+      "step": 1021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.814394766945788e-05,
+      "loss": 0.6249,
+      "step": 1022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.813748721701091e-05,
+      "loss": 0.6538,
+      "step": 1023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8131016089535735e-05,
+      "loss": 0.6364,
+      "step": 1024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.812453429084101e-05,
+      "loss": 0.6172,
+      "step": 1025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.811804182474167e-05,
+      "loss": 0.6225,
+      "step": 1026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.811153869505893e-05,
+      "loss": 0.6667,
+      "step": 1027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.81050249056203e-05,
+      "loss": 0.6452,
+      "step": 1028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8098500460259524e-05,
+      "loss": 0.6528,
+      "step": 1029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.809196536281665e-05,
+      "loss": 0.6578,
+      "step": 1030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.808541961713798e-05,
+      "loss": 0.6002,
+      "step": 1031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8078863227076094e-05,
+      "loss": 0.6513,
+      "step": 1032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8072296196489825e-05,
+      "loss": 0.6262,
+      "step": 1033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8065718529244285e-05,
+      "loss": 0.6414,
+      "step": 1034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.805913022921083e-05,
+      "loss": 0.6521,
+      "step": 1035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8052531300267075e-05,
+      "loss": 0.6362,
+      "step": 1036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.80459217462969e-05,
+      "loss": 0.652,
+      "step": 1037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8039301571190444e-05,
+      "loss": 0.6519,
+      "step": 1038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8032670778844075e-05,
+      "loss": 0.6423,
+      "step": 1039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.802602937316043e-05,
+      "loss": 0.6554,
+      "step": 1040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.801937735804838e-05,
+      "loss": 0.5945,
+      "step": 1041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.801271473742306e-05,
+      "loss": 0.6217,
+      "step": 1042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.800604151520581e-05,
+      "loss": 0.6741,
+      "step": 1043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.799935769532425e-05,
+      "loss": 0.6558,
+      "step": 1044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.7992663281712206e-05,
+      "loss": 0.6219,
+      "step": 1045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.798595827830976e-05,
+      "loss": 0.6016,
+      "step": 1046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.7979242689063226e-05,
+      "loss": 0.6551,
+      "step": 1047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.7972516517925126e-05,
+      "loss": 0.6576,
+      "step": 1048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.796577976885424e-05,
+      "loss": 0.654,
+      "step": 1049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.795903244581555e-05,
+      "loss": 0.6096,
+      "step": 1050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.795227455278029e-05,
+      "loss": 0.6294,
+      "step": 1051
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.7945506093725863e-05,
+      "loss": 0.6395,
+      "step": 1052
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.793872707263595e-05,
+      "loss": 0.6404,
+      "step": 1053
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.793193749350042e-05,
+      "loss": 0.647,
+      "step": 1054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.792513736031535e-05,
+      "loss": 0.6593,
+      "step": 1055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.791832667708304e-05,
+      "loss": 0.6365,
+      "step": 1056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.7911505447812e-05,
+      "loss": 0.6324,
+      "step": 1057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.790467367651694e-05,
+      "loss": 0.6402,
+      "step": 1058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.7897831367218774e-05,
+      "loss": 0.628,
+      "step": 1059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.789097852394463e-05,
+      "loss": 0.6863,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.788411515072783e-05,
+      "loss": 0.6481,
+      "step": 1061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.787724125160789e-05,
+      "loss": 0.6408,
+      "step": 1062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.7870356830630516e-05,
+      "loss": 0.645,
+      "step": 1063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.786346189184762e-05,
+      "loss": 0.6467,
+      "step": 1064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.785655643931728e-05,
+      "loss": 0.6524,
+      "step": 1065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.784964047710381e-05,
+      "loss": 0.631,
+      "step": 1066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.7842714009277675e-05,
+      "loss": 0.6558,
+      "step": 1067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.783577703991551e-05,
+      "loss": 0.6309,
+      "step": 1068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.782882957310015e-05,
+      "loss": 0.6398,
+      "step": 1069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.782187161292063e-05,
+      "loss": 0.6433,
+      "step": 1070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7814903163472105e-05,
+      "loss": 0.6529,
+      "step": 1071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.780792422885597e-05,
+      "loss": 0.6716,
+      "step": 1072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7800934813179736e-05,
+      "loss": 0.6224,
+      "step": 1073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.779393492055711e-05,
+      "loss": 0.6351,
+      "step": 1074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.778692455510796e-05,
+      "loss": 0.6637,
+      "step": 1075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.777990372095833e-05,
+      "loss": 0.6453,
+      "step": 1076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.777287242224039e-05,
+      "loss": 0.6452,
+      "step": 1077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7765830663092505e-05,
+      "loss": 0.6414,
+      "step": 1078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7758778447659184e-05,
+      "loss": 0.6357,
+      "step": 1079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.775171578009109e-05,
+      "loss": 0.6452,
+      "step": 1080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.774464266454503e-05,
+      "loss": 0.6366,
+      "step": 1081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7737559105183986e-05,
+      "loss": 0.6191,
+      "step": 1082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.773046510617705e-05,
+      "loss": 0.6187,
+      "step": 1083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.772336067169948e-05,
+      "loss": 0.6354,
+      "step": 1084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.771624580593269e-05,
+      "loss": 0.647,
+      "step": 1085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7709120513064196e-05,
+      "loss": 0.617,
+      "step": 1086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.770198479728769e-05,
+      "loss": 0.624,
+      "step": 1087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769483866280297e-05,
+      "loss": 0.655,
+      "step": 1088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.768768211381597e-05,
+      "loss": 0.6181,
+      "step": 1089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.768051515453877e-05,
+      "loss": 0.6285,
+      "step": 1090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.767333778918956e-05,
+      "loss": 0.6584,
+      "step": 1091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7666150021992674e-05,
+      "loss": 0.6143,
+      "step": 1092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7658951857178544e-05,
+      "loss": 0.6367,
+      "step": 1093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7651743298983726e-05,
+      "loss": 0.6286,
+      "step": 1094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7644524351650926e-05,
+      "loss": 0.6484,
+      "step": 1095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7637295019428916e-05,
+      "loss": 0.636,
+      "step": 1096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.763005530657261e-05,
+      "loss": 0.6133,
+      "step": 1097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7622805217343025e-05,
+      "loss": 0.641,
+      "step": 1098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.761554475600729e-05,
+      "loss": 0.6252,
+      "step": 1099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.760827392683863e-05,
+      "loss": 0.6766,
+      "step": 1100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7600992734116374e-05,
+      "loss": 0.6265,
+      "step": 1101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.759370118212595e-05,
+      "loss": 0.6168,
+      "step": 1102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.75863992751589e-05,
+      "loss": 0.6565,
+      "step": 1103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.757908701751282e-05,
+      "loss": 0.6106,
+      "step": 1104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7571764413491446e-05,
+      "loss": 0.6257,
+      "step": 1105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.756443146740457e-05,
+      "loss": 0.6506,
+      "step": 1106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.755708818356809e-05,
+      "loss": 0.6432,
+      "step": 1107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7549734566303986e-05,
+      "loss": 0.6234,
+      "step": 1108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7542370619940297e-05,
+      "loss": 0.6228,
+      "step": 1109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7534996348811164e-05,
+      "loss": 0.6639,
+      "step": 1110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.752761175725681e-05,
+      "loss": 0.641,
+      "step": 1111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.752021684962351e-05,
+      "loss": 0.6156,
+      "step": 1112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.751281163026363e-05,
+      "loss": 0.6392,
+      "step": 1113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.75053961035356e-05,
+      "loss": 0.6654,
+      "step": 1114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7497970273803905e-05,
+      "loss": 0.639,
+      "step": 1115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.749053414543911e-05,
+      "loss": 0.6585,
+      "step": 1116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7483087722817834e-05,
+      "loss": 0.6427,
+      "step": 1117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7475631010322745e-05,
+      "loss": 0.621,
+      "step": 1118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.74681640123426e-05,
+      "loss": 0.6464,
+      "step": 1119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.746068673327217e-05,
+      "loss": 0.631,
+      "step": 1120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.745319917751229e-05,
+      "loss": 0.6445,
+      "step": 1121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.744570134946986e-05,
+      "loss": 0.6193,
+      "step": 1122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7438193253557806e-05,
+      "loss": 0.6523,
+      "step": 1123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7430674894195114e-05,
+      "loss": 0.6706,
+      "step": 1124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7423146275806786e-05,
+      "loss": 0.6407,
+      "step": 1125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.741560740282389e-05,
+      "loss": 0.6527,
+      "step": 1126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.740805827968352e-05,
+      "loss": 0.6207,
+      "step": 1127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.740049891082879e-05,
+      "loss": 0.6107,
+      "step": 1128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.739292930070885e-05,
+      "loss": 0.6328,
+      "step": 1129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.73853494537789e-05,
+      "loss": 0.6133,
+      "step": 1130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7377759374500135e-05,
+      "loss": 0.6184,
+      "step": 1131
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.737015906733978e-05,
+      "loss": 0.6363,
+      "step": 1132
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.736254853677109e-05,
+      "loss": 0.6394,
+      "step": 1133
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7354927787273334e-05,
+      "loss": 0.6233,
+      "step": 1134
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.734729682333179e-05,
+      "loss": 0.6549,
+      "step": 1135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.733965564943774e-05,
+      "loss": 0.6394,
+      "step": 1136
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.73320042700885e-05,
+      "loss": 0.6141,
+      "step": 1137
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.732434268978737e-05,
+      "loss": 0.6415,
+      "step": 1138
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.731667091304367e-05,
+      "loss": 0.6593,
+      "step": 1139
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7308988944372705e-05,
+      "loss": 0.6215,
+      "step": 1140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.730129678829578e-05,
+      "loss": 0.6116,
+      "step": 1141
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.729359444934022e-05,
+      "loss": 0.6218,
+      "step": 1142
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7285881932039307e-05,
+      "loss": 0.6658,
+      "step": 1143
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7278159240932355e-05,
+      "loss": 0.6031,
+      "step": 1144
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7270426380564625e-05,
+      "loss": 0.6432,
+      "step": 1145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.726268335548738e-05,
+      "loss": 0.6329,
+      "step": 1146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.725493017025788e-05,
+      "loss": 0.6373,
+      "step": 1147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7247166829439346e-05,
+      "loss": 0.6426,
+      "step": 1148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.723939333760099e-05,
+      "loss": 0.6556,
+      "step": 1149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.723160969931797e-05,
+      "loss": 0.6546,
+      "step": 1150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.722381591917146e-05,
+      "loss": 0.6119,
+      "step": 1151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7216012001748564e-05,
+      "loss": 0.616,
+      "step": 1152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7208197951642376e-05,
+      "loss": 0.662,
+      "step": 1153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7200373773451944e-05,
+      "loss": 0.6264,
+      "step": 1154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7192539471782275e-05,
+      "loss": 0.6486,
+      "step": 1155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.718469505124434e-05,
+      "loss": 0.6109,
+      "step": 1156
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.717684051645506e-05,
+      "loss": 0.6594,
+      "step": 1157
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.716897587203733e-05,
+      "loss": 0.6475,
+      "step": 1158
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7161101122619954e-05,
+      "loss": 0.6312,
+      "step": 1159
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.715321627283772e-05,
+      "loss": 0.6489,
+      "step": 1160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7145321327331345e-05,
+      "loss": 0.6368,
+      "step": 1161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7137416290747486e-05,
+      "loss": 0.6506,
+      "step": 1162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.712950116773875e-05,
+      "loss": 0.6224,
+      "step": 1163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7121575962963676e-05,
+      "loss": 0.6098,
+      "step": 1164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711364068108673e-05,
+      "loss": 0.6533,
+      "step": 1165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.710569532677832e-05,
+      "loss": 0.6207,
+      "step": 1166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.709773990471476e-05,
+      "loss": 0.6428,
+      "step": 1167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.708977441957833e-05,
+      "loss": 0.6312,
+      "step": 1168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.708179887605719e-05,
+      "loss": 0.6373,
+      "step": 1169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.707381327884545e-05,
+      "loss": 0.6275,
+      "step": 1170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7065817632643115e-05,
+      "loss": 0.6188,
+      "step": 1171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.705781194215613e-05,
+      "loss": 0.6708,
+      "step": 1172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.704979621209632e-05,
+      "loss": 0.6137,
+      "step": 1173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.704177044718145e-05,
+      "loss": 0.6383,
+      "step": 1174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.703373465213516e-05,
+      "loss": 0.637,
+      "step": 1175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702568883168703e-05,
+      "loss": 0.6251,
+      "step": 1176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70176329905725e-05,
+      "loss": 0.6244,
+      "step": 1177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700956713353296e-05,
+      "loss": 0.633,
+      "step": 1178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700149126531563e-05,
+      "loss": 0.6436,
+      "step": 1179
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699340539067368e-05,
+      "loss": 0.6473,
+      "step": 1180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6985309514366115e-05,
+      "loss": 0.6096,
+      "step": 1181
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697720364115789e-05,
+      "loss": 0.6435,
+      "step": 1182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.69690877758198e-05,
+      "loss": 0.61,
+      "step": 1183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696096192312852e-05,
+      "loss": 0.6246,
+      "step": 1184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695282608786663e-05,
+      "loss": 0.5988,
+      "step": 1185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694468027482256e-05,
+      "loss": 0.6531,
+      "step": 1186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6936524488790635e-05,
+      "loss": 0.6346,
+      "step": 1187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6928358734571016e-05,
+      "loss": 0.6259,
+      "step": 1188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.692018301696977e-05,
+      "loss": 0.6156,
+      "step": 1189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.691199734079881e-05,
+      "loss": 0.6502,
+      "step": 1190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.69038017108759e-05,
+      "loss": 0.6495,
+      "step": 1191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6895596132024676e-05,
+      "loss": 0.6211,
+      "step": 1192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6887380609074633e-05,
+      "loss": 0.6444,
+      "step": 1193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6879155146861115e-05,
+      "loss": 0.6209,
+      "step": 1194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687091975022529e-05,
+      "loss": 0.627,
+      "step": 1195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686267442401422e-05,
+      "loss": 0.6318,
+      "step": 1196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6854419173080784e-05,
+      "loss": 0.6036,
+      "step": 1197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6846154002283696e-05,
+      "loss": 0.6434,
+      "step": 1198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683787891648753e-05,
+      "loss": 0.6229,
+      "step": 1199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682959392056267e-05,
+      "loss": 0.6724,
+      "step": 1200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682129901938536e-05,
+      "loss": 0.622,
+      "step": 1201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681299421783765e-05,
+      "loss": 0.6383,
+      "step": 1202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680467952080743e-05,
+      "loss": 0.6425,
+      "step": 1203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.67963549331884e-05,
+      "loss": 0.623,
+      "step": 1204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678802045988012e-05,
+      "loss": 0.6228,
+      "step": 1205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.677967610578792e-05,
+      "loss": 0.6109,
+      "step": 1206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6771321875822976e-05,
+      "loss": 0.6418,
+      "step": 1207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676295777490225e-05,
+      "loss": 0.6514,
+      "step": 1208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675458380794855e-05,
+      "loss": 0.6166,
+      "step": 1209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674619997989047e-05,
+      "loss": 0.6062,
+      "step": 1210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6737806295662396e-05,
+      "loss": 0.6174,
+      "step": 1211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6729402760204535e-05,
+      "loss": 0.6226,
+      "step": 1212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672098937846289e-05,
+      "loss": 0.6307,
+      "step": 1213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671256615538925e-05,
+      "loss": 0.6353,
+      "step": 1214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670413309594119e-05,
+      "loss": 0.6296,
+      "step": 1215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66956902050821e-05,
+      "loss": 0.6302,
+      "step": 1216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668723748778113e-05,
+      "loss": 0.6391,
+      "step": 1217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667877494901325e-05,
+      "loss": 0.6052,
+      "step": 1218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667030259375915e-05,
+      "loss": 0.6158,
+      "step": 1219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666182042700535e-05,
+      "loss": 0.622,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6653328453744125e-05,
+      "loss": 0.6546,
+      "step": 1221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6644826678973536e-05,
+      "loss": 0.6341,
+      "step": 1222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.663631510769739e-05,
+      "loss": 0.6132,
+      "step": 1223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6627793744925256e-05,
+      "loss": 0.6425,
+      "step": 1224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.661926259567251e-05,
+      "loss": 0.6031,
+      "step": 1225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6610721664960236e-05,
+      "loss": 0.6087,
+      "step": 1226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660217095781531e-05,
+      "loss": 0.6098,
+      "step": 1227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6593610479270346e-05,
+      "loss": 0.6122,
+      "step": 1228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658504023436372e-05,
+      "loss": 0.6183,
+      "step": 1229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657646022813954e-05,
+      "loss": 0.6399,
+      "step": 1230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656787046564766e-05,
+      "loss": 0.6311,
+      "step": 1231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65592709519437e-05,
+      "loss": 0.6321,
+      "step": 1232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6550661692089e-05,
+      "loss": 0.6423,
+      "step": 1233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654204269115063e-05,
+      "loss": 0.6567,
+      "step": 1234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6533413954201426e-05,
+      "loss": 0.6176,
+      "step": 1235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65247754863199e-05,
+      "loss": 0.653,
+      "step": 1236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.651612729259034e-05,
+      "loss": 0.6255,
+      "step": 1237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6507469378102745e-05,
+      "loss": 0.6277,
+      "step": 1238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.649880174795282e-05,
+      "loss": 0.639,
+      "step": 1239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6490124407242007e-05,
+      "loss": 0.6448,
+      "step": 1240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648143736107745e-05,
+      "loss": 0.6255,
+      "step": 1241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.647274061457201e-05,
+      "loss": 0.6314,
+      "step": 1242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.646403417284426e-05,
+      "loss": 0.6196,
+      "step": 1243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.645531804101848e-05,
+      "loss": 0.6458,
+      "step": 1244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.644659222422464e-05,
+      "loss": 0.6244,
+      "step": 1245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643785672759844e-05,
+      "loss": 0.6326,
+      "step": 1246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642911155628124e-05,
+      "loss": 0.6234,
+      "step": 1247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6420356715420115e-05,
+      "loss": 0.6439,
+      "step": 1248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6411592210167834e-05,
+      "loss": 0.6085,
+      "step": 1249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6402818045682844e-05,
+      "loss": 0.6439,
+      "step": 1250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639403422712928e-05,
+      "loss": 0.6248,
+      "step": 1251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.638524075967697e-05,
+      "loss": 0.6544,
+      "step": 1252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6376437648501395e-05,
+      "loss": 0.648,
+      "step": 1253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.636762489878374e-05,
+      "loss": 0.612,
+      "step": 1254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6358802515710854e-05,
+      "loss": 0.6185,
+      "step": 1255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.634997050447523e-05,
+      "loss": 0.6018,
+      "step": 1256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6341128870275084e-05,
+      "loss": 0.6243,
+      "step": 1257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.633227761831424e-05,
+      "loss": 0.6406,
+      "step": 1258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.63234167538022e-05,
+      "loss": 0.6213,
+      "step": 1259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6314546281954146e-05,
+      "loss": 0.6467,
+      "step": 1260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6305666207990886e-05,
+      "loss": 0.6174,
+      "step": 1261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6296776537138905e-05,
+      "loss": 0.65,
+      "step": 1262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62878772746303e-05,
+      "loss": 0.6084,
+      "step": 1263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627896842570285e-05,
+      "loss": 0.6298,
+      "step": 1264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627004999559995e-05,
+      "loss": 0.6054,
+      "step": 1265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626112198957065e-05,
+      "loss": 0.6359,
+      "step": 1266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6252184412869637e-05,
+      "loss": 0.6205,
+      "step": 1267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.624323727075723e-05,
+      "loss": 0.6256,
+      "step": 1268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.623428056849936e-05,
+      "loss": 0.6218,
+      "step": 1269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6225314311367596e-05,
+      "loss": 0.609,
+      "step": 1270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.621633850463915e-05,
+      "loss": 0.6508,
+      "step": 1271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6207353153596814e-05,
+      "loss": 0.6218,
+      "step": 1272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.619835826352904e-05,
+      "loss": 0.6001,
+      "step": 1273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6189353839729864e-05,
+      "loss": 0.6285,
+      "step": 1274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6180339887498953e-05,
+      "loss": 0.6599,
+      "step": 1275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.617131641214156e-05,
+      "loss": 0.6399,
+      "step": 1276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.616228341896857e-05,
+      "loss": 0.6239,
+      "step": 1277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.615324091329645e-05,
+      "loss": 0.6344,
+      "step": 1278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.614418890044726e-05,
+      "loss": 0.6298,
+      "step": 1279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6135127385748683e-05,
+      "loss": 0.631,
+      "step": 1280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6126056374533974e-05,
+      "loss": 0.6536,
+      "step": 1281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6116975872141984e-05,
+      "loss": 0.6412,
+      "step": 1282
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.610788588391714e-05,
+      "loss": 0.6421,
+      "step": 1283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.609878641520947e-05,
+      "loss": 0.6169,
+      "step": 1284
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.608967747137456e-05,
+      "loss": 0.6265,
+      "step": 1285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.608055905777359e-05,
+      "loss": 0.6424,
+      "step": 1286
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.607143117977331e-05,
+      "loss": 0.6506,
+      "step": 1287
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.606229384274604e-05,
+      "loss": 0.6273,
+      "step": 1288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.605314705206966e-05,
+      "loss": 0.6126,
+      "step": 1289
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.604399081312762e-05,
+      "loss": 0.6143,
+      "step": 1290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6034825131308936e-05,
+      "loss": 0.6295,
+      "step": 1291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.602565001200817e-05,
+      "loss": 0.6201,
+      "step": 1292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.601646546062545e-05,
+      "loss": 0.6053,
+      "step": 1293
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6007271482566444e-05,
+      "loss": 0.5934,
+      "step": 1294
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.599806808324238e-05,
+      "loss": 0.6232,
+      "step": 1295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.598885526807003e-05,
+      "loss": 0.6336,
+      "step": 1296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.597963304247169e-05,
+      "loss": 0.621,
+      "step": 1297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.597040141187521e-05,
+      "loss": 0.6657,
+      "step": 1298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.596116038171398e-05,
+      "loss": 0.6447,
+      "step": 1299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5951909957426916e-05,
+      "loss": 0.6147,
+      "step": 1300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5942650144458454e-05,
+      "loss": 0.6651,
+      "step": 1301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.593338094825857e-05,
+      "loss": 0.6317,
+      "step": 1302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5924102374282754e-05,
+      "loss": 0.6098,
+      "step": 1303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.591481442799202e-05,
+      "loss": 0.6367,
+      "step": 1304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.59055171148529e-05,
+      "loss": 0.6224,
+      "step": 1305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5896210440337434e-05,
+      "loss": 0.6275,
+      "step": 1306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.588689440992317e-05,
+      "loss": 0.6138,
+      "step": 1307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.587756902909316e-05,
+      "loss": 0.6631,
+      "step": 1308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.586823430333597e-05,
+      "loss": 0.6452,
+      "step": 1309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5858890238145674e-05,
+      "loss": 0.6397,
+      "step": 1310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.584953683902181e-05,
+      "loss": 0.6324,
+      "step": 1311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.584017411146944e-05,
+      "loss": 0.6506,
+      "step": 1312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5830802060999106e-05,
+      "loss": 0.6758,
+      "step": 1313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5821420693126834e-05,
+      "loss": 0.6641,
+      "step": 1314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5812030013374126e-05,
+      "loss": 0.6173,
+      "step": 1315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5802630027267993e-05,
+      "loss": 0.6138,
+      "step": 1316
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5793220740340904e-05,
+      "loss": 0.6252,
+      "step": 1317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.578380215813079e-05,
+      "loss": 0.582,
+      "step": 1318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.577437428618108e-05,
+      "loss": 0.6346,
+      "step": 1319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5764937130040654e-05,
+      "loss": 0.5875,
+      "step": 1320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.575549069526386e-05,
+      "loss": 0.6389,
+      "step": 1321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5746034987410504e-05,
+      "loss": 0.5843,
+      "step": 1322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.573657001204584e-05,
+      "loss": 0.6282,
+      "step": 1323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.572709577474062e-05,
+      "loss": 0.6219,
+      "step": 1324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5717612281070995e-05,
+      "loss": 0.6114,
+      "step": 1325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5708119536618584e-05,
+      "loss": 0.6404,
+      "step": 1326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.569861754697045e-05,
+      "loss": 0.6249,
+      "step": 1327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.568910631771911e-05,
+      "loss": 0.6557,
+      "step": 1328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5679585854462506e-05,
+      "loss": 0.5998,
+      "step": 1329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5670056162804024e-05,
+      "loss": 0.6262,
+      "step": 1330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.566051724835245e-05,
+      "loss": 0.6036,
+      "step": 1331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.565096911672204e-05,
+      "loss": 0.6163,
+      "step": 1332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5641411773532453e-05,
+      "loss": 0.6249,
+      "step": 1333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.563184522440878e-05,
+      "loss": 0.6329,
+      "step": 1334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.562226947498152e-05,
+      "loss": 0.6348,
+      "step": 1335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.561268453088659e-05,
+      "loss": 0.6172,
+      "step": 1336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.560309039776532e-05,
+      "loss": 0.6528,
+      "step": 1337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.559348708126445e-05,
+      "loss": 0.6058,
+      "step": 1338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.558387458703612e-05,
+      "loss": 0.6381,
+      "step": 1339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.557425292073788e-05,
+      "loss": 0.6256,
+      "step": 1340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.556462208803268e-05,
+      "loss": 0.6347,
+      "step": 1341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.555498209458883e-05,
+      "loss": 0.6185,
+      "step": 1342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5545332946080095e-05,
+      "loss": 0.632,
+      "step": 1343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.553567464818557e-05,
+      "loss": 0.636,
+      "step": 1344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.552600720658976e-05,
+      "loss": 0.6453,
+      "step": 1345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.551633062698256e-05,
+      "loss": 0.6403,
+      "step": 1346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.550664491505923e-05,
+      "loss": 0.6333,
+      "step": 1347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.54969500765204e-05,
+      "loss": 0.6078,
+      "step": 1348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.548724611707209e-05,
+      "loss": 0.6216,
+      "step": 1349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.547753304242567e-05,
+      "loss": 0.618,
+      "step": 1350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.546781085829788e-05,
+      "loss": 0.6034,
+      "step": 1351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.545807957041084e-05,
+      "loss": 0.6469,
+      "step": 1352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.544833918449199e-05,
+      "loss": 0.6516,
+      "step": 1353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.543858970627417e-05,
+      "loss": 0.6449,
+      "step": 1354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.542883114149553e-05,
+      "loss": 0.628,
+      "step": 1355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.541906349589959e-05,
+      "loss": 0.6489,
+      "step": 1356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5409286775235224e-05,
+      "loss": 0.6202,
+      "step": 1357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.539950098525662e-05,
+      "loss": 0.6305,
+      "step": 1358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.538970613172332e-05,
+      "loss": 0.6275,
+      "step": 1359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5379902220400204e-05,
+      "loss": 0.6203,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5370089257057476e-05,
+      "loss": 0.6066,
+      "step": 1361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.536026724747067e-05,
+      "loss": 0.6592,
+      "step": 1362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.535043619742063e-05,
+      "loss": 0.6212,
+      "step": 1363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5340596112693553e-05,
+      "loss": 0.6529,
+      "step": 1364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.533074699908093e-05,
+      "loss": 0.62,
+      "step": 1365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.6133,
+      "step": 1366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.531102170839158e-05,
+      "loss": 0.6383,
+      "step": 1367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.530114554292441e-05,
+      "loss": 0.617,
+      "step": 1368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5291260371790776e-05,
+      "loss": 0.6319,
+      "step": 1369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.528136620080872e-05,
+      "loss": 0.6361,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5271463035801563e-05,
+      "loss": 0.6425,
+      "step": 1371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5261550882597934e-05,
+      "loss": 0.654,
+      "step": 1372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.525162974703174e-05,
+      "loss": 0.6484,
+      "step": 1373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.524169963494219e-05,
+      "loss": 0.6339,
+      "step": 1374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.523176055217375e-05,
+      "loss": 0.6279,
+      "step": 1375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5221812504576193e-05,
+      "loss": 0.5963,
+      "step": 1376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.521185549800456e-05,
+      "loss": 0.6487,
+      "step": 1377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5201889538319154e-05,
+      "loss": 0.646,
+      "step": 1378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5191914631385565e-05,
+      "loss": 0.6471,
+      "step": 1379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.518193078307463e-05,
+      "loss": 0.6368,
+      "step": 1380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.517193799926247e-05,
+      "loss": 0.6461,
+      "step": 1381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5161936285830446e-05,
+      "loss": 0.6193,
+      "step": 1382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.515192564866519e-05,
+      "loss": 0.6742,
+      "step": 1383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.514190609365857e-05,
+      "loss": 0.6086,
+      "step": 1384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.513187762670771e-05,
+      "loss": 0.5982,
+      "step": 1385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.512184025371498e-05,
+      "loss": 0.6458,
+      "step": 1386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5111793980588006e-05,
+      "loss": 0.6521,
+      "step": 1387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5101738813239626e-05,
+      "loss": 0.6264,
+      "step": 1388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5091674757587925e-05,
+      "loss": 0.6382,
+      "step": 1389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5081601819556224e-05,
+      "loss": 0.6127,
+      "step": 1390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.507152000507307e-05,
+      "loss": 0.6567,
+      "step": 1391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5061429320072225e-05,
+      "loss": 0.6047,
+      "step": 1392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.505132977049268e-05,
+      "loss": 0.6442,
+      "step": 1393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5041221362278644e-05,
+      "loss": 0.594,
+      "step": 1394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.503110410137954e-05,
+      "loss": 0.6089,
+      "step": 1395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.502097799374998e-05,
+      "loss": 0.618,
+      "step": 1396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.501084304534984e-05,
+      "loss": 0.646,
+      "step": 1397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.500069926214413e-05,
+      "loss": 0.6403,
+      "step": 1398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.49905466501031e-05,
+      "loss": 0.6015,
+      "step": 1399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4980385215202185e-05,
+      "loss": 0.6205,
+      "step": 1400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.497021496342203e-05,
+      "loss": 0.6253,
+      "step": 1401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.496003590074843e-05,
+      "loss": 0.6279,
+      "step": 1402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4949848033172406e-05,
+      "loss": 0.6189,
+      "step": 1403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4939651366690144e-05,
+      "loss": 0.641,
+      "step": 1404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.492944590730301e-05,
+      "loss": 0.6491,
+      "step": 1405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.491923166101754e-05,
+      "loss": 0.614,
+      "step": 1406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.490900863384545e-05,
+      "loss": 0.6363,
+      "step": 1407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.489877683180362e-05,
+      "loss": 0.6216,
+      "step": 1408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.48885362609141e-05,
+      "loss": 0.6398,
+      "step": 1409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4878286927204104e-05,
+      "loss": 0.6509,
+      "step": 1410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.486802883670597e-05,
+      "loss": 0.6269,
+      "step": 1411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.485776199545724e-05,
+      "loss": 0.6137,
+      "step": 1412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4847486409500575e-05,
+      "loss": 0.6215,
+      "step": 1413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4837202084883785e-05,
+      "loss": 0.6185,
+      "step": 1414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.482690902765984e-05,
+      "loss": 0.6418,
+      "step": 1415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4816607243886834e-05,
+      "loss": 0.6164,
+      "step": 1416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.480629673962799e-05,
+      "loss": 0.6232,
+      "step": 1417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4795977520951684e-05,
+      "loss": 0.625,
+      "step": 1418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.478564959393141e-05,
+      "loss": 0.6255,
+      "step": 1419
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4775312964645795e-05,
+      "loss": 0.5972,
+      "step": 1420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.476496763917857e-05,
+      "loss": 0.6187,
+      "step": 1421
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.475461362361861e-05,
+      "loss": 0.5953,
+      "step": 1422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4744250924059876e-05,
+      "loss": 0.623,
+      "step": 1423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4733879546601466e-05,
+      "loss": 0.6423,
+      "step": 1424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.472349949734757e-05,
+      "loss": 0.6241,
+      "step": 1425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.471311078240749e-05,
+      "loss": 0.6487,
+      "step": 1426
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.470271340789562e-05,
+      "loss": 0.6553,
+      "step": 1427
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.469230737993146e-05,
+      "loss": 0.6057,
+      "step": 1428
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.468189270463959e-05,
+      "loss": 0.6515,
+      "step": 1429
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.467146938814969e-05,
+      "loss": 0.5852,
+      "step": 1430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4661037436596526e-05,
+      "loss": 0.6353,
+      "step": 1431
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.465059685611995e-05,
+      "loss": 0.6188,
+      "step": 1432
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4640147652864884e-05,
+      "loss": 0.6427,
+      "step": 1433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.462968983298132e-05,
+      "loss": 0.6141,
+      "step": 1434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.461922340262433e-05,
+      "loss": 0.6459,
+      "step": 1435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4608748367954064e-05,
+      "loss": 0.6219,
+      "step": 1436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.459826473513571e-05,
+      "loss": 0.6291,
+      "step": 1437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.458777251033955e-05,
+      "loss": 0.6353,
+      "step": 1438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.457727169974089e-05,
+      "loss": 0.6468,
+      "step": 1439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.45667623095201e-05,
+      "loss": 0.6439,
+      "step": 1440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4556244345862614e-05,
+      "loss": 0.6424,
+      "step": 1441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.45457178149589e-05,
+      "loss": 0.6418,
+      "step": 1442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4535182723004466e-05,
+      "loss": 0.6128,
+      "step": 1443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.452463907619986e-05,
+      "loss": 0.6205,
+      "step": 1444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.451408688075068e-05,
+      "loss": 0.6574,
+      "step": 1445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.450352614286753e-05,
+      "loss": 0.6283,
+      "step": 1446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.449295686876606e-05,
+      "loss": 0.6409,
+      "step": 1447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.448237906466694e-05,
+      "loss": 0.6259,
+      "step": 1448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4471792736795853e-05,
+      "loss": 0.6135,
+      "step": 1449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.446119789138351e-05,
+      "loss": 0.6156,
+      "step": 1450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.445059453466563e-05,
+      "loss": 0.6229,
+      "step": 1451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.443998267288294e-05,
+      "loss": 0.6323,
+      "step": 1452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4429362312281175e-05,
+      "loss": 0.6164,
+      "step": 1453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.441873345911106e-05,
+      "loss": 0.6171,
+      "step": 1454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.440809611962835e-05,
+      "loss": 0.6218,
+      "step": 1455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.439745030009377e-05,
+      "loss": 0.6182,
+      "step": 1456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.438679600677303e-05,
+      "loss": 0.6492,
+      "step": 1457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.437613324593684e-05,
+      "loss": 0.6389,
+      "step": 1458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4365462023860895e-05,
+      "loss": 0.6255,
+      "step": 1459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.435478234682586e-05,
+      "loss": 0.6484,
+      "step": 1460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.43440942211174e-05,
+      "loss": 0.5902,
+      "step": 1461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.433339765302612e-05,
+      "loss": 0.638,
+      "step": 1462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.432269264884761e-05,
+      "loss": 0.6167,
+      "step": 1463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.431197921488242e-05,
+      "loss": 0.631,
+      "step": 1464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4301257357436085e-05,
+      "loss": 0.6389,
+      "step": 1465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.429052708281906e-05,
+      "loss": 0.6275,
+      "step": 1466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.427978839734678e-05,
+      "loss": 0.6297,
+      "step": 1467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4269041307339615e-05,
+      "loss": 0.6471,
+      "step": 1468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.42582858191229e-05,
+      "loss": 0.6218,
+      "step": 1469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.42475219390269e-05,
+      "loss": 0.6043,
+      "step": 1470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.423674967338681e-05,
+      "loss": 0.605,
+      "step": 1471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.422596902854279e-05,
+      "loss": 0.634,
+      "step": 1472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.42151800108399e-05,
+      "loss": 0.6465,
+      "step": 1473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.420438262662814e-05,
+      "loss": 0.6302,
+      "step": 1474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4193576882262455e-05,
+      "loss": 0.6367,
+      "step": 1475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4182762784102665e-05,
+      "loss": 0.6067,
+      "step": 1476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4171940338513546e-05,
+      "loss": 0.6386,
+      "step": 1477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.416110955186477e-05,
+      "loss": 0.635,
+      "step": 1478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.415027043053093e-05,
+      "loss": 0.6196,
+      "step": 1479
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4139422980891516e-05,
+      "loss": 0.6173,
+      "step": 1480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.41285672093309e-05,
+      "loss": 0.6143,
+      "step": 1481
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4117703122238396e-05,
+      "loss": 0.6081,
+      "step": 1482
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.410683072600818e-05,
+      "loss": 0.6394,
+      "step": 1483
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.409595002703932e-05,
+      "loss": 0.609,
+      "step": 1484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4085061031735794e-05,
+      "loss": 0.6306,
+      "step": 1485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4074163746506425e-05,
+      "loss": 0.6215,
+      "step": 1486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.406325817776495e-05,
+      "loss": 0.5881,
+      "step": 1487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4052344331929965e-05,
+      "loss": 0.5941,
+      "step": 1488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4041422215424943e-05,
+      "loss": 0.6087,
+      "step": 1489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.403049183467822e-05,
+      "loss": 0.6212,
+      "step": 1490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.401955319612299e-05,
+      "loss": 0.6281,
+      "step": 1491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4008606306197336e-05,
+      "loss": 0.6555,
+      "step": 1492
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.399765117134415e-05,
+      "loss": 0.6587,
+      "step": 1493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.398668779801122e-05,
+      "loss": 0.6275,
+      "step": 1494
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.397571619265117e-05,
+      "loss": 0.6066,
+      "step": 1495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.396473636172146e-05,
+      "loss": 0.5995,
+      "step": 1496
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.39537483116844e-05,
+      "loss": 0.6505,
+      "step": 1497
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3942752049007116e-05,
+      "loss": 0.6714,
+      "step": 1498
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.393174758016161e-05,
+      "loss": 0.6228,
+      "step": 1499
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.392073491162467e-05,
+      "loss": 0.6053,
+      "step": 1500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.390971404987795e-05,
+      "loss": 0.5939,
+      "step": 1501
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.389868500140789e-05,
+      "loss": 0.6364,
+      "step": 1502
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.388764777270578e-05,
+      "loss": 0.6083,
+      "step": 1503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3876602370267685e-05,
+      "loss": 0.6422,
+      "step": 1504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.386554880059452e-05,
+      "loss": 0.6324,
+      "step": 1505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.385448707019199e-05,
+      "loss": 0.6016,
+      "step": 1506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.38434171855706e-05,
+      "loss": 0.6242,
+      "step": 1507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.383233915324567e-05,
+      "loss": 0.6203,
+      "step": 1508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.38212529797373e-05,
+      "loss": 0.6216,
+      "step": 1509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.381015867157038e-05,
+      "loss": 0.6062,
+      "step": 1510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.379905623527459e-05,
+      "loss": 0.6244,
+      "step": 1511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.37879456773844e-05,
+      "loss": 0.6212,
+      "step": 1512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.377682700443907e-05,
+      "loss": 0.6165,
+      "step": 1513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.376570022298262e-05,
+      "loss": 0.6355,
+      "step": 1514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.375456533956383e-05,
+      "loss": 0.6147,
+      "step": 1515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3743422360736275e-05,
+      "loss": 0.6367,
+      "step": 1516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.373227129305829e-05,
+      "loss": 0.6166,
+      "step": 1517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.372111214309296e-05,
+      "loss": 0.5627,
+      "step": 1518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.370994491740813e-05,
+      "loss": 0.5991,
+      "step": 1519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3698769622576404e-05,
+      "loss": 0.6186,
+      "step": 1520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.368758626517514e-05,
+      "loss": 0.6045,
+      "step": 1521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.36763948517864e-05,
+      "loss": 0.5791,
+      "step": 1522
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.366519538899706e-05,
+      "loss": 0.5786,
+      "step": 1523
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.365398788339867e-05,
+      "loss": 0.5912,
+      "step": 1524
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3642772341587536e-05,
+      "loss": 0.5976,
+      "step": 1525
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.363154877016471e-05,
+      "loss": 0.5856,
+      "step": 1526
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3620317175735945e-05,
+      "loss": 0.5852,
+      "step": 1527
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3609077564911724e-05,
+      "loss": 0.5716,
+      "step": 1528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.359782994430726e-05,
+      "loss": 0.5779,
+      "step": 1529
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.358657432054246e-05,
+      "loss": 0.5936,
+      "step": 1530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.357531070024196e-05,
+      "loss": 0.597,
+      "step": 1531
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.356403909003509e-05,
+      "loss": 0.5667,
+      "step": 1532
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.355275949655588e-05,
+      "loss": 0.5846,
+      "step": 1533
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3541471926443084e-05,
+      "loss": 0.5981,
+      "step": 1534
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.353017638634011e-05,
+      "loss": 0.5722,
+      "step": 1535
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.351887288289509e-05,
+      "loss": 0.5802,
+      "step": 1536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.350756142276083e-05,
+      "loss": 0.5906,
+      "step": 1537
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.349624201259483e-05,
+      "loss": 0.6028,
+      "step": 1538
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.348491465905925e-05,
+      "loss": 0.5825,
+      "step": 1539
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3473579368820934e-05,
+      "loss": 0.5652,
+      "step": 1540
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.34622361485514e-05,
+      "loss": 0.5831,
+      "step": 1541
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3450885004926846e-05,
+      "loss": 0.5927,
+      "step": 1542
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3439525944628096e-05,
+      "loss": 0.5865,
+      "step": 1543
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.342815897434067e-05,
+      "loss": 0.5939,
+      "step": 1544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.341678410075473e-05,
+      "loss": 0.5821,
+      "step": 1545
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3405401330565084e-05,
+      "loss": 0.5754,
+      "step": 1546
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.339401067047119e-05,
+      "loss": 0.579,
+      "step": 1547
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3382612127177166e-05,
+      "loss": 0.5957,
+      "step": 1548
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.337120570739175e-05,
+      "loss": 0.592,
+      "step": 1549
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.335979141782832e-05,
+      "loss": 0.6011,
+      "step": 1550
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3348369265204894e-05,
+      "loss": 0.5517,
+      "step": 1551
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3336939256244105e-05,
+      "loss": 0.5973,
+      "step": 1552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.332550139767322e-05,
+      "loss": 0.5874,
+      "step": 1553
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.331405569622413e-05,
+      "loss": 0.5705,
+      "step": 1554
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.330260215863332e-05,
+      "loss": 0.599,
+      "step": 1555
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.329114079164192e-05,
+      "loss": 0.6044,
+      "step": 1556
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.327967160199563e-05,
+      "loss": 0.5932,
+      "step": 1557
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3268194596444794e-05,
+      "loss": 0.5571,
+      "step": 1558
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.325670978174433e-05,
+      "loss": 0.6231,
+      "step": 1559
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3245217164653764e-05,
+      "loss": 0.5746,
+      "step": 1560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.323371675193719e-05,
+      "loss": 0.6091,
+      "step": 1561
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.322220855036333e-05,
+      "loss": 0.5841,
+      "step": 1562
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.321069256670546e-05,
+      "loss": 0.5724,
+      "step": 1563
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3199168807741445e-05,
+      "loss": 0.5911,
+      "step": 1564
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.318763728025373e-05,
+      "loss": 0.5852,
+      "step": 1565
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3176097991029315e-05,
+      "loss": 0.6015,
+      "step": 1566
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3164550946859806e-05,
+      "loss": 0.5814,
+      "step": 1567
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.315299615454134e-05,
+      "loss": 0.5753,
+      "step": 1568
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.314143362087462e-05,
+      "loss": 0.577,
+      "step": 1569
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.312986335266491e-05,
+      "loss": 0.5976,
+      "step": 1570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3118285356722034e-05,
+      "loss": 0.6,
+      "step": 1571
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.310669963986033e-05,
+      "loss": 0.5762,
+      "step": 1572
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3095106208898746e-05,
+      "loss": 0.6043,
+      "step": 1573
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.308350507066069e-05,
+      "loss": 0.5883,
+      "step": 1574
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3071896231974176e-05,
+      "loss": 0.5916,
+      "step": 1575
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.30602796996717e-05,
+      "loss": 0.5728,
+      "step": 1576
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3048655480590324e-05,
+      "loss": 0.584,
+      "step": 1577
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.303702358157161e-05,
+      "loss": 0.6096,
+      "step": 1578
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3025384009461636e-05,
+      "loss": 0.5645,
+      "step": 1579
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.301373677111103e-05,
+      "loss": 0.5856,
+      "step": 1580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.300208187337489e-05,
+      "loss": 0.5948,
+      "step": 1581
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.299041932311284e-05,
+      "loss": 0.6117,
+      "step": 1582
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.297874912718902e-05,
+      "loss": 0.5682,
+      "step": 1583
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.296707129247206e-05,
+      "loss": 0.5926,
+      "step": 1584
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.295538582583507e-05,
+      "loss": 0.6016,
+      "step": 1585
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.294369273415568e-05,
+      "loss": 0.5773,
+      "step": 1586
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.293199202431599e-05,
+      "loss": 0.5762,
+      "step": 1587
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2920283703202596e-05,
+      "loss": 0.592,
+      "step": 1588
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2908567777706555e-05,
+      "loss": 0.583,
+      "step": 1589
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2896844254723414e-05,
+      "loss": 0.626,
+      "step": 1590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.288511314115319e-05,
+      "loss": 0.6099,
+      "step": 1591
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.287337444390037e-05,
+      "loss": 0.576,
+      "step": 1592
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.286162816987388e-05,
+      "loss": 0.577,
+      "step": 1593
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2849874325987156e-05,
+      "loss": 0.558,
+      "step": 1594
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.283811291915804e-05,
+      "loss": 0.6132,
+      "step": 1595
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.282634395630884e-05,
+      "loss": 0.619,
+      "step": 1596
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.281456744436634e-05,
+      "loss": 0.5902,
+      "step": 1597
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2802783390261707e-05,
+      "loss": 0.5888,
+      "step": 1598
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.279099180093061e-05,
+      "loss": 0.582,
+      "step": 1599
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.277919268331312e-05,
+      "loss": 0.5781,
+      "step": 1600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.276738604435374e-05,
+      "loss": 0.6137,
+      "step": 1601
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.27555718910014e-05,
+      "loss": 0.6166,
+      "step": 1602
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2743750230209474e-05,
+      "loss": 0.6016,
+      "step": 1603
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.273192106893572e-05,
+      "loss": 0.5964,
+      "step": 1604
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.272008441414233e-05,
+      "loss": 0.5662,
+      "step": 1605
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.270824027279591e-05,
+      "loss": 0.597,
+      "step": 1606
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2696388651867464e-05,
+      "loss": 0.5833,
+      "step": 1607
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.26845295583324e-05,
+      "loss": 0.5929,
+      "step": 1608
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2672662999170526e-05,
+      "loss": 0.5638,
+      "step": 1609
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.266078898136605e-05,
+      "loss": 0.596,
+      "step": 1610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2648907511907544e-05,
+      "loss": 0.5785,
+      "step": 1611
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.263701859778801e-05,
+      "loss": 0.5881,
+      "step": 1612
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.262512224600478e-05,
+      "loss": 0.574,
+      "step": 1613
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2613218463559615e-05,
+      "loss": 0.5845,
+      "step": 1614
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.26013072574586e-05,
+      "loss": 0.5567,
+      "step": 1615
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2589388634712224e-05,
+      "loss": 0.5959,
+      "step": 1616
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2577462602335334e-05,
+      "loss": 0.5804,
+      "step": 1617
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.256552916734713e-05,
+      "loss": 0.6097,
+      "step": 1618
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.255358833677118e-05,
+      "loss": 0.5619,
+      "step": 1619
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.254164011763539e-05,
+      "loss": 0.6268,
+      "step": 1620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.252968451697203e-05,
+      "loss": 0.557,
+      "step": 1621
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2517721541817696e-05,
+      "loss": 0.5857,
+      "step": 1622
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.250575119921335e-05,
+      "loss": 0.5933,
+      "step": 1623
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2493773496204264e-05,
+      "loss": 0.6046,
+      "step": 1624
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.248178843984006e-05,
+      "loss": 0.5699,
+      "step": 1625
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.246979603717467e-05,
+      "loss": 0.6157,
+      "step": 1626
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2457796295266384e-05,
+      "loss": 0.563,
+      "step": 1627
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2445789221177775e-05,
+      "loss": 0.613,
+      "step": 1628
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2433774821975744e-05,
+      "loss": 0.6049,
+      "step": 1629
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.242175310473151e-05,
+      "loss": 0.6078,
+      "step": 1630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2409724076520585e-05,
+      "loss": 0.5926,
+      "step": 1631
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.239768774442281e-05,
+      "loss": 0.6058,
+      "step": 1632
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2385644115522295e-05,
+      "loss": 0.6173,
+      "step": 1633
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.237359319690746e-05,
+      "loss": 0.5895,
+      "step": 1634
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2361534995671005e-05,
+      "loss": 0.5626,
+      "step": 1635
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2349469518909935e-05,
+      "loss": 0.5654,
+      "step": 1636
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.233739677372553e-05,
+      "loss": 0.6118,
+      "step": 1637
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.232531676722333e-05,
+      "loss": 0.5817,
+      "step": 1638
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2313229506513167e-05,
+      "loss": 0.6104,
+      "step": 1639
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2301134998709146e-05,
+      "loss": 0.5578,
+      "step": 1640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.228903325092963e-05,
+      "loss": 0.5803,
+      "step": 1641
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.227692427029722e-05,
+      "loss": 0.5877,
+      "step": 1642
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.226480806393883e-05,
+      "loss": 0.5781,
+      "step": 1643
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.225268463898557e-05,
+      "loss": 0.5942,
+      "step": 1644
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.224055400257284e-05,
+      "loss": 0.6428,
+      "step": 1645
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.222841616184025e-05,
+      "loss": 0.565,
+      "step": 1646
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2216271123931667e-05,
+      "loss": 0.5977,
+      "step": 1647
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.220411889599521e-05,
+      "loss": 0.5739,
+      "step": 1648
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.219195948518319e-05,
+      "loss": 0.6093,
+      "step": 1649
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.21797928986522e-05,
+      "loss": 0.6007,
+      "step": 1650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2167619143562996e-05,
+      "loss": 0.5714,
+      "step": 1651
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2155438227080607e-05,
+      "loss": 0.5922,
+      "step": 1652
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2143250156374226e-05,
+      "loss": 0.5711,
+      "step": 1653
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.21310549386173e-05,
+      "loss": 0.5965,
+      "step": 1654
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.211885258098747e-05,
+      "loss": 0.5958,
+      "step": 1655
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.210664309066656e-05,
+      "loss": 0.574,
+      "step": 1656
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2094426474840626e-05,
+      "loss": 0.5812,
+      "step": 1657
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.208220274069988e-05,
+      "loss": 0.6108,
+      "step": 1658
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2069971895438756e-05,
+      "loss": 0.5724,
+      "step": 1659
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2057733946255844e-05,
+      "loss": 0.5929,
+      "step": 1660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2045488900353946e-05,
+      "loss": 0.6072,
+      "step": 1661
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.203323676494003e-05,
+      "loss": 0.6117,
+      "step": 1662
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.20209775472252e-05,
+      "loss": 0.5827,
+      "step": 1663
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.20087112544248e-05,
+      "loss": 0.6048,
+      "step": 1664
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.199643789375828e-05,
+      "loss": 0.567,
+      "step": 1665
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.198415747244926e-05,
+      "loss": 0.543,
+      "step": 1666
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.197186999772555e-05,
+      "loss": 0.6013,
+      "step": 1667
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1959575476819054e-05,
+      "loss": 0.5862,
+      "step": 1668
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.194727391696588e-05,
+      "loss": 0.5755,
+      "step": 1669
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.193496532540624e-05,
+      "loss": 0.5896,
+      "step": 1670
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.192264970938451e-05,
+      "loss": 0.591,
+      "step": 1671
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1910327076149176e-05,
+      "loss": 0.5806,
+      "step": 1672
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.189799743295287e-05,
+      "loss": 0.6003,
+      "step": 1673
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.188566078705235e-05,
+      "loss": 0.5674,
+      "step": 1674
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1873317145708497e-05,
+      "loss": 0.5956,
+      "step": 1675
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.186096651618629e-05,
+      "loss": 0.5823,
+      "step": 1676
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1848608905754835e-05,
+      "loss": 0.613,
+      "step": 1677
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.183624432168736e-05,
+      "loss": 0.5784,
+      "step": 1678
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.182387277126118e-05,
+      "loss": 0.6092,
+      "step": 1679
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1811494261757706e-05,
+      "loss": 0.5725,
+      "step": 1680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1799108800462466e-05,
+      "loss": 0.5724,
+      "step": 1681
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1786716394665056e-05,
+      "loss": 0.5913,
+      "step": 1682
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1774317051659176e-05,
+      "loss": 0.5937,
+      "step": 1683
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1761910778742595e-05,
+      "loss": 0.591,
+      "step": 1684
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1749497583217173e-05,
+      "loss": 0.5842,
+      "step": 1685
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1737077472388854e-05,
+      "loss": 0.5845,
+      "step": 1686
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.172465045356762e-05,
+      "loss": 0.5991,
+      "step": 1687
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1712216534067536e-05,
+      "loss": 0.6118,
+      "step": 1688
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.169977572120675e-05,
+      "loss": 0.5785,
+      "step": 1689
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.168732802230742e-05,
+      "loss": 0.5817,
+      "step": 1690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1674873444695804e-05,
+      "loss": 0.6242,
+      "step": 1691
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.166241199570217e-05,
+      "loss": 0.598,
+      "step": 1692
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.164994368266087e-05,
+      "loss": 0.5654,
+      "step": 1693
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.163746851291025e-05,
+      "loss": 0.5914,
+      "step": 1694
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1624986493792735e-05,
+      "loss": 0.5797,
+      "step": 1695
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.161249763265476e-05,
+      "loss": 0.5812,
+      "step": 1696
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.160000193684677e-05,
+      "loss": 0.6093,
+      "step": 1697
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1587499413723275e-05,
+      "loss": 0.5953,
+      "step": 1698
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1574990070642766e-05,
+      "loss": 0.5983,
+      "step": 1699
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.156247391496777e-05,
+      "loss": 0.5799,
+      "step": 1700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.15499509540648e-05,
+      "loss": 0.5733,
+      "step": 1701
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.153742119530441e-05,
+      "loss": 0.5862,
+      "step": 1702
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.152488464606112e-05,
+      "loss": 0.5475,
+      "step": 1703
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.151234131371348e-05,
+      "loss": 0.5472,
+      "step": 1704
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.149979120564398e-05,
+      "loss": 0.5699,
+      "step": 1705
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1487234329239164e-05,
+      "loss": 0.5947,
+      "step": 1706
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1474670691889515e-05,
+      "loss": 0.5894,
+      "step": 1707
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1462100300989504e-05,
+      "loss": 0.5897,
+      "step": 1708
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.144952316393758e-05,
+      "loss": 0.5532,
+      "step": 1709
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.143693928813617e-05,
+      "loss": 0.5912,
+      "step": 1710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.142434868099166e-05,
+      "loss": 0.6249,
+      "step": 1711
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1411751349914395e-05,
+      "loss": 0.6007,
+      "step": 1712
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.139914730231867e-05,
+      "loss": 0.5873,
+      "step": 1713
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.138653654562275e-05,
+      "loss": 0.6147,
+      "step": 1714
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.137391908724885e-05,
+      "loss": 0.6165,
+      "step": 1715
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.136129493462312e-05,
+      "loss": 0.5649,
+      "step": 1716
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.134866409517564e-05,
+      "loss": 0.5805,
+      "step": 1717
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.133602657634044e-05,
+      "loss": 0.5803,
+      "step": 1718
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.132338238555549e-05,
+      "loss": 0.5742,
+      "step": 1719
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.131073153026266e-05,
+      "loss": 0.5971,
+      "step": 1720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.129807401790777e-05,
+      "loss": 0.6102,
+      "step": 1721
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.128540985594053e-05,
+      "loss": 0.5956,
+      "step": 1722
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1272739051814594e-05,
+      "loss": 0.5823,
+      "step": 1723
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.126006161298751e-05,
+      "loss": 0.5629,
+      "step": 1724
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1247377546920716e-05,
+      "loss": 0.6277,
+      "step": 1725
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.123468686107958e-05,
+      "loss": 0.5987,
+      "step": 1726
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.122198956293334e-05,
+      "loss": 0.6036,
+      "step": 1727
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.120928565995517e-05,
+      "loss": 0.5995,
+      "step": 1728
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1196575159622055e-05,
+      "loss": 0.5662,
+      "step": 1729
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1183858069414936e-05,
+      "loss": 0.5906,
+      "step": 1730
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.11711343968186e-05,
+      "loss": 0.5662,
+      "step": 1731
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1158404149321704e-05,
+      "loss": 0.5746,
+      "step": 1732
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1145667334416785e-05,
+      "loss": 0.5886,
+      "step": 1733
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.113292395960024e-05,
+      "loss": 0.5829,
+      "step": 1734
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.112017403237234e-05,
+      "loss": 0.5883,
+      "step": 1735
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.11074175602372e-05,
+      "loss": 0.5894,
+      "step": 1736
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.109465455070278e-05,
+      "loss": 0.5792,
+      "step": 1737
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1081885011280904e-05,
+      "loss": 0.6038,
+      "step": 1738
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.106910894948723e-05,
+      "loss": 0.5883,
+      "step": 1739
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1056326372841264e-05,
+      "loss": 0.598,
+      "step": 1740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.104353728886633e-05,
+      "loss": 0.5625,
+      "step": 1741
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.10307417050896e-05,
+      "loss": 0.5693,
+      "step": 1742
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.101793962904205e-05,
+      "loss": 0.5765,
+      "step": 1743
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1005131068258506e-05,
+      "loss": 0.6135,
+      "step": 1744
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0992316030277594e-05,
+      "loss": 0.5728,
+      "step": 1745
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.097949452264173e-05,
+      "loss": 0.5941,
+      "step": 1746
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.096666655289719e-05,
+      "loss": 0.5738,
+      "step": 1747
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.095383212859401e-05,
+      "loss": 0.5807,
+      "step": 1748
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.094099125728604e-05,
+      "loss": 0.5991,
+      "step": 1749
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.092814394653092e-05,
+      "loss": 0.583,
+      "step": 1750
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.091529020389009e-05,
+      "loss": 0.5967,
+      "step": 1751
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.090243003692877e-05,
+      "loss": 0.5763,
+      "step": 1752
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.088956345321596e-05,
+      "loss": 0.5896,
+      "step": 1753
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.087669046032443e-05,
+      "loss": 0.5964,
+      "step": 1754
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0863811065830734e-05,
+      "loss": 0.5774,
+      "step": 1755
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0850925277315193e-05,
+      "loss": 0.599,
+      "step": 1756
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.083803310236188e-05,
+      "loss": 0.6236,
+      "step": 1757
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.082513454855863e-05,
+      "loss": 0.5719,
+      "step": 1758
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.081222962349705e-05,
+      "loss": 0.5876,
+      "step": 1759
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.079931833477247e-05,
+      "loss": 0.5558,
+      "step": 1760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0786400689983986e-05,
+      "loss": 0.629,
+      "step": 1761
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0773476696734425e-05,
+      "loss": 0.5926,
+      "step": 1762
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.076054636263036e-05,
+      "loss": 0.6155,
+      "step": 1763
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.074760969528208e-05,
+      "loss": 0.5855,
+      "step": 1764
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.073466670230361e-05,
+      "loss": 0.5943,
+      "step": 1765
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.07217173913127e-05,
+      "loss": 0.601,
+      "step": 1766
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.070876176993082e-05,
+      "loss": 0.5614,
+      "step": 1767
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.069579984578315e-05,
+      "loss": 0.5737,
+      "step": 1768
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.068283162649858e-05,
+      "loss": 0.5851,
+      "step": 1769
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.06698571197097e-05,
+      "loss": 0.5833,
+      "step": 1770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.065687633305282e-05,
+      "loss": 0.5766,
+      "step": 1771
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0643889274167926e-05,
+      "loss": 0.5864,
+      "step": 1772
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.063089595069869e-05,
+      "loss": 0.6067,
+      "step": 1773
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.06178963702925e-05,
+      "loss": 0.5582,
+      "step": 1774
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.060489054060041e-05,
+      "loss": 0.5932,
+      "step": 1775
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.059187846927713e-05,
+      "loss": 0.5719,
+      "step": 1776
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.057886016398109e-05,
+      "loss": 0.5608,
+      "step": 1777
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0565835632374355e-05,
+      "loss": 0.5757,
+      "step": 1778
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.055280488212266e-05,
+      "loss": 0.6325,
+      "step": 1779
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0539767920895404e-05,
+      "loss": 0.6047,
+      "step": 1780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.052672475636565e-05,
+      "loss": 0.562,
+      "step": 1781
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0513675396210094e-05,
+      "loss": 0.5809,
+      "step": 1782
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.05006198481091e-05,
+      "loss": 0.5973,
+      "step": 1783
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0487558119746645e-05,
+      "loss": 0.5975,
+      "step": 1784
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.047449021881038e-05,
+      "loss": 0.582,
+      "step": 1785
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0461416152991555e-05,
+      "loss": 0.6167,
+      "step": 1786
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0448335929985066e-05,
+      "loss": 0.608,
+      "step": 1787
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0435249557489436e-05,
+      "loss": 0.5738,
+      "step": 1788
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.042215704320679e-05,
+      "loss": 0.5844,
+      "step": 1789
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.040905839484289e-05,
+      "loss": 0.6049,
+      "step": 1790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.039595362010709e-05,
+      "loss": 0.6006,
+      "step": 1791
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0382842726712366e-05,
+      "loss": 0.6015,
+      "step": 1792
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0369725722375274e-05,
+      "loss": 0.5736,
+      "step": 1793
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0356602614815984e-05,
+      "loss": 0.6059,
+      "step": 1794
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0343473411758255e-05,
+      "loss": 0.5627,
+      "step": 1795
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.033033812092943e-05,
+      "loss": 0.6299,
+      "step": 1796
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0317196750060435e-05,
+      "loss": 0.6115,
+      "step": 1797
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0304049306885787e-05,
+      "loss": 0.6059,
+      "step": 1798
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.029089579914355e-05,
+      "loss": 0.5674,
+      "step": 1799
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0277736234575378e-05,
+      "loss": 0.5917,
+      "step": 1800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0264570620926487e-05,
+      "loss": 0.5764,
+      "step": 1801
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0251398965945654e-05,
+      "loss": 0.5796,
+      "step": 1802
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0238221277385203e-05,
+      "loss": 0.5721,
+      "step": 1803
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0225037563001022e-05,
+      "loss": 0.5876,
+      "step": 1804
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0211847830552545e-05,
+      "loss": 0.5679,
+      "step": 1805
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0198652087802722e-05,
+      "loss": 0.5701,
+      "step": 1806
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0185450342518075e-05,
+      "loss": 0.5983,
+      "step": 1807
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0172242602468637e-05,
+      "loss": 0.5998,
+      "step": 1808
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0159028875427982e-05,
+      "loss": 0.5949,
+      "step": 1809
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0145809169173198e-05,
+      "loss": 0.5968,
+      "step": 1810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0132583491484895e-05,
+      "loss": 0.5866,
+      "step": 1811
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.011935185014721e-05,
+      "loss": 0.5928,
+      "step": 1812
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0106114252947763e-05,
+      "loss": 0.596,
+      "step": 1813
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.009287070767771e-05,
+      "loss": 0.562,
+      "step": 1814
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.007962122213168e-05,
+      "loss": 0.6103,
+      "step": 1815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0066365804107813e-05,
+      "loss": 0.5723,
+      "step": 1816
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.005310446140775e-05,
+      "loss": 0.5732,
+      "step": 1817
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.00398372018366e-05,
+      "loss": 0.5804,
+      "step": 1818
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0026564033202966e-05,
+      "loss": 0.5709,
+      "step": 1819
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0013284963318915e-05,
+      "loss": 0.5952,
+      "step": 1820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5615,
+      "step": 1821
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9986709151065257e-05,
+      "loss": 0.5667,
+      "step": 1822
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9973412424337146e-05,
+      "loss": 0.5912,
+      "step": 1823
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9960109827641628e-05,
+      "loss": 0.5762,
+      "step": 1824
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9946801368808088e-05,
+      "loss": 0.5869,
+      "step": 1825
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9933487055669382e-05,
+      "loss": 0.5867,
+      "step": 1826
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9920166896061795e-05,
+      "loss": 0.5914,
+      "step": 1827
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.990684089782507e-05,
+      "loss": 0.5684,
+      "step": 1828
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9893509068802368e-05,
+      "loss": 0.5973,
+      "step": 1829
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.98801714168403e-05,
+      "loss": 0.5878,
+      "step": 1830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9866827949788886e-05,
+      "loss": 0.5831,
+      "step": 1831
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.985347867550159e-05,
+      "loss": 0.6072,
+      "step": 1832
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9840123601835275e-05,
+      "loss": 0.6081,
+      "step": 1833
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.982676273665023e-05,
+      "loss": 0.5683,
+      "step": 1834
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9813396087810134e-05,
+      "loss": 0.5794,
+      "step": 1835
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.98000236631821e-05,
+      "loss": 0.5792,
+      "step": 1836
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9786645470636615e-05,
+      "loss": 0.5807,
+      "step": 1837
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9773261518047568e-05,
+      "loss": 0.5703,
+      "step": 1838
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9759871813292237e-05,
+      "loss": 0.574,
+      "step": 1839
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9746476364251294e-05,
+      "loss": 0.5886,
+      "step": 1840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.973307517880878e-05,
+      "loss": 0.5975,
+      "step": 1841
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.971966826485212e-05,
+      "loss": 0.5868,
+      "step": 1842
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9706255630272114e-05,
+      "loss": 0.6163,
+      "step": 1843
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9692837282962922e-05,
+      "loss": 0.6133,
+      "step": 1844
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9679413230822063e-05,
+      "loss": 0.6387,
+      "step": 1845
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9665983481750413e-05,
+      "loss": 0.5855,
+      "step": 1846
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.965254804365222e-05,
+      "loss": 0.5801,
+      "step": 1847
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9639106924435057e-05,
+      "loss": 0.5451,
+      "step": 1848
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.962566013200986e-05,
+      "loss": 0.575,
+      "step": 1849
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.961220767429088e-05,
+      "loss": 0.5863,
+      "step": 1850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.959874955919573e-05,
+      "loss": 0.586,
+      "step": 1851
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9585285794645333e-05,
+      "loss": 0.5916,
+      "step": 1852
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9571816388563944e-05,
+      "loss": 0.5654,
+      "step": 1853
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.955834134887914e-05,
+      "loss": 0.5684,
+      "step": 1854
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9544860683521812e-05,
+      "loss": 0.6163,
+      "step": 1855
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9531374400426158e-05,
+      "loss": 0.5709,
+      "step": 1856
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9517882507529692e-05,
+      "loss": 0.5688,
+      "step": 1857
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9504385012773224e-05,
+      "loss": 0.6002,
+      "step": 1858
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9490881924100847e-05,
+      "loss": 0.5905,
+      "step": 1859
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9477373249459974e-05,
+      "loss": 0.5981,
+      "step": 1860
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.946385899680129e-05,
+      "loss": 0.5653,
+      "step": 1861
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9450339174078767e-05,
+      "loss": 0.5555,
+      "step": 1862
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.943681378924964e-05,
+      "loss": 0.6002,
+      "step": 1863
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.942328285027446e-05,
+      "loss": 0.5708,
+      "step": 1864
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9409746365116984e-05,
+      "loss": 0.5692,
+      "step": 1865
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9396204341744285e-05,
+      "loss": 0.5731,
+      "step": 1866
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.938265678812667e-05,
+      "loss": 0.5784,
+      "step": 1867
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9369103712237718e-05,
+      "loss": 0.6032,
+      "step": 1868
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9355545122054243e-05,
+      "loss": 0.6076,
+      "step": 1869
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.934198102555631e-05,
+      "loss": 0.5609,
+      "step": 1870
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.932841143072723e-05,
+      "loss": 0.6019,
+      "step": 1871
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.931483634555354e-05,
+      "loss": 0.5857,
+      "step": 1872
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9301255778025014e-05,
+      "loss": 0.5984,
+      "step": 1873
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9287669736134653e-05,
+      "loss": 0.6351,
+      "step": 1874
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9274078227878678e-05,
+      "loss": 0.6195,
+      "step": 1875
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.926048126125654e-05,
+      "loss": 0.5589,
+      "step": 1876
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.924687884427087e-05,
+      "loss": 0.5904,
+      "step": 1877
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.923327098492755e-05,
+      "loss": 0.5964,
+      "step": 1878
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9219657691235626e-05,
+      "loss": 0.5922,
+      "step": 1879
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.920603897120737e-05,
+      "loss": 0.5893,
+      "step": 1880
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9192414832858237e-05,
+      "loss": 0.5857,
+      "step": 1881
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.917878528420686e-05,
+      "loss": 0.6,
+      "step": 1882
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9165150333275084e-05,
+      "loss": 0.5926,
+      "step": 1883
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9151509988087912e-05,
+      "loss": 0.5523,
+      "step": 1884
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9137864256673517e-05,
+      "loss": 0.5933,
+      "step": 1885
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9124213147063263e-05,
+      "loss": 0.6164,
+      "step": 1886
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9110556667291667e-05,
+      "loss": 0.6005,
+      "step": 1887
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9096894825396408e-05,
+      "loss": 0.5894,
+      "step": 1888
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9083227629418318e-05,
+      "loss": 0.5898,
+      "step": 1889
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9069555087401387e-05,
+      "loss": 0.6098,
+      "step": 1890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9055877207392752e-05,
+      "loss": 0.5781,
+      "step": 1891
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9042193997442677e-05,
+      "loss": 0.5838,
+      "step": 1892
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.902850546560458e-05,
+      "loss": 0.6036,
+      "step": 1893
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9014811619935007e-05,
+      "loss": 0.5856,
+      "step": 1894
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9001112468493624e-05,
+      "loss": 0.5764,
+      "step": 1895
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.898740801934323e-05,
+      "loss": 0.5903,
+      "step": 1896
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8973698280549734e-05,
+      "loss": 0.6113,
+      "step": 1897
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8959983260182166e-05,
+      "loss": 0.6201,
+      "step": 1898
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8946262966312652e-05,
+      "loss": 0.6009,
+      "step": 1899
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.893253740701644e-05,
+      "loss": 0.5819,
+      "step": 1900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8918806590371866e-05,
+      "loss": 0.5815,
+      "step": 1901
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8905070524460348e-05,
+      "loss": 0.5726,
+      "step": 1902
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8891329217366423e-05,
+      "loss": 0.5925,
+      "step": 1903
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8877582677177688e-05,
+      "loss": 0.5641,
+      "step": 1904
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.886383091198483e-05,
+      "loss": 0.6086,
+      "step": 1905
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.885007392988161e-05,
+      "loss": 0.5498,
+      "step": 1906
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8836311738964852e-05,
+      "loss": 0.5762,
+      "step": 1907
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.882254434733447e-05,
+      "loss": 0.599,
+      "step": 1908
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8808771763093407e-05,
+      "loss": 0.5931,
+      "step": 1909
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8794993994347683e-05,
+      "loss": 0.5954,
+      "step": 1910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.878121104920636e-05,
+      "loss": 0.5818,
+      "step": 1911
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.876742293578155e-05,
+      "loss": 0.5965,
+      "step": 1912
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.875362966218841e-05,
+      "loss": 0.5553,
+      "step": 1913
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8739831236545135e-05,
+      "loss": 0.5985,
+      "step": 1914
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8726027666972936e-05,
+      "loss": 0.5922,
+      "step": 1915
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.871221896159608e-05,
+      "loss": 0.6141,
+      "step": 1916
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8698405128541824e-05,
+      "loss": 0.5584,
+      "step": 1917
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8684586175940468e-05,
+      "loss": 0.5885,
+      "step": 1918
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8670762111925313e-05,
+      "loss": 0.5725,
+      "step": 1919
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8656932944632675e-05,
+      "loss": 0.5862,
+      "step": 1920
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8643098682201875e-05,
+      "loss": 0.5573,
+      "step": 1921
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8629259332775218e-05,
+      "loss": 0.6172,
+      "step": 1922
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8615414904498026e-05,
+      "loss": 0.5732,
+      "step": 1923
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8601565405518584e-05,
+      "loss": 0.5818,
+      "step": 1924
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8587710843988187e-05,
+      "loss": 0.5557,
+      "step": 1925
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8573851228061084e-05,
+      "loss": 0.6035,
+      "step": 1926
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.855998656589453e-05,
+      "loss": 0.574,
+      "step": 1927
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8546116865648717e-05,
+      "loss": 0.5619,
+      "step": 1928
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8532242135486826e-05,
+      "loss": 0.5845,
+      "step": 1929
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8518362383574988e-05,
+      "loss": 0.5901,
+      "step": 1930
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8504477618082295e-05,
+      "loss": 0.5871,
+      "step": 1931
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.849058784718078e-05,
+      "loss": 0.5932,
+      "step": 1932
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8476693079045432e-05,
+      "loss": 0.5814,
+      "step": 1933
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8462793321854182e-05,
+      "loss": 0.6062,
+      "step": 1934
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8448888583787882e-05,
+      "loss": 0.5773,
+      "step": 1935
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8434978873030338e-05,
+      "loss": 0.592,
+      "step": 1936
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.842106419776827e-05,
+      "loss": 0.5763,
+      "step": 1937
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8407144566191315e-05,
+      "loss": 0.5739,
+      "step": 1938
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8393219986492036e-05,
+      "loss": 0.5935,
+      "step": 1939
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8379290466865906e-05,
+      "loss": 0.5979,
+      "step": 1940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8365356015511305e-05,
+      "loss": 0.5588,
+      "step": 1941
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8351416640629515e-05,
+      "loss": 0.621,
+      "step": 1942
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8337472350424728e-05,
+      "loss": 0.5917,
+      "step": 1943
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8323523153104e-05,
+      "loss": 0.5635,
+      "step": 1944
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.83095690568773e-05,
+      "loss": 0.6297,
+      "step": 1945
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8295610069957472e-05,
+      "loss": 0.5993,
+      "step": 1946
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.828164620056024e-05,
+      "loss": 0.5832,
+      "step": 1947
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8267677456904206e-05,
+      "loss": 0.564,
+      "step": 1948
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8253703847210833e-05,
+      "loss": 0.5815,
+      "step": 1949
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.823972537970445e-05,
+      "loss": 0.5867,
+      "step": 1950
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8225742062612236e-05,
+      "loss": 0.5785,
+      "step": 1951
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.821175390416425e-05,
+      "loss": 0.555,
+      "step": 1952
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8197760912593372e-05,
+      "loss": 0.615,
+      "step": 1953
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.818376309613535e-05,
+      "loss": 0.5861,
+      "step": 1954
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8169760463028758e-05,
+      "loss": 0.5689,
+      "step": 1955
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8155753021515003e-05,
+      "loss": 0.542,
+      "step": 1956
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8141740779838343e-05,
+      "loss": 0.5641,
+      "step": 1957
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.812772374624582e-05,
+      "loss": 0.6028,
+      "step": 1958
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8113701928987336e-05,
+      "loss": 0.5969,
+      "step": 1959
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.809967533631559e-05,
+      "loss": 0.6118,
+      "step": 1960
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.80856439764861e-05,
+      "loss": 0.58,
+      "step": 1961
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.807160785775719e-05,
+      "loss": 0.6201,
+      "step": 1962
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8057566988389964e-05,
+      "loss": 0.5937,
+      "step": 1963
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.804352137664835e-05,
+      "loss": 0.6241,
+      "step": 1964
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8029471030799054e-05,
+      "loss": 0.5642,
+      "step": 1965
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8015415959111568e-05,
+      "loss": 0.6147,
+      "step": 1966
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8001356169858172e-05,
+      "loss": 0.5905,
+      "step": 1967
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.798729167131391e-05,
+      "loss": 0.5976,
+      "step": 1968
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7973222471756618e-05,
+      "loss": 0.6068,
+      "step": 1969
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.795914857946687e-05,
+      "loss": 0.5574,
+      "step": 1970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7945070002728028e-05,
+      "loss": 0.5873,
+      "step": 1971
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.79309867498262e-05,
+      "loss": 0.6099,
+      "step": 1972
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7916898829050243e-05,
+      "loss": 0.553,
+      "step": 1973
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7902806248691777e-05,
+      "loss": 0.5692,
+      "step": 1974
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7888709017045146e-05,
+      "loss": 0.5923,
+      "step": 1975
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7874607142407436e-05,
+      "loss": 0.571,
+      "step": 1976
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7860500633078475e-05,
+      "loss": 0.5773,
+      "step": 1977
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7846389497360807e-05,
+      "loss": 0.556,
+      "step": 1978
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.783227374355971e-05,
+      "loss": 0.5806,
+      "step": 1979
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7818153379983173e-05,
+      "loss": 0.5819,
+      "step": 1980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7804028414941898e-05,
+      "loss": 0.6065,
+      "step": 1981
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7789898856749297e-05,
+      "loss": 0.5812,
+      "step": 1982
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7775764713721485e-05,
+      "loss": 0.603,
+      "step": 1983
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7761625994177275e-05,
+      "loss": 0.579,
+      "step": 1984
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7747482706438176e-05,
+      "loss": 0.6223,
+      "step": 1985
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7733334858828376e-05,
+      "loss": 0.5722,
+      "step": 1986
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7719182459674773e-05,
+      "loss": 0.5776,
+      "step": 1987
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7705025517306902e-05,
+      "loss": 0.6273,
+      "step": 1988
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7690864040057023e-05,
+      "loss": 0.5768,
+      "step": 1989
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7676698036260013e-05,
+      "loss": 0.5773,
+      "step": 1990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.766252751425344e-05,
+      "loss": 0.5956,
+      "step": 1991
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7648352482377542e-05,
+      "loss": 0.5611,
+      "step": 1992
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.763417294897519e-05,
+      "loss": 0.5732,
+      "step": 1993
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7619988922391917e-05,
+      "loss": 0.5881,
+      "step": 1994
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7605800410975896e-05,
+      "loss": 0.5613,
+      "step": 1995
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7591607423077932e-05,
+      "loss": 0.5978,
+      "step": 1996
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7577409967051477e-05,
+      "loss": 0.5664,
+      "step": 1997
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.756320805125261e-05,
+      "loss": 0.5861,
+      "step": 1998
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.754900168404003e-05,
+      "loss": 0.5792,
+      "step": 1999
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7534790873775064e-05,
+      "loss": 0.5587,
+      "step": 2000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7520575628821644e-05,
+      "loss": 0.5861,
+      "step": 2001
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7506355957546324e-05,
+      "loss": 0.5876,
+      "step": 2002
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7492131868318247e-05,
+      "loss": 0.5711,
+      "step": 2003
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.747790336950917e-05,
+      "loss": 0.5813,
+      "step": 2004
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7463670469493436e-05,
+      "loss": 0.5478,
+      "step": 2005
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.744943317664799e-05,
+      "loss": 0.6355,
+      "step": 2006
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7435191499352356e-05,
+      "loss": 0.5624,
+      "step": 2007
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7420945445988634e-05,
+      "loss": 0.5969,
+      "step": 2008
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.740669502494151e-05,
+      "loss": 0.5952,
+      "step": 2009
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.739244024459822e-05,
+      "loss": 0.5784,
+      "step": 2010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7378181113348597e-05,
+      "loss": 0.6212,
+      "step": 2011
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7363917639585004e-05,
+      "loss": 0.5857,
+      "step": 2012
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7349649831702387e-05,
+      "loss": 0.6087,
+      "step": 2013
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.733537769809822e-05,
+      "loss": 0.5858,
+      "step": 2014
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7321101247172533e-05,
+      "loss": 0.5852,
+      "step": 2015
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7306820487327906e-05,
+      "loss": 0.6111,
+      "step": 2016
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7292535426969436e-05,
+      "loss": 0.6155,
+      "step": 2017
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.727824607450476e-05,
+      "loss": 0.5967,
+      "step": 2018
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7263952438344047e-05,
+      "loss": 0.6164,
+      "step": 2019
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7249654526899977e-05,
+      "loss": 0.571,
+      "step": 2020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.723535234858776e-05,
+      "loss": 0.6191,
+      "step": 2021
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7221045911825102e-05,
+      "loss": 0.5902,
+      "step": 2022
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7206735225032223e-05,
+      "loss": 0.5799,
+      "step": 2023
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7192420296631835e-05,
+      "loss": 0.5478,
+      "step": 2024
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7178101135049177e-05,
+      "loss": 0.6101,
+      "step": 2025
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7163777748711937e-05,
+      "loss": 0.591,
+      "step": 2026
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.714945014605032e-05,
+      "loss": 0.5877,
+      "step": 2027
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7135118335497007e-05,
+      "loss": 0.5619,
+      "step": 2028
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.712078232548714e-05,
+      "loss": 0.5643,
+      "step": 2029
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7106442124458342e-05,
+      "loss": 0.6081,
+      "step": 2030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7092097740850712e-05,
+      "loss": 0.5701,
+      "step": 2031
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7077749183106806e-05,
+      "loss": 0.6003,
+      "step": 2032
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7063396459671626e-05,
+      "loss": 0.5668,
+      "step": 2033
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.704903957899264e-05,
+      "loss": 0.5805,
+      "step": 2034
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.703467854951975e-05,
+      "loss": 0.5978,
+      "step": 2035
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7020313379705312e-05,
+      "loss": 0.598,
+      "step": 2036
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7005944078004106e-05,
+      "loss": 0.562,
+      "step": 2037
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6991570652873357e-05,
+      "loss": 0.5876,
+      "step": 2038
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6977193112772714e-05,
+      "loss": 0.5611,
+      "step": 2039
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6962811466164236e-05,
+      "loss": 0.5969,
+      "step": 2040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6948425721512418e-05,
+      "loss": 0.6025,
+      "step": 2041
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.693403588728415e-05,
+      "loss": 0.5859,
+      "step": 2042
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6919641971948726e-05,
+      "loss": 0.5843,
+      "step": 2043
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.690524398397787e-05,
+      "loss": 0.5942,
+      "step": 2044
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6890841931845674e-05,
+      "loss": 0.6224,
+      "step": 2045
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6876435824028635e-05,
+      "loss": 0.5866,
+      "step": 2046
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.686202566900564e-05,
+      "loss": 0.5898,
+      "step": 2047
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6847611475257946e-05,
+      "loss": 0.5984,
+      "step": 2048
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.683319325126919e-05,
+      "loss": 0.5956,
+      "step": 2049
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.681877100552539e-05,
+      "loss": 0.5841,
+      "step": 2050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.680434474651493e-05,
+      "loss": 0.5807,
+      "step": 2051
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6789914482728546e-05,
+      "loss": 0.5565,
+      "step": 2052
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6775480222659337e-05,
+      "loss": 0.5441,
+      "step": 2053
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.676104197480276e-05,
+      "loss": 0.5967,
+      "step": 2054
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6746599747656605e-05,
+      "loss": 0.5813,
+      "step": 2055
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6732153549721018e-05,
+      "loss": 0.5954,
+      "step": 2056
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6717703389498467e-05,
+      "loss": 0.5753,
+      "step": 2057
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.670324927549377e-05,
+      "loss": 0.5949,
+      "step": 2058
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6688791216214064e-05,
+      "loss": 0.606,
+      "step": 2059
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6674329220168805e-05,
+      "loss": 0.6158,
+      "step": 2060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6659863295869768e-05,
+      "loss": 0.5872,
+      "step": 2061
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6645393451831042e-05,
+      "loss": 0.5721,
+      "step": 2062
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6630919696569014e-05,
+      "loss": 0.5952,
+      "step": 2063
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6616442038602384e-05,
+      "loss": 0.5548,
+      "step": 2064
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.660196048645215e-05,
+      "loss": 0.5767,
+      "step": 2065
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6587475048641596e-05,
+      "loss": 0.5998,
+      "step": 2066
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6572985733696287e-05,
+      "loss": 0.6271,
+      "step": 2067
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6558492550144092e-05,
+      "loss": 0.5887,
+      "step": 2068
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6543995506515126e-05,
+      "loss": 0.592,
+      "step": 2069
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6529494611341802e-05,
+      "loss": 0.5701,
+      "step": 2070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6514989873158783e-05,
+      "loss": 0.5979,
+      "step": 2071
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6500481300503006e-05,
+      "loss": 0.5877,
+      "step": 2072
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6485968901913658e-05,
+      "loss": 0.5844,
+      "step": 2073
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6471452685932177e-05,
+      "loss": 0.6212,
+      "step": 2074
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6456932661102258e-05,
+      "loss": 0.595,
+      "step": 2075
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6442408835969814e-05,
+      "loss": 0.5742,
+      "step": 2076
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6427881219083028e-05,
+      "loss": 0.5775,
+      "step": 2077
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6413349818992288e-05,
+      "loss": 0.5811,
+      "step": 2078
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6398814644250218e-05,
+      "loss": 0.5563,
+      "step": 2079
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6384275703411666e-05,
+      "loss": 0.5774,
+      "step": 2080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6369733005033693e-05,
+      "loss": 0.5878,
+      "step": 2081
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6355186557675572e-05,
+      "loss": 0.5822,
+      "step": 2082
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.634063636989878e-05,
+      "loss": 0.5834,
+      "step": 2083
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6326082450267013e-05,
+      "loss": 0.5617,
+      "step": 2084
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.631152480734613e-05,
+      "loss": 0.5825,
+      "step": 2085
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6296963449704206e-05,
+      "loss": 0.6149,
+      "step": 2086
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6282398385911503e-05,
+      "loss": 0.5822,
+      "step": 2087
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6267829624540454e-05,
+      "loss": 0.5526,
+      "step": 2088
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6253257174165675e-05,
+      "loss": 0.5984,
+      "step": 2089
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6238681043363948e-05,
+      "loss": 0.5716,
+      "step": 2090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.622410124071422e-05,
+      "loss": 0.6026,
+      "step": 2091
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6209517774797605e-05,
+      "loss": 0.5622,
+      "step": 2092
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.619493065419738e-05,
+      "loss": 0.5828,
+      "step": 2093
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.618033988749895e-05,
+      "loss": 0.578,
+      "step": 2094
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6165745483289887e-05,
+      "loss": 0.5978,
+      "step": 2095
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6151147450159898e-05,
+      "loss": 0.6084,
+      "step": 2096
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.613654579670082e-05,
+      "loss": 0.6038,
+      "step": 2097
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6121940531506626e-05,
+      "loss": 0.5652,
+      "step": 2098
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6107331663173416e-05,
+      "loss": 0.555,
+      "step": 2099
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.609271920029941e-05,
+      "loss": 0.5896,
+      "step": 2100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.607810315148494e-05,
+      "loss": 0.5946,
+      "step": 2101
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6063483525332452e-05,
+      "loss": 0.6044,
+      "step": 2102
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.604886033044649e-05,
+      "loss": 0.5982,
+      "step": 2103
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6034233575433712e-05,
+      "loss": 0.5807,
+      "step": 2104
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6019603268902866e-05,
+      "loss": 0.5971,
+      "step": 2105
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.600496941946478e-05,
+      "loss": 0.5275,
+      "step": 2106
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5990332035732388e-05,
+      "loss": 0.6258,
+      "step": 2107
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5975691126320678e-05,
+      "loss": 0.5654,
+      "step": 2108
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5961046699846736e-05,
+      "loss": 0.5963,
+      "step": 2109
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.59463987649297e-05,
+      "loss": 0.586,
+      "step": 2110
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5931747330190793e-05,
+      "loss": 0.5905,
+      "step": 2111
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5917092404253275e-05,
+      "loss": 0.5999,
+      "step": 2112
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.590243399574249e-05,
+      "loss": 0.5989,
+      "step": 2113
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.588777211328579e-05,
+      "loss": 0.5573,
+      "step": 2114
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.587310676551262e-05,
+      "loss": 0.5901,
+      "step": 2115
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5858437961054418e-05,
+      "loss": 0.6046,
+      "step": 2116
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5843765708544696e-05,
+      "loss": 0.5754,
+      "step": 2117
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.582909001661897e-05,
+      "loss": 0.6054,
+      "step": 2118
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5814410893914793e-05,
+      "loss": 0.5913,
+      "step": 2119
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5799728349071727e-05,
+      "loss": 0.5898,
+      "step": 2120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5785042390731358e-05,
+      "loss": 0.5837,
+      "step": 2121
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5770353027537276e-05,
+      "loss": 0.5747,
+      "step": 2122
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.575566026813507e-05,
+      "loss": 0.603,
+      "step": 2123
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5740964121172347e-05,
+      "loss": 0.5705,
+      "step": 2124
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.572626459529868e-05,
+      "loss": 0.5845,
+      "step": 2125
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5711561699165656e-05,
+      "loss": 0.6111,
+      "step": 2126
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5696855441426834e-05,
+      "loss": 0.606,
+      "step": 2127
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5682145830737747e-05,
+      "loss": 0.5752,
+      "step": 2128
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5667432875755904e-05,
+      "loss": 0.5728,
+      "step": 2129
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5652716585140795e-05,
+      "loss": 0.5775,
+      "step": 2130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5637996967553857e-05,
+      "loss": 0.588,
+      "step": 2131
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.562327403165849e-05,
+      "loss": 0.5775,
+      "step": 2132
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5608547786120056e-05,
+      "loss": 0.6073,
+      "step": 2133
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5593818239605856e-05,
+      "loss": 0.554,
+      "step": 2134
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5579085400785128e-05,
+      "loss": 0.5793,
+      "step": 2135
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5564349278329056e-05,
+      "loss": 0.5846,
+      "step": 2136
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5549609880910763e-05,
+      "loss": 0.5821,
+      "step": 2137
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5534867217205284e-05,
+      "loss": 0.6086,
+      "step": 2138
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.552012129588959e-05,
+      "loss": 0.5781,
+      "step": 2139
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5505372125642564e-05,
+      "loss": 0.557,
+      "step": 2140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5490619715144996e-05,
+      "loss": 0.5788,
+      "step": 2141
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.547586407307959e-05,
+      "loss": 0.5778,
+      "step": 2142
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5461105208130953e-05,
+      "loss": 0.5508,
+      "step": 2143
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5446343128985583e-05,
+      "loss": 0.6017,
+      "step": 2144
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5431577844331874e-05,
+      "loss": 0.5539,
+      "step": 2145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5416809362860107e-05,
+      "loss": 0.5993,
+      "step": 2146
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5402037693262443e-05,
+      "loss": 0.5819,
+      "step": 2147
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.538726284423291e-05,
+      "loss": 0.5808,
+      "step": 2148
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5372484824467428e-05,
+      "loss": 0.5536,
+      "step": 2149
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5357703642663766e-05,
+      "loss": 0.5981,
+      "step": 2150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5342919307521562e-05,
+      "loss": 0.5642,
+      "step": 2151
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5328131827742308e-05,
+      "loss": 0.5748,
+      "step": 2152
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5313341212029344e-05,
+      "loss": 0.5607,
+      "step": 2153
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5298547469087856e-05,
+      "loss": 0.5647,
+      "step": 2154
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5283750607624882e-05,
+      "loss": 0.5786,
+      "step": 2155
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5268950636349275e-05,
+      "loss": 0.5751,
+      "step": 2156
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.525414756397174e-05,
+      "loss": 0.5858,
+      "step": 2157
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5239341399204788e-05,
+      "loss": 0.5712,
+      "step": 2158
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.522453215076277e-05,
+      "loss": 0.5805,
+      "step": 2159
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5209719827361826e-05,
+      "loss": 0.611,
+      "step": 2160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5194904437719935e-05,
+      "loss": 0.5695,
+      "step": 2161
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.518008599055685e-05,
+      "loss": 0.5906,
+      "step": 2162
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5165264494594155e-05,
+      "loss": 0.5704,
+      "step": 2163
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5150439958555205e-05,
+      "loss": 0.6076,
+      "step": 2164
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5135612391165155e-05,
+      "loss": 0.5865,
+      "step": 2165
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5120781801150944e-05,
+      "loss": 0.5722,
+      "step": 2166
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5105948197241282e-05,
+      "loss": 0.5838,
+      "step": 2167
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5091111588166663e-05,
+      "loss": 0.5935,
+      "step": 2168
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.507627198265934e-05,
+      "loss": 0.6069,
+      "step": 2169
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5061429389453338e-05,
+      "loss": 0.6159,
+      "step": 2170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5046583817284437e-05,
+      "loss": 0.5852,
+      "step": 2171
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5031735274890176e-05,
+      "loss": 0.6068,
+      "step": 2172
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5016883771009824e-05,
+      "loss": 0.599,
+      "step": 2173
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5002029314384415e-05,
+      "loss": 0.5862,
+      "step": 2174
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.4987171913756708e-05,
+      "loss": 0.5599,
+      "step": 2175
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.4972311577871198e-05,
+      "loss": 0.6175,
+      "step": 2176
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.495744831547411e-05,
+      "loss": 0.5931,
+      "step": 2177
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4942582135313393e-05,
+      "loss": 0.6,
+      "step": 2178
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4927713046138705e-05,
+      "loss": 0.6055,
+      "step": 2179
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.491284105670142e-05,
+      "loss": 0.5748,
+      "step": 2180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.489796617575462e-05,
+      "loss": 0.58,
+      "step": 2181
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.488308841205309e-05,
+      "loss": 0.6417,
+      "step": 2182
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4868207774353305e-05,
+      "loss": 0.5836,
+      "step": 2183
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.485332427141344e-05,
+      "loss": 0.5574,
+      "step": 2184
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4838437911993355e-05,
+      "loss": 0.573,
+      "step": 2185
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4823548704854593e-05,
+      "loss": 0.5775,
+      "step": 2186
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4808656658760364e-05,
+      "loss": 0.6017,
+      "step": 2187
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4793761782475548e-05,
+      "loss": 0.5853,
+      "step": 2188
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4778864084766713e-05,
+      "loss": 0.5784,
+      "step": 2189
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4763963574402055e-05,
+      "loss": 0.5873,
+      "step": 2190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.474906026015145e-05,
+      "loss": 0.5693,
+      "step": 2191
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.473415415078642e-05,
+      "loss": 0.606,
+      "step": 2192
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4719245255080118e-05,
+      "loss": 0.5574,
+      "step": 2193
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.470433358180736e-05,
+      "loss": 0.5826,
+      "step": 2194
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4689419139744573e-05,
+      "loss": 0.5621,
+      "step": 2195
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4674501937669825e-05,
+      "loss": 0.5752,
+      "step": 2196
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4659581984362814e-05,
+      "loss": 0.6106,
+      "step": 2197
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4644659288604853e-05,
+      "loss": 0.5735,
+      "step": 2198
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4629733859178867e-05,
+      "loss": 0.5944,
+      "step": 2199
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.461480570486938e-05,
+      "loss": 0.573,
+      "step": 2200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4599874834462538e-05,
+      "loss": 0.6126,
+      "step": 2201
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.458494125674608e-05,
+      "loss": 0.5453,
+      "step": 2202
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.457000498050933e-05,
+      "loss": 0.5974,
+      "step": 2203
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.455506601454321e-05,
+      "loss": 0.5872,
+      "step": 2204
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4540124367640222e-05,
+      "loss": 0.5966,
+      "step": 2205
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4525180048594452e-05,
+      "loss": 0.5852,
+      "step": 2206
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4510233066201532e-05,
+      "loss": 0.5754,
+      "step": 2207
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.44952834292587e-05,
+      "loss": 0.5846,
+      "step": 2208
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.448033114656473e-05,
+      "loss": 0.5889,
+      "step": 2209
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4465376226919963e-05,
+      "loss": 0.583,
+      "step": 2210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.445041867912629e-05,
+      "loss": 0.5606,
+      "step": 2211
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4435458511987153e-05,
+      "loss": 0.5805,
+      "step": 2212
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4420495734307527e-05,
+      "loss": 0.5756,
+      "step": 2213
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.440553035489393e-05,
+      "loss": 0.6036,
+      "step": 2214
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4390562382554407e-05,
+      "loss": 0.6073,
+      "step": 2215
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.437559182609853e-05,
+      "loss": 0.5781,
+      "step": 2216
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4360618694337397e-05,
+      "loss": 0.5708,
+      "step": 2217
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.434564299608363e-05,
+      "loss": 0.598,
+      "step": 2218
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4330664740151324e-05,
+      "loss": 0.5755,
+      "step": 2219
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4315683935356127e-05,
+      "loss": 0.5605,
+      "step": 2220
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.430070059051515e-05,
+      "loss": 0.5866,
+      "step": 2221
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4285714714447024e-05,
+      "loss": 0.5544,
+      "step": 2222
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4270726315971857e-05,
+      "loss": 0.5801,
+      "step": 2223
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4255735403911243e-05,
+      "loss": 0.5813,
+      "step": 2224
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4240741987088265e-05,
+      "loss": 0.5627,
+      "step": 2225
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4225746074327456e-05,
+      "loss": 0.5888,
+      "step": 2226
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.421074767445485e-05,
+      "loss": 0.5795,
+      "step": 2227
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4195746796297922e-05,
+      "loss": 0.5603,
+      "step": 2228
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.418074344868561e-05,
+      "loss": 0.5804,
+      "step": 2229
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4165737640448312e-05,
+      "loss": 0.5668,
+      "step": 2230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4150729380417874e-05,
+      "loss": 0.5949,
+      "step": 2231
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4135718677427572e-05,
+      "loss": 0.5682,
+      "step": 2232
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4120705540312134e-05,
+      "loss": 0.5989,
+      "step": 2233
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4105689977907722e-05,
+      "loss": 0.5627,
+      "step": 2234
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4090671999051906e-05,
+      "loss": 0.5924,
+      "step": 2235
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.40756516125837e-05,
+      "loss": 0.5847,
+      "step": 2236
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4060628827343525e-05,
+      "loss": 0.5673,
+      "step": 2237
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.4045603652173218e-05,
+      "loss": 0.6082,
+      "step": 2238
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.4030576095916013e-05,
+      "loss": 0.5632,
+      "step": 2239
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.401554616741656e-05,
+      "loss": 0.5501,
+      "step": 2240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.4000513875520892e-05,
+      "loss": 0.557,
+      "step": 2241
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.398547922907644e-05,
+      "loss": 0.5759,
+      "step": 2242
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3970442236932017e-05,
+      "loss": 0.5558,
+      "step": 2243
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3955402907937825e-05,
+      "loss": 0.6041,
+      "step": 2244
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3940361250945436e-05,
+      "loss": 0.5716,
+      "step": 2245
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3925317274807784e-05,
+      "loss": 0.5789,
+      "step": 2246
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.391027098837918e-05,
+      "loss": 0.6172,
+      "step": 2247
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3895222400515282e-05,
+      "loss": 0.5705,
+      "step": 2248
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3880171520073123e-05,
+      "loss": 0.5846,
+      "step": 2249
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3865118355911066e-05,
+      "loss": 0.5612,
+      "step": 2250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3850062916888832e-05,
+      "loss": 0.5979,
+      "step": 2251
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3835005211867465e-05,
+      "loss": 0.5783,
+      "step": 2252
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3819945249709357e-05,
+      "loss": 0.5763,
+      "step": 2253
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.380488303927822e-05,
+      "loss": 0.5627,
+      "step": 2254
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3789818589439094e-05,
+      "loss": 0.5645,
+      "step": 2255
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3774751909058335e-05,
+      "loss": 0.5903,
+      "step": 2256
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3759683007003616e-05,
+      "loss": 0.5855,
+      "step": 2257
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3744611892143913e-05,
+      "loss": 0.5672,
+      "step": 2258
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3729538573349498e-05,
+      "loss": 0.5898,
+      "step": 2259
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3714463059491956e-05,
+      "loss": 0.5741,
+      "step": 2260
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3699385359444148e-05,
+      "loss": 0.5735,
+      "step": 2261
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3684305482080233e-05,
+      "loss": 0.5821,
+      "step": 2262
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.366922343627565e-05,
+      "loss": 0.603,
+      "step": 2263
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.365413923090711e-05,
+      "loss": 0.6049,
+      "step": 2264
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3639052874852597e-05,
+      "loss": 0.6079,
+      "step": 2265
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3623964376991354e-05,
+      "loss": 0.6124,
+      "step": 2266
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.36088737462039e-05,
+      "loss": 0.5829,
+      "step": 2267
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3593780991371996e-05,
+      "loss": 0.5629,
+      "step": 2268
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.357868612137866e-05,
+      "loss": 0.5871,
+      "step": 2269
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.356358914510815e-05,
+      "loss": 0.5849,
+      "step": 2270
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3548490071445968e-05,
+      "loss": 0.5891,
+      "step": 2271
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3533388909278855e-05,
+      "loss": 0.5903,
+      "step": 2272
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3518285667494763e-05,
+      "loss": 0.5648,
+      "step": 2273
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3503180354982883e-05,
+      "loss": 0.5546,
+      "step": 2274
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3488072980633624e-05,
+      "loss": 0.5864,
+      "step": 2275
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.5508,
+      "step": 2276
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3457852081990665e-05,
+      "loss": 0.591,
+      "step": 2277
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3442738575483826e-05,
+      "loss": 0.5474,
+      "step": 2278
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.342762304271331e-05,
+      "loss": 0.5727,
+      "step": 2279
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3412505492575544e-05,
+      "loss": 0.5721,
+      "step": 2280
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3397385933968143e-05,
+      "loss": 0.567,
+      "step": 2281
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3382264375789897e-05,
+      "loss": 0.5389,
+      "step": 2282
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3367140826940768e-05,
+      "loss": 0.5871,
+      "step": 2283
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3352015296321907e-05,
+      "loss": 0.5888,
+      "step": 2284
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3336887792835613e-05,
+      "loss": 0.5999,
+      "step": 2285
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.332175832538535e-05,
+      "loss": 0.5843,
+      "step": 2286
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3306626902875745e-05,
+      "loss": 0.5556,
+      "step": 2287
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3291493534212573e-05,
+      "loss": 0.5756,
+      "step": 2288
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3276358228302757e-05,
+      "loss": 0.6068,
+      "step": 2289
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.326122099405435e-05,
+      "loss": 0.5743,
+      "step": 2290
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3246081840376548e-05,
+      "loss": 0.5829,
+      "step": 2291
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3230940776179672e-05,
+      "loss": 0.586,
+      "step": 2292
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3215797810375178e-05,
+      "loss": 0.5979,
+      "step": 2293
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.320065295187563e-05,
+      "loss": 0.5995,
+      "step": 2294
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.318550620959471e-05,
+      "loss": 0.5809,
+      "step": 2295
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3170357592447214e-05,
+      "loss": 0.5731,
+      "step": 2296
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.315520710934903e-05,
+      "loss": 0.598,
+      "step": 2297
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3140054769217145e-05,
+      "loss": 0.5975,
+      "step": 2298
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.312490058096966e-05,
+      "loss": 0.552,
+      "step": 2299
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3109744553525735e-05,
+      "loss": 0.5691,
+      "step": 2300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3094586695805625e-05,
+      "loss": 0.5819,
+      "step": 2301
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.307942701673067e-05,
+      "loss": 0.5959,
+      "step": 2302
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3064265525223277e-05,
+      "loss": 0.5685,
+      "step": 2303
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.304910223020691e-05,
+      "loss": 0.5821,
+      "step": 2304
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3033937140606105e-05,
+      "loss": 0.581,
+      "step": 2305
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3018770265346452e-05,
+      "loss": 0.5729,
+      "step": 2306
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3003601613354586e-05,
+      "loss": 0.5836,
+      "step": 2307
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2988431193558196e-05,
+      "loss": 0.5873,
+      "step": 2308
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2973259014886008e-05,
+      "loss": 0.5797,
+      "step": 2309
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.295808508626779e-05,
+      "loss": 0.5634,
+      "step": 2310
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2942909416634326e-05,
+      "loss": 0.6084,
+      "step": 2311
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.292773201491743e-05,
+      "loss": 0.5628,
+      "step": 2312
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.291255289004995e-05,
+      "loss": 0.5814,
+      "step": 2313
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2897372050965717e-05,
+      "loss": 0.5675,
+      "step": 2314
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2882189506599604e-05,
+      "loss": 0.595,
+      "step": 2315
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.286700526588747e-05,
+      "loss": 0.5739,
+      "step": 2316
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.285181933776618e-05,
+      "loss": 0.5589,
+      "step": 2317
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2836631731173577e-05,
+      "loss": 0.5736,
+      "step": 2318
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.282144245504851e-05,
+      "loss": 0.5776,
+      "step": 2319
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2806251518330802e-05,
+      "loss": 0.5723,
+      "step": 2320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2791058929961255e-05,
+      "loss": 0.5767,
+      "step": 2321
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2775864698881645e-05,
+      "loss": 0.5374,
+      "step": 2322
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.276066883403471e-05,
+      "loss": 0.5947,
+      "step": 2323
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2745471344364157e-05,
+      "loss": 0.6024,
+      "step": 2324
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2730272238814636e-05,
+      "loss": 0.5685,
+      "step": 2325
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.271507152633176e-05,
+      "loss": 0.5928,
+      "step": 2326
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2699869215862086e-05,
+      "loss": 0.5834,
+      "step": 2327
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2684665316353112e-05,
+      "loss": 0.6191,
+      "step": 2328
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2669459836753266e-05,
+      "loss": 0.5913,
+      "step": 2329
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.265425278601191e-05,
+      "loss": 0.5557,
+      "step": 2330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2639044173079333e-05,
+      "loss": 0.6081,
+      "step": 2331
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2623834006906732e-05,
+      "loss": 0.5961,
+      "step": 2332
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.260862229644623e-05,
+      "loss": 0.5862,
+      "step": 2333
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.259340905065086e-05,
+      "loss": 0.5917,
+      "step": 2334
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.257819427847455e-05,
+      "loss": 0.5715,
+      "step": 2335
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2562977988872136e-05,
+      "loss": 0.5869,
+      "step": 2336
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2547760190799333e-05,
+      "loss": 0.608,
+      "step": 2337
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.253254089321276e-05,
+      "loss": 0.5561,
+      "step": 2338
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.25173201050699e-05,
+      "loss": 0.5863,
+      "step": 2339
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.250209783532914e-05,
+      "loss": 0.5555,
+      "step": 2340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2486874092949708e-05,
+      "loss": 0.5807,
+      "step": 2341
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.247164888689172e-05,
+      "loss": 0.5843,
+      "step": 2342
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.245642222611615e-05,
+      "loss": 0.562,
+      "step": 2343
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2441194119584824e-05,
+      "loss": 0.5898,
+      "step": 2344
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.242596457626042e-05,
+      "loss": 0.6003,
+      "step": 2345
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2410733605106462e-05,
+      "loss": 0.5647,
+      "step": 2346
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2395501215087316e-05,
+      "loss": 0.549,
+      "step": 2347
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2380267415168174e-05,
+      "loss": 0.5849,
+      "step": 2348
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2365032214315085e-05,
+      "loss": 0.5963,
+      "step": 2349
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.234979562149488e-05,
+      "loss": 0.6209,
+      "step": 2350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2334557645675242e-05,
+      "loss": 0.5846,
+      "step": 2351
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.231931829582466e-05,
+      "loss": 0.5942,
+      "step": 2352
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2304077580912423e-05,
+      "loss": 0.5926,
+      "step": 2353
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.228883550990864e-05,
+      "loss": 0.5836,
+      "step": 2354
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2273592091784203e-05,
+      "loss": 0.5869,
+      "step": 2355
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.225834733551081e-05,
+      "loss": 0.5744,
+      "step": 2356
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2243101250060918e-05,
+      "loss": 0.5988,
+      "step": 2357
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2227853844407812e-05,
+      "loss": 0.5772,
+      "step": 2358
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.221260512752551e-05,
+      "loss": 0.5907,
+      "step": 2359
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2197355108388835e-05,
+      "loss": 0.5633,
+      "step": 2360
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2182103795973354e-05,
+      "loss": 0.5998,
+      "step": 2361
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.216685119925541e-05,
+      "loss": 0.5689,
+      "step": 2362
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2151597327212092e-05,
+      "loss": 0.5618,
+      "step": 2363
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2136342188821237e-05,
+      "loss": 0.5911,
+      "step": 2364
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2121085793061444e-05,
+      "loss": 0.6065,
+      "step": 2365
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2105828148912032e-05,
+      "loss": 0.5555,
+      "step": 2366
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.209056926535307e-05,
+      "loss": 0.5783,
+      "step": 2367
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.207530915136535e-05,
+      "loss": 0.5416,
+      "step": 2368
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2060047815930393e-05,
+      "loss": 0.5829,
+      "step": 2369
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2044785268030432e-05,
+      "loss": 0.6009,
+      "step": 2370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2029521516648407e-05,
+      "loss": 0.5619,
+      "step": 2371
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2014256570767985e-05,
+      "loss": 0.5878,
+      "step": 2372
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.1998990439373523e-05,
+      "loss": 0.566,
+      "step": 2373
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.1983723131450088e-05,
+      "loss": 0.5868,
+      "step": 2374
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.196845465598342e-05,
+      "loss": 0.6197,
+      "step": 2375
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1953185021959964e-05,
+      "loss": 0.6003,
+      "step": 2376
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.193791423836684e-05,
+      "loss": 0.5759,
+      "step": 2377
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1922642314191833e-05,
+      "loss": 0.5641,
+      "step": 2378
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1907369258423428e-05,
+      "loss": 0.5916,
+      "step": 2379
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1892095080050746e-05,
+      "loss": 0.5742,
+      "step": 2380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1876819788063586e-05,
+      "loss": 0.5899,
+      "step": 2381
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1861543391452402e-05,
+      "loss": 0.546,
+      "step": 2382
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1846265899208295e-05,
+      "loss": 0.5901,
+      "step": 2383
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1830987320323e-05,
+      "loss": 0.5993,
+      "step": 2384
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1815707663788907e-05,
+      "loss": 0.5463,
+      "step": 2385
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.180042693859903e-05,
+      "loss": 0.5761,
+      "step": 2386
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1785145153747033e-05,
+      "loss": 0.5867,
+      "step": 2387
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.176986231822717e-05,
+      "loss": 0.6044,
+      "step": 2388
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.175457844103434e-05,
+      "loss": 0.5748,
+      "step": 2389
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1739293531164046e-05,
+      "loss": 0.5727,
+      "step": 2390
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.172400759761239e-05,
+      "loss": 0.5749,
+      "step": 2391
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1708720649376097e-05,
+      "loss": 0.5847,
+      "step": 2392
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1693432695452467e-05,
+      "loss": 0.5704,
+      "step": 2393
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1678143744839407e-05,
+      "loss": 0.5834,
+      "step": 2394
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.166285380653541e-05,
+      "loss": 0.5937,
+      "step": 2395
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.164756288953955e-05,
+      "loss": 0.5946,
+      "step": 2396
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1632271002851452e-05,
+      "loss": 0.5599,
+      "step": 2397
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1616978155471352e-05,
+      "loss": 0.5885,
+      "step": 2398
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1601684356400025e-05,
+      "loss": 0.5793,
+      "step": 2399
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.158638961463882e-05,
+      "loss": 0.5976,
+      "step": 2400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1571093939189625e-05,
+      "loss": 0.5848,
+      "step": 2401
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1555797339054898e-05,
+      "loss": 0.5892,
+      "step": 2402
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.154049982323763e-05,
+      "loss": 0.5956,
+      "step": 2403
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1525201400741338e-05,
+      "loss": 0.5737,
+      "step": 2404
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1509902080570093e-05,
+      "loss": 0.6266,
+      "step": 2405
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.149460187172849e-05,
+      "loss": 0.5687,
+      "step": 2406
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1479300783221642e-05,
+      "loss": 0.5467,
+      "step": 2407
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1463998824055177e-05,
+      "loss": 0.6043,
+      "step": 2408
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1448696003235252e-05,
+      "loss": 0.5841,
+      "step": 2409
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1433392329768505e-05,
+      "loss": 0.5839,
+      "step": 2410
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1418087812662104e-05,
+      "loss": 0.5813,
+      "step": 2411
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1402782460923693e-05,
+      "loss": 0.589,
+      "step": 2412
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.138747628356141e-05,
+      "loss": 0.559,
+      "step": 2413
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1372169289583892e-05,
+      "loss": 0.5778,
+      "step": 2414
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1356861488000247e-05,
+      "loss": 0.5877,
+      "step": 2415
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1341552887820048e-05,
+      "loss": 0.6333,
+      "step": 2416
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1326243498053357e-05,
+      "loss": 0.5689,
+      "step": 2417
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1310933327710698e-05,
+      "loss": 0.5465,
+      "step": 2418
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1295622385803036e-05,
+      "loss": 0.6026,
+      "step": 2419
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1280310681341815e-05,
+      "loss": 0.5565,
+      "step": 2420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1264998223338907e-05,
+      "loss": 0.6255,
+      "step": 2421
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1249685020806653e-05,
+      "loss": 0.5375,
+      "step": 2422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.123437108275779e-05,
+      "loss": 0.577,
+      "step": 2423
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1219056418205534e-05,
+      "loss": 0.5521,
+      "step": 2424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1203741036163493e-05,
+      "loss": 0.577,
+      "step": 2425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.118842494564573e-05,
+      "loss": 0.5565,
+      "step": 2426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1173108155666693e-05,
+      "loss": 0.577,
+      "step": 2427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1157790675241267e-05,
+      "loss": 0.5856,
+      "step": 2428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1142472513384728e-05,
+      "loss": 0.5861,
+      "step": 2429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.112715367911275e-05,
+      "loss": 0.5769,
+      "step": 2430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1111834181441423e-05,
+      "loss": 0.5804,
+      "step": 2431
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1096514029387204e-05,
+      "loss": 0.5782,
+      "step": 2432
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1081193231966954e-05,
+      "loss": 0.5684,
+      "step": 2433
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1065871798197898e-05,
+      "loss": 0.5666,
+      "step": 2434
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1050549737097654e-05,
+      "loss": 0.5707,
+      "step": 2435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1035227057684193e-05,
+      "loss": 0.5408,
+      "step": 2436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1019903768975852e-05,
+      "loss": 0.5783,
+      "step": 2437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1004579879991333e-05,
+      "loss": 0.5769,
+      "step": 2438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0989255399749687e-05,
+      "loss": 0.5802,
+      "step": 2439
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.097393033727032e-05,
+      "loss": 0.5731,
+      "step": 2440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0958604701572966e-05,
+      "loss": 0.5841,
+      "step": 2441
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0943278501677716e-05,
+      "loss": 0.5911,
+      "step": 2442
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0927951746604977e-05,
+      "loss": 0.5829,
+      "step": 2443
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0912624445375483e-05,
+      "loss": 0.5967,
+      "step": 2444
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.08972966070103e-05,
+      "loss": 0.5634,
+      "step": 2445
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0881968240530804e-05,
+      "loss": 0.5951,
+      "step": 2446
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0866639354958692e-05,
+      "loss": 0.5677,
+      "step": 2447
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0851309959315942e-05,
+      "loss": 0.5605,
+      "step": 2448
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0835980062624856e-05,
+      "loss": 0.5948,
+      "step": 2449
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0820649673908018e-05,
+      "loss": 0.56,
+      "step": 2450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0805318802188307e-05,
+      "loss": 0.5705,
+      "step": 2451
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0789987456488885e-05,
+      "loss": 0.5931,
+      "step": 2452
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.07746556458332e-05,
+      "loss": 0.5621,
+      "step": 2453
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0759323379244954e-05,
+      "loss": 0.6003,
+      "step": 2454
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0743990665748143e-05,
+      "loss": 0.5765,
+      "step": 2455
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0728657514367008e-05,
+      "loss": 0.5724,
+      "step": 2456
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0713323934126047e-05,
+      "loss": 0.5561,
+      "step": 2457
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0697989934050025e-05,
+      "loss": 0.5764,
+      "step": 2458
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0682655523163938e-05,
+      "loss": 0.5911,
+      "step": 2459
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0667320710493034e-05,
+      "loss": 0.6017,
+      "step": 2460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.06519855050628e-05,
+      "loss": 0.5645,
+      "step": 2461
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.063664991589895e-05,
+      "loss": 0.5624,
+      "step": 2462
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.062131395202741e-05,
+      "loss": 0.5524,
+      "step": 2463
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.060597762247434e-05,
+      "loss": 0.588,
+      "step": 2464
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0590640936266132e-05,
+      "loss": 0.5822,
+      "step": 2465
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.057530390242935e-05,
+      "loss": 0.5666,
+      "step": 2466
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.05599665299908e-05,
+      "loss": 0.5357,
+      "step": 2467
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0544628827977453e-05,
+      "loss": 0.5756,
+      "step": 2468
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.052929080541651e-05,
+      "loss": 0.587,
+      "step": 2469
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0513952471335328e-05,
+      "loss": 0.5616,
+      "step": 2470
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0498613834761462e-05,
+      "loss": 0.5796,
+      "step": 2471
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0483274904722647e-05,
+      "loss": 0.5857,
+      "step": 2472
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0467935690246793e-05,
+      "loss": 0.5876,
+      "step": 2473
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0452596200361968e-05,
+      "loss": 0.5867,
+      "step": 2474
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0437256444096404e-05,
+      "loss": 0.5655,
+      "step": 2475
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0421916430478497e-05,
+      "loss": 0.5802,
+      "step": 2476
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0406576168536785e-05,
+      "loss": 0.5695,
+      "step": 2477
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.039123566729996e-05,
+      "loss": 0.5856,
+      "step": 2478
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.037589493579685e-05,
+      "loss": 0.5838,
+      "step": 2479
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.036055398305642e-05,
+      "loss": 0.605,
+      "step": 2480
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0345212818107774e-05,
+      "loss": 0.5864,
+      "step": 2481
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0329871449980118e-05,
+      "loss": 0.5627,
+      "step": 2482
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.03145298877028e-05,
+      "loss": 0.5963,
+      "step": 2483
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0299188140305276e-05,
+      "loss": 0.563,
+      "step": 2484
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0283846216817103e-05,
+      "loss": 0.5978,
+      "step": 2485
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0268504126267952e-05,
+      "loss": 0.5531,
+      "step": 2486
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.025316187768759e-05,
+      "loss": 0.5902,
+      "step": 2487
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0237819480105867e-05,
+      "loss": 0.5799,
+      "step": 2488
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0222476942552737e-05,
+      "loss": 0.5747,
+      "step": 2489
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0207134274058223e-05,
+      "loss": 0.5889,
+      "step": 2490
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0191791483652432e-05,
+      "loss": 0.5988,
+      "step": 2491
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0176448580365546e-05,
+      "loss": 0.6058,
+      "step": 2492
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0161105573227798e-05,
+      "loss": 0.5712,
+      "step": 2493
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.014576247126951e-05,
+      "loss": 0.5778,
+      "step": 2494
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0130419283521025e-05,
+      "loss": 0.5696,
+      "step": 2495
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0115076019012757e-05,
+      "loss": 0.5616,
+      "step": 2496
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0099732686775165e-05,
+      "loss": 0.6113,
+      "step": 2497
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0084389295838746e-05,
+      "loss": 0.583,
+      "step": 2498
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.006904585523403e-05,
+      "loss": 0.6062,
+      "step": 2499
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.005370237399157e-05,
+      "loss": 0.5313,
+      "step": 2500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0038358861141968e-05,
+      "loss": 0.5857,
+      "step": 2501
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0023015325715806e-05,
+      "loss": 0.5714,
+      "step": 2502
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0007671776743706e-05,
+      "loss": 0.5619,
+      "step": 2503
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9992328223256297e-05,
+      "loss": 0.5572,
+      "step": 2504
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9976984674284198e-05,
+      "loss": 0.5874,
+      "step": 2505
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.996164113885804e-05,
+      "loss": 0.5658,
+      "step": 2506
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9946297626008432e-05,
+      "loss": 0.5889,
+      "step": 2507
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9930954144765978e-05,
+      "loss": 0.5991,
+      "step": 2508
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.991561070416126e-05,
+      "loss": 0.6314,
+      "step": 2509
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9900267313224842e-05,
+      "loss": 0.579,
+      "step": 2510
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9884923980987253e-05,
+      "loss": 0.5705,
+      "step": 2511
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9869580716478986e-05,
+      "loss": 0.5791,
+      "step": 2512
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9854237528730497e-05,
+      "loss": 0.5795,
+      "step": 2513
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9838894426772205e-05,
+      "loss": 0.5689,
+      "step": 2514
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.982355141963446e-05,
+      "loss": 0.5369,
+      "step": 2515
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.980820851634757e-05,
+      "loss": 0.5971,
+      "step": 2516
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9792865725941784e-05,
+      "loss": 0.5526,
+      "step": 2517
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.977752305744727e-05,
+      "loss": 0.5815,
+      "step": 2518
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.976218051989414e-05,
+      "loss": 0.5707,
+      "step": 2519
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9746838122312418e-05,
+      "loss": 0.5846,
+      "step": 2520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9731495873732055e-05,
+      "loss": 0.5791,
+      "step": 2521
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9716153783182907e-05,
+      "loss": 0.5753,
+      "step": 2522
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9700811859694734e-05,
+      "loss": 0.59,
+      "step": 2523
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9685470112297203e-05,
+      "loss": 0.5946,
+      "step": 2524
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9670128550019885e-05,
+      "loss": 0.5835,
+      "step": 2525
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9654787181892233e-05,
+      "loss": 0.5795,
+      "step": 2526
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9639446016943586e-05,
+      "loss": 0.5641,
+      "step": 2527
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9624105064203157e-05,
+      "loss": 0.5934,
+      "step": 2528
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9608764332700046e-05,
+      "loss": 0.5588,
+      "step": 2529
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9593423831463222e-05,
+      "loss": 0.6063,
+      "step": 2530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.957808356952151e-05,
+      "loss": 0.5772,
+      "step": 2531
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9562743555903603e-05,
+      "loss": 0.5622,
+      "step": 2532
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.954740379963804e-05,
+      "loss": 0.5825,
+      "step": 2533
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9532064309753217e-05,
+      "loss": 0.5832,
+      "step": 2534
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.951672509527736e-05,
+      "loss": 0.589,
+      "step": 2535
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9501386165238548e-05,
+      "loss": 0.5904,
+      "step": 2536
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9486047528664685e-05,
+      "loss": 0.5909,
+      "step": 2537
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.94707091945835e-05,
+      "loss": 0.5837,
+      "step": 2538
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9455371172022553e-05,
+      "loss": 0.5922,
+      "step": 2539
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944003347000921e-05,
+      "loss": 0.6099,
+      "step": 2540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.942469609757066e-05,
+      "loss": 0.5663,
+      "step": 2541
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.940935906373388e-05,
+      "loss": 0.5894,
+      "step": 2542
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.939402237752567e-05,
+      "loss": 0.6027,
+      "step": 2543
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9378686047972605e-05,
+      "loss": 0.5904,
+      "step": 2544
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9363350084101063e-05,
+      "loss": 0.5687,
+      "step": 2545
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9348014494937204e-05,
+      "loss": 0.5572,
+      "step": 2546
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.933267928950697e-05,
+      "loss": 0.5977,
+      "step": 2547
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9317344476836072e-05,
+      "loss": 0.603,
+      "step": 2548
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.930201006594999e-05,
+      "loss": 0.5842,
+      "step": 2549
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9286676065873963e-05,
+      "loss": 0.5614,
+      "step": 2550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9271342485633006e-05,
+      "loss": 0.6044,
+      "step": 2551
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9256009334251864e-05,
+      "loss": 0.5404,
+      "step": 2552
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9240676620755053e-05,
+      "loss": 0.5683,
+      "step": 2553
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9225344354166814e-05,
+      "loss": 0.5731,
+      "step": 2554
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9210012543511122e-05,
+      "loss": 0.5981,
+      "step": 2555
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9194681197811703e-05,
+      "loss": 0.5765,
+      "step": 2556
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9179350326091996e-05,
+      "loss": 0.5528,
+      "step": 2557
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9164019937375157e-05,
+      "loss": 0.5582,
+      "step": 2558
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9148690040684064e-05,
+      "loss": 0.5711,
+      "step": 2559
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9133360645041318e-05,
+      "loss": 0.5588,
+      "step": 2560
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9118031759469202e-05,
+      "loss": 0.5947,
+      "step": 2561
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.910270339298971e-05,
+      "loss": 0.6065,
+      "step": 2562
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9087375554624527e-05,
+      "loss": 0.5802,
+      "step": 2563
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9072048253395033e-05,
+      "loss": 0.5648,
+      "step": 2564
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9056721498322294e-05,
+      "loss": 0.5685,
+      "step": 2565
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9041395298427038e-05,
+      "loss": 0.5998,
+      "step": 2566
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9026069662729687e-05,
+      "loss": 0.567,
+      "step": 2567
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9010744600250316e-05,
+      "loss": 0.558,
+      "step": 2568
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.899542012000867e-05,
+      "loss": 0.5831,
+      "step": 2569
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.898009623102415e-05,
+      "loss": 0.5632,
+      "step": 2570
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.896477294231581e-05,
+      "loss": 0.5958,
+      "step": 2571
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.894945026290235e-05,
+      "loss": 0.5817,
+      "step": 2572
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8934128201802102e-05,
+      "loss": 0.5724,
+      "step": 2573
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.891880676803305e-05,
+      "loss": 0.5782,
+      "step": 2574
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.89034859706128e-05,
+      "loss": 0.592,
+      "step": 2575
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.888816581855858e-05,
+      "loss": 0.5962,
+      "step": 2576
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.887284632088725e-05,
+      "loss": 0.5703,
+      "step": 2577
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.885752748661528e-05,
+      "loss": 0.5824,
+      "step": 2578
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8842209324758736e-05,
+      "loss": 0.5838,
+      "step": 2579
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8826891844333307e-05,
+      "loss": 0.5682,
+      "step": 2580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8811575054354273e-05,
+      "loss": 0.5872,
+      "step": 2581
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8796258963836504e-05,
+      "loss": 0.5688,
+      "step": 2582
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.878094358179447e-05,
+      "loss": 0.5729,
+      "step": 2583
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8765628917242213e-05,
+      "loss": 0.574,
+      "step": 2584
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8750314979193357e-05,
+      "loss": 0.5536,
+      "step": 2585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8735001776661093e-05,
+      "loss": 0.5712,
+      "step": 2586
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.871968931865819e-05,
+      "loss": 0.593,
+      "step": 2587
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8704377614196963e-05,
+      "loss": 0.5315,
+      "step": 2588
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8689066672289306e-05,
+      "loss": 0.5584,
+      "step": 2589
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.867375650194664e-05,
+      "loss": 0.5754,
+      "step": 2590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8658447112179952e-05,
+      "loss": 0.5995,
+      "step": 2591
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.864313851199976e-05,
+      "loss": 0.5831,
+      "step": 2592
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8627830710416108e-05,
+      "loss": 0.5834,
+      "step": 2593
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.861252371643859e-05,
+      "loss": 0.5671,
+      "step": 2594
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.859721753907631e-05,
+      "loss": 0.5691,
+      "step": 2595
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.85819121873379e-05,
+      "loss": 0.5433,
+      "step": 2596
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8566607670231494e-05,
+      "loss": 0.5414,
+      "step": 2597
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8551303996764755e-05,
+      "loss": 0.6079,
+      "step": 2598
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8536001175944826e-05,
+      "loss": 0.5915,
+      "step": 2599
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8520699216778365e-05,
+      "loss": 0.5711,
+      "step": 2600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8505398128271517e-05,
+      "loss": 0.6042,
+      "step": 2601
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8490097919429913e-05,
+      "loss": 0.5775,
+      "step": 2602
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.847479859925867e-05,
+      "loss": 0.5523,
+      "step": 2603
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8459500176762378e-05,
+      "loss": 0.5585,
+      "step": 2604
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8444202660945105e-05,
+      "loss": 0.5846,
+      "step": 2605
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8428906060810378e-05,
+      "loss": 0.5552,
+      "step": 2606
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8413610385361183e-05,
+      "loss": 0.5711,
+      "step": 2607
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8398315643599978e-05,
+      "loss": 0.5527,
+      "step": 2608
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.838302184452865e-05,
+      "loss": 0.5934,
+      "step": 2609
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.836772899714855e-05,
+      "loss": 0.5987,
+      "step": 2610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.835243711046046e-05,
+      "loss": 0.5528,
+      "step": 2611
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8337146193464595e-05,
+      "loss": 0.5814,
+      "step": 2612
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8321856255160596e-05,
+      "loss": 0.5394,
+      "step": 2613
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8306567304547537e-05,
+      "loss": 0.5754,
+      "step": 2614
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.829127935062391e-05,
+      "loss": 0.5727,
+      "step": 2615
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8275992402387613e-05,
+      "loss": 0.5723,
+      "step": 2616
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8260706468835958e-05,
+      "loss": 0.5876,
+      "step": 2617
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8245421558965662e-05,
+      "loss": 0.5676,
+      "step": 2618
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8230137681772836e-05,
+      "loss": 0.5871,
+      "step": 2619
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8214854846252974e-05,
+      "loss": 0.5747,
+      "step": 2620
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.819957306140097e-05,
+      "loss": 0.5611,
+      "step": 2621
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.81842923362111e-05,
+      "loss": 0.5748,
+      "step": 2622
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.816901267967701e-05,
+      "loss": 0.5924,
+      "step": 2623
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.815373410079171e-05,
+      "loss": 0.5519,
+      "step": 2624
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8138456608547604e-05,
+      "loss": 0.574,
+      "step": 2625
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8123180211936417e-05,
+      "loss": 0.5452,
+      "step": 2626
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.810790491994926e-05,
+      "loss": 0.583,
+      "step": 2627
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.809263074157658e-05,
+      "loss": 0.5774,
+      "step": 2628
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.807735768580817e-05,
+      "loss": 0.5898,
+      "step": 2629
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8062085761633168e-05,
+      "loss": 0.5594,
+      "step": 2630
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.804681497804004e-05,
+      "loss": 0.5499,
+      "step": 2631
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.803154534401659e-05,
+      "loss": 0.5782,
+      "step": 2632
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.801627686854992e-05,
+      "loss": 0.5848,
+      "step": 2633
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.800100956062648e-05,
+      "loss": 0.5515,
+      "step": 2634
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7985743429232022e-05,
+      "loss": 0.5422,
+      "step": 2635
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.79704784833516e-05,
+      "loss": 0.5835,
+      "step": 2636
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7955214731969578e-05,
+      "loss": 0.5783,
+      "step": 2637
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.793995218406961e-05,
+      "loss": 0.5693,
+      "step": 2638
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7924690848634653e-05,
+      "loss": 0.5898,
+      "step": 2639
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7909430734646936e-05,
+      "loss": 0.5767,
+      "step": 2640
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7894171851087975e-05,
+      "loss": 0.5482,
+      "step": 2641
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7878914206938566e-05,
+      "loss": 0.5734,
+      "step": 2642
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.786365781117877e-05,
+      "loss": 0.5706,
+      "step": 2643
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7848402672787918e-05,
+      "loss": 0.5652,
+      "step": 2644
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.78331488007446e-05,
+      "loss": 0.5865,
+      "step": 2645
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7817896204026652e-05,
+      "loss": 0.5994,
+      "step": 2646
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.780264489161117e-05,
+      "loss": 0.6073,
+      "step": 2647
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7787394872474495e-05,
+      "loss": 0.5948,
+      "step": 2648
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7772146155592195e-05,
+      "loss": 0.5632,
+      "step": 2649
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7756898749939086e-05,
+      "loss": 0.5819,
+      "step": 2650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.77416526644892e-05,
+      "loss": 0.5791,
+      "step": 2651
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7726407908215803e-05,
+      "loss": 0.5726,
+      "step": 2652
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7711164490091365e-05,
+      "loss": 0.5664,
+      "step": 2653
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.769592241908758e-05,
+      "loss": 0.5781,
+      "step": 2654
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7680681704175348e-05,
+      "loss": 0.5672,
+      "step": 2655
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7665442354324765e-05,
+      "loss": 0.5897,
+      "step": 2656
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7650204378505126e-05,
+      "loss": 0.5671,
+      "step": 2657
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7634967785684922e-05,
+      "loss": 0.5951,
+      "step": 2658
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.761973258483183e-05,
+      "loss": 0.5609,
+      "step": 2659
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.760449878491269e-05,
+      "loss": 0.5727,
+      "step": 2660
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.758926639489354e-05,
+      "loss": 0.5683,
+      "step": 2661
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7574035423739586e-05,
+      "loss": 0.5596,
+      "step": 2662
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7558805880415182e-05,
+      "loss": 0.572,
+      "step": 2663
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7543577773883856e-05,
+      "loss": 0.5648,
+      "step": 2664
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7528351113108283e-05,
+      "loss": 0.574,
+      "step": 2665
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7513125907050302e-05,
+      "loss": 0.5989,
+      "step": 2666
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7497902164670872e-05,
+      "loss": 0.5817,
+      "step": 2667
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.748267989493011e-05,
+      "loss": 0.5706,
+      "step": 2668
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.746745910678725e-05,
+      "loss": 0.5618,
+      "step": 2669
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7452239809200673e-05,
+      "loss": 0.5969,
+      "step": 2670
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.743702201112787e-05,
+      "loss": 0.5724,
+      "step": 2671
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7421805721525457e-05,
+      "loss": 0.5793,
+      "step": 2672
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7406590949349147e-05,
+      "loss": 0.5735,
+      "step": 2673
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7391377703553777e-05,
+      "loss": 0.5948,
+      "step": 2674
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7376165993093278e-05,
+      "loss": 0.5877,
+      "step": 2675
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7360955826920677e-05,
+      "loss": 0.5556,
+      "step": 2676
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.73457472139881e-05,
+      "loss": 0.6271,
+      "step": 2677
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.733054016324674e-05,
+      "loss": 0.5582,
+      "step": 2678
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7315334683646898e-05,
+      "loss": 0.6154,
+      "step": 2679
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.730013078413792e-05,
+      "loss": 0.5561,
+      "step": 2680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.728492847366825e-05,
+      "loss": 0.5934,
+      "step": 2681
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7269727761185374e-05,
+      "loss": 0.5824,
+      "step": 2682
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7254528655635853e-05,
+      "loss": 0.5853,
+      "step": 2683
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7239331165965295e-05,
+      "loss": 0.5684,
+      "step": 2684
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.722413530111836e-05,
+      "loss": 0.5778,
+      "step": 2685
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.720894107003875e-05,
+      "loss": 0.6042,
+      "step": 2686
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7193748481669205e-05,
+      "loss": 0.5778,
+      "step": 2687
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.71785575449515e-05,
+      "loss": 0.5564,
+      "step": 2688
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7163368268826433e-05,
+      "loss": 0.5972,
+      "step": 2689
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.714818066223383e-05,
+      "loss": 0.608,
+      "step": 2690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7132994734112536e-05,
+      "loss": 0.5549,
+      "step": 2691
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7117810493400403e-05,
+      "loss": 0.5716,
+      "step": 2692
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7102627949034293e-05,
+      "loss": 0.5951,
+      "step": 2693
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7087447109950064e-05,
+      "loss": 0.5499,
+      "step": 2694
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.707226798508258e-05,
+      "loss": 0.5994,
+      "step": 2695
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7057090583365678e-05,
+      "loss": 0.5475,
+      "step": 2696
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7041914913732213e-05,
+      "loss": 0.5694,
+      "step": 2697
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7026740985113992e-05,
+      "loss": 0.5742,
+      "step": 2698
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7011568806441808e-05,
+      "loss": 0.567,
+      "step": 2699
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6996398386645417e-05,
+      "loss": 0.5528,
+      "step": 2700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.698122973465355e-05,
+      "loss": 0.5642,
+      "step": 2701
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.69660628593939e-05,
+      "loss": 0.572,
+      "step": 2702
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6950897769793093e-05,
+      "loss": 0.5917,
+      "step": 2703
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6935734474776726e-05,
+      "loss": 0.5669,
+      "step": 2704
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.692057298326933e-05,
+      "loss": 0.5986,
+      "step": 2705
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6905413304194375e-05,
+      "loss": 0.5867,
+      "step": 2706
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.689025544647427e-05,
+      "loss": 0.5768,
+      "step": 2707
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6875099419030344e-05,
+      "loss": 0.5594,
+      "step": 2708
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.685994523078285e-05,
+      "loss": 0.5828,
+      "step": 2709
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6844792890650976e-05,
+      "loss": 0.5529,
+      "step": 2710
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.682964240755279e-05,
+      "loss": 0.5582,
+      "step": 2711
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6814493790405292e-05,
+      "loss": 0.5804,
+      "step": 2712
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6799347048124372e-05,
+      "loss": 0.5728,
+      "step": 2713
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6784202189624822e-05,
+      "loss": 0.5968,
+      "step": 2714
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6769059223820328e-05,
+      "loss": 0.5927,
+      "step": 2715
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6753918159623462e-05,
+      "loss": 0.5511,
+      "step": 2716
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.673877900594566e-05,
+      "loss": 0.582,
+      "step": 2717
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6723641771697246e-05,
+      "loss": 0.5666,
+      "step": 2718
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6708506465787427e-05,
+      "loss": 0.5688,
+      "step": 2719
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6693373097124255e-05,
+      "loss": 0.5819,
+      "step": 2720
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.667824167461465e-05,
+      "loss": 0.5686,
+      "step": 2721
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.666311220716439e-05,
+      "loss": 0.5907,
+      "step": 2722
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6647984703678096e-05,
+      "loss": 0.5764,
+      "step": 2723
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6632859173059232e-05,
+      "loss": 0.5594,
+      "step": 2724
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6617735624210106e-05,
+      "loss": 0.5717,
+      "step": 2725
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6602614066031857e-05,
+      "loss": 0.5629,
+      "step": 2726
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.658749450742446e-05,
+      "loss": 0.5988,
+      "step": 2727
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6572376957286694e-05,
+      "loss": 0.5615,
+      "step": 2728
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.655726142451618e-05,
+      "loss": 0.5714,
+      "step": 2729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6542147918009342e-05,
+      "loss": 0.5785,
+      "step": 2730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.5629,
+      "step": 2731
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.651192701936638e-05,
+      "loss": 0.5601,
+      "step": 2732
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.649681964501712e-05,
+      "loss": 0.6043,
+      "step": 2733
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6481714332505247e-05,
+      "loss": 0.5634,
+      "step": 2734
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.646661109072115e-05,
+      "loss": 0.5793,
+      "step": 2735
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6451509928554035e-05,
+      "loss": 0.5915,
+      "step": 2736
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6436410854891856e-05,
+      "loss": 0.5838,
+      "step": 2737
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6421313878621344e-05,
+      "loss": 0.5356,
+      "step": 2738
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.640621900862801e-05,
+      "loss": 0.5967,
+      "step": 2739
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6391126253796107e-05,
+      "loss": 0.5984,
+      "step": 2740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.637603562300865e-05,
+      "loss": 0.576,
+      "step": 2741
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.636094712514741e-05,
+      "loss": 0.553,
+      "step": 2742
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6345860769092898e-05,
+      "loss": 0.5667,
+      "step": 2743
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6330776563724354e-05,
+      "loss": 0.5746,
+      "step": 2744
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.631569451791977e-05,
+      "loss": 0.5439,
+      "step": 2745
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.630061464055586e-05,
+      "loss": 0.6046,
+      "step": 2746
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.628553694050805e-05,
+      "loss": 0.5854,
+      "step": 2747
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6270461426650505e-05,
+      "loss": 0.607,
+      "step": 2748
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6255388107856094e-05,
+      "loss": 0.5721,
+      "step": 2749
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.624031699299639e-05,
+      "loss": 0.5654,
+      "step": 2750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.622524809094167e-05,
+      "loss": 0.5913,
+      "step": 2751
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6210181410560912e-05,
+      "loss": 0.5542,
+      "step": 2752
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6195116960721785e-05,
+      "loss": 0.5808,
+      "step": 2753
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.618005475029065e-05,
+      "loss": 0.5843,
+      "step": 2754
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6164994788132538e-05,
+      "loss": 0.568,
+      "step": 2755
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6149937083111175e-05,
+      "loss": 0.6075,
+      "step": 2756
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.613488164408894e-05,
+      "loss": 0.5875,
+      "step": 2757
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6119828479926883e-05,
+      "loss": 0.5448,
+      "step": 2758
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.610477759948472e-05,
+      "loss": 0.5858,
+      "step": 2759
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.608972901162083e-05,
+      "loss": 0.5779,
+      "step": 2760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6074682725192223e-05,
+      "loss": 0.5761,
+      "step": 2761
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6059638749054568e-05,
+      "loss": 0.5503,
+      "step": 2762
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.604459709206218e-05,
+      "loss": 0.5219,
+      "step": 2763
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6029557763067986e-05,
+      "loss": 0.5744,
+      "step": 2764
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6014520770923566e-05,
+      "loss": 0.5795,
+      "step": 2765
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.5999486124479115e-05,
+      "loss": 0.559,
+      "step": 2766
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.5984453832583447e-05,
+      "loss": 0.591,
+      "step": 2767
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.596942390408399e-05,
+      "loss": 0.59,
+      "step": 2768
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.595439634782679e-05,
+      "loss": 0.589,
+      "step": 2769
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.593937117265648e-05,
+      "loss": 0.5604,
+      "step": 2770
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5924348387416306e-05,
+      "loss": 0.5511,
+      "step": 2771
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.59093280009481e-05,
+      "loss": 0.5787,
+      "step": 2772
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5894310022092288e-05,
+      "loss": 0.585,
+      "step": 2773
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.587929445968787e-05,
+      "loss": 0.5765,
+      "step": 2774
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.586428132257243e-05,
+      "loss": 0.5587,
+      "step": 2775
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.584927061958213e-05,
+      "loss": 0.5604,
+      "step": 2776
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5834262359551694e-05,
+      "loss": 0.5739,
+      "step": 2777
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5819256551314398e-05,
+      "loss": 0.5712,
+      "step": 2778
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5804253203702088e-05,
+      "loss": 0.5625,
+      "step": 2779
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5789252325545157e-05,
+      "loss": 0.6043,
+      "step": 2780
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5774253925672547e-05,
+      "loss": 0.5627,
+      "step": 2781
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5759258012911742e-05,
+      "loss": 0.5756,
+      "step": 2782
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5744264596088763e-05,
+      "loss": 0.5635,
+      "step": 2783
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.572927368402815e-05,
+      "loss": 0.5504,
+      "step": 2784
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5714285285552982e-05,
+      "loss": 0.5575,
+      "step": 2785
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5699299409484856e-05,
+      "loss": 0.5536,
+      "step": 2786
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.568431606464388e-05,
+      "loss": 0.5722,
+      "step": 2787
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.566933525984868e-05,
+      "loss": 0.576,
+      "step": 2788
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5654357003916378e-05,
+      "loss": 0.5854,
+      "step": 2789
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5639381305662606e-05,
+      "loss": 0.5734,
+      "step": 2790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5624408173901476e-05,
+      "loss": 0.5853,
+      "step": 2791
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5609437617445603e-05,
+      "loss": 0.5786,
+      "step": 2792
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.559446964510608e-05,
+      "loss": 0.5827,
+      "step": 2793
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.557950426569248e-05,
+      "loss": 0.5705,
+      "step": 2794
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5564541488012854e-05,
+      "loss": 0.5557,
+      "step": 2795
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5549581320873715e-05,
+      "loss": 0.5576,
+      "step": 2796
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5534623773080047e-05,
+      "loss": 0.532,
+      "step": 2797
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5519668853435282e-05,
+      "loss": 0.5529,
+      "step": 2798
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.550471657074131e-05,
+      "loss": 0.5742,
+      "step": 2799
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5489766933798478e-05,
+      "loss": 0.535,
+      "step": 2800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.547481995140556e-05,
+      "loss": 0.5866,
+      "step": 2801
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5459875632359785e-05,
+      "loss": 0.5891,
+      "step": 2802
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5444933985456795e-05,
+      "loss": 0.5622,
+      "step": 2803
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.542999501949068e-05,
+      "loss": 0.5825,
+      "step": 2804
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.541505874325393e-05,
+      "loss": 0.5701,
+      "step": 2805
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.540012516553747e-05,
+      "loss": 0.5599,
+      "step": 2806
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.538519429513063e-05,
+      "loss": 0.5735,
+      "step": 2807
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5370266140821143e-05,
+      "loss": 0.5764,
+      "step": 2808
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5355340711395154e-05,
+      "loss": 0.5785,
+      "step": 2809
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5340418015637193e-05,
+      "loss": 0.5647,
+      "step": 2810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5325498062330185e-05,
+      "loss": 0.5566,
+      "step": 2811
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.531058086025544e-05,
+      "loss": 0.5407,
+      "step": 2812
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.529566641819265e-05,
+      "loss": 0.5906,
+      "step": 2813
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.528075474491989e-05,
+      "loss": 0.6024,
+      "step": 2814
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5265845849213588e-05,
+      "loss": 0.58,
+      "step": 2815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5250939739848556e-05,
+      "loss": 0.5783,
+      "step": 2816
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5236036425597957e-05,
+      "loss": 0.5877,
+      "step": 2817
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.52211359152333e-05,
+      "loss": 0.5755,
+      "step": 2818
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.520623821752446e-05,
+      "loss": 0.5808,
+      "step": 2819
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5191343341239648e-05,
+      "loss": 0.587,
+      "step": 2820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5176451295145416e-05,
+      "loss": 0.5906,
+      "step": 2821
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5161562088006649e-05,
+      "loss": 0.5496,
+      "step": 2822
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5146675728586565e-05,
+      "loss": 0.5672,
+      "step": 2823
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.51317922256467e-05,
+      "loss": 0.5721,
+      "step": 2824
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5116911587946917e-05,
+      "loss": 0.5612,
+      "step": 2825
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5102033824245383e-05,
+      "loss": 0.588,
+      "step": 2826
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.508715894329858e-05,
+      "loss": 0.5706,
+      "step": 2827
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.50722869538613e-05,
+      "loss": 0.6069,
+      "step": 2828
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5057417864686607e-05,
+      "loss": 0.5527,
+      "step": 2829
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.5042551684525888e-05,
+      "loss": 0.573,
+      "step": 2830
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.50276884221288e-05,
+      "loss": 0.5749,
+      "step": 2831
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.5012828086243293e-05,
+      "loss": 0.5543,
+      "step": 2832
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4997970685615588e-05,
+      "loss": 0.5935,
+      "step": 2833
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4983116228990183e-05,
+      "loss": 0.5441,
+      "step": 2834
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.496826472510983e-05,
+      "loss": 0.5715,
+      "step": 2835
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4953416182715566e-05,
+      "loss": 0.5837,
+      "step": 2836
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4938570610546665e-05,
+      "loss": 0.588,
+      "step": 2837
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4923728017340662e-05,
+      "loss": 0.5879,
+      "step": 2838
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4908888411833339e-05,
+      "loss": 0.5727,
+      "step": 2839
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4894051802758718e-05,
+      "loss": 0.5555,
+      "step": 2840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4879218198849059e-05,
+      "loss": 0.5969,
+      "step": 2841
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4864387608834845e-05,
+      "loss": 0.5503,
+      "step": 2842
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4849560041444795e-05,
+      "loss": 0.5885,
+      "step": 2843
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4834735505405845e-05,
+      "loss": 0.5674,
+      "step": 2844
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4819914009443149e-05,
+      "loss": 0.581,
+      "step": 2845
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.480509556228007e-05,
+      "loss": 0.5629,
+      "step": 2846
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4790280172638173e-05,
+      "loss": 0.5479,
+      "step": 2847
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4775467849237234e-05,
+      "loss": 0.5865,
+      "step": 2848
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4760658600795212e-05,
+      "loss": 0.5642,
+      "step": 2849
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4745852436028262e-05,
+      "loss": 0.5264,
+      "step": 2850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4731049363650725e-05,
+      "loss": 0.6103,
+      "step": 2851
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4716249392375122e-05,
+      "loss": 0.5597,
+      "step": 2852
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4701452530912145e-05,
+      "loss": 0.5544,
+      "step": 2853
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4686658787970664e-05,
+      "loss": 0.5596,
+      "step": 2854
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4671868172257702e-05,
+      "loss": 0.5772,
+      "step": 2855
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4657080692478445e-05,
+      "loss": 0.5905,
+      "step": 2856
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.464229635733624e-05,
+      "loss": 0.5446,
+      "step": 2857
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4627515175532579e-05,
+      "loss": 0.5594,
+      "step": 2858
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4612737155767093e-05,
+      "loss": 0.5756,
+      "step": 2859
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4597962306737565e-05,
+      "loss": 0.5379,
+      "step": 2860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4583190637139901e-05,
+      "loss": 0.5484,
+      "step": 2861
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4568422155668131e-05,
+      "loss": 0.5648,
+      "step": 2862
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.455365687101442e-05,
+      "loss": 0.581,
+      "step": 2863
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4538894791869052e-05,
+      "loss": 0.5637,
+      "step": 2864
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4524135926920415e-05,
+      "loss": 0.5837,
+      "step": 2865
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.450938028485501e-05,
+      "loss": 0.5754,
+      "step": 2866
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4494627874357443e-05,
+      "loss": 0.5613,
+      "step": 2867
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4479878704110417e-05,
+      "loss": 0.5637,
+      "step": 2868
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4465132782794721e-05,
+      "loss": 0.5584,
+      "step": 2869
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4450390119089245e-05,
+      "loss": 0.5689,
+      "step": 2870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.443565072167095e-05,
+      "loss": 0.5752,
+      "step": 2871
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4420914599214879e-05,
+      "loss": 0.5359,
+      "step": 2872
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4406181760394149e-05,
+      "loss": 0.5793,
+      "step": 2873
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4391452213879949e-05,
+      "loss": 0.5875,
+      "step": 2874
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4376725968341512e-05,
+      "loss": 0.5961,
+      "step": 2875
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4362003032446148e-05,
+      "loss": 0.5638,
+      "step": 2876
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4347283414859212e-05,
+      "loss": 0.5606,
+      "step": 2877
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.43325671242441e-05,
+      "loss": 0.5504,
+      "step": 2878
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4317854169262262e-05,
+      "loss": 0.5583,
+      "step": 2879
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4303144558573172e-05,
+      "loss": 0.5716,
+      "step": 2880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.428843830083435e-05,
+      "loss": 0.5632,
+      "step": 2881
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4273735404701325e-05,
+      "loss": 0.5698,
+      "step": 2882
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4259035878827661e-05,
+      "loss": 0.546,
+      "step": 2883
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4244339731864934e-05,
+      "loss": 0.5949,
+      "step": 2884
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4229646972462732e-05,
+      "loss": 0.5628,
+      "step": 2885
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4214957609268648e-05,
+      "loss": 0.5691,
+      "step": 2886
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4200271650928277e-05,
+      "loss": 0.6048,
+      "step": 2887
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4185589106085215e-05,
+      "loss": 0.5943,
+      "step": 2888
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4170909983381037e-05,
+      "loss": 0.5501,
+      "step": 2889
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4156234291455309e-05,
+      "loss": 0.5601,
+      "step": 2890
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4141562038945584e-05,
+      "loss": 0.5523,
+      "step": 2891
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.412689323448739e-05,
+      "loss": 0.5581,
+      "step": 2892
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4112227886714212e-05,
+      "loss": 0.5862,
+      "step": 2893
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4097566004257517e-05,
+      "loss": 0.573,
+      "step": 2894
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4082907595746726e-05,
+      "loss": 0.5601,
+      "step": 2895
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4068252669809215e-05,
+      "loss": 0.5567,
+      "step": 2896
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4053601235070305e-05,
+      "loss": 0.5541,
+      "step": 2897
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.403895330015327e-05,
+      "loss": 0.5754,
+      "step": 2898
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4024308873679327e-05,
+      "loss": 0.5511,
+      "step": 2899
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4009667964267617e-05,
+      "loss": 0.5512,
+      "step": 2900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3995030580535223e-05,
+      "loss": 0.583,
+      "step": 2901
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.398039673109714e-05,
+      "loss": 0.6039,
+      "step": 2902
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3965766424566291e-05,
+      "loss": 0.5703,
+      "step": 2903
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3951139669553515e-05,
+      "loss": 0.583,
+      "step": 2904
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3936516474667557e-05,
+      "loss": 0.5633,
+      "step": 2905
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3921896848515064e-05,
+      "loss": 0.586,
+      "step": 2906
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3907280799700594e-05,
+      "loss": 0.5506,
+      "step": 2907
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3892668336826591e-05,
+      "loss": 0.5603,
+      "step": 2908
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3878059468493379e-05,
+      "loss": 0.5438,
+      "step": 2909
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3863454203299188e-05,
+      "loss": 0.5671,
+      "step": 2910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3848852549840109e-05,
+      "loss": 0.573,
+      "step": 2911
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3834254516710116e-05,
+      "loss": 0.5739,
+      "step": 2912
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3819660112501054e-05,
+      "loss": 0.579,
+      "step": 2913
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3805069345802627e-05,
+      "loss": 0.5632,
+      "step": 2914
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.37904822252024e-05,
+      "loss": 0.5751,
+      "step": 2915
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3775898759285786e-05,
+      "loss": 0.5469,
+      "step": 2916
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.376131895663606e-05,
+      "loss": 0.5667,
+      "step": 2917
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3746742825834332e-05,
+      "loss": 0.5468,
+      "step": 2918
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3732170375459551e-05,
+      "loss": 0.5693,
+      "step": 2919
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.37176016140885e-05,
+      "loss": 0.5614,
+      "step": 2920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3703036550295802e-05,
+      "loss": 0.574,
+      "step": 2921
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3688475192653882e-05,
+      "loss": 0.5743,
+      "step": 2922
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3673917549733e-05,
+      "loss": 0.5803,
+      "step": 2923
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3659363630101225e-05,
+      "loss": 0.5663,
+      "step": 2924
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3644813442324436e-05,
+      "loss": 0.5737,
+      "step": 2925
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3630266994966314e-05,
+      "loss": 0.551,
+      "step": 2926
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3615724296588342e-05,
+      "loss": 0.5513,
+      "step": 2927
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.360118535574979e-05,
+      "loss": 0.5592,
+      "step": 2928
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3586650181007724e-05,
+      "loss": 0.5665,
+      "step": 2929
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.357211878091698e-05,
+      "loss": 0.5576,
+      "step": 2930
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3557591164030193e-05,
+      "loss": 0.5657,
+      "step": 2931
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3543067338897752e-05,
+      "loss": 0.5602,
+      "step": 2932
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3528547314067831e-05,
+      "loss": 0.5534,
+      "step": 2933
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3514031098086349e-05,
+      "loss": 0.556,
+      "step": 2934
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3499518699497004e-05,
+      "loss": 0.5669,
+      "step": 2935
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3485010126841225e-05,
+      "loss": 0.5844,
+      "step": 2936
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3470505388658208e-05,
+      "loss": 0.5581,
+      "step": 2937
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3456004493484882e-05,
+      "loss": 0.6081,
+      "step": 2938
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3441507449855914e-05,
+      "loss": 0.562,
+      "step": 2939
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3427014266303715e-05,
+      "loss": 0.5731,
+      "step": 2940
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.341252495135841e-05,
+      "loss": 0.6012,
+      "step": 2941
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3398039513547861e-05,
+      "loss": 0.5931,
+      "step": 2942
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3383557961397624e-05,
+      "loss": 0.5671,
+      "step": 2943
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3369080303430999e-05,
+      "loss": 0.5612,
+      "step": 2944
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3354606548168972e-05,
+      "loss": 0.5467,
+      "step": 2945
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.334013670413024e-05,
+      "loss": 0.5524,
+      "step": 2946
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3325670779831201e-05,
+      "loss": 0.5541,
+      "step": 2947
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3311208783785945e-05,
+      "loss": 0.5543,
+      "step": 2948
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3296750724506238e-05,
+      "loss": 0.5794,
+      "step": 2949
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3282296610501542e-05,
+      "loss": 0.5817,
+      "step": 2950
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3267846450278996e-05,
+      "loss": 0.5761,
+      "step": 2951
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3253400252343403e-05,
+      "loss": 0.5841,
+      "step": 2952
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3238958025197246e-05,
+      "loss": 0.5536,
+      "step": 2953
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3224519777340666e-05,
+      "loss": 0.5723,
+      "step": 2954
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3210085517271459e-05,
+      "loss": 0.55,
+      "step": 2955
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3195655253485074e-05,
+      "loss": 0.5835,
+      "step": 2956
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3181228994474609e-05,
+      "loss": 0.5774,
+      "step": 2957
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3166806748730812e-05,
+      "loss": 0.5789,
+      "step": 2958
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3152388524742061e-05,
+      "loss": 0.5613,
+      "step": 2959
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3137974330994366e-05,
+      "loss": 0.5848,
+      "step": 2960
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3123564175971363e-05,
+      "loss": 0.5682,
+      "step": 2961
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3109158068154329e-05,
+      "loss": 0.5809,
+      "step": 2962
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.309475601602213e-05,
+      "loss": 0.5407,
+      "step": 2963
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3080358028051273e-05,
+      "loss": 0.5683,
+      "step": 2964
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3065964112715857e-05,
+      "loss": 0.5536,
+      "step": 2965
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3051574278487587e-05,
+      "loss": 0.5646,
+      "step": 2966
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3037188533835766e-05,
+      "loss": 0.5723,
+      "step": 2967
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3022806887227288e-05,
+      "loss": 0.57,
+      "step": 2968
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3008429347126641e-05,
+      "loss": 0.5675,
+      "step": 2969
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2994055921995893e-05,
+      "loss": 0.5223,
+      "step": 2970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2979686620294691e-05,
+      "loss": 0.5792,
+      "step": 2971
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2965321450480255e-05,
+      "loss": 0.5502,
+      "step": 2972
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2950960421007366e-05,
+      "loss": 0.5627,
+      "step": 2973
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2936603540328376e-05,
+      "loss": 0.5807,
+      "step": 2974
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2922250816893196e-05,
+      "loss": 0.5691,
+      "step": 2975
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2907902259149287e-05,
+      "loss": 0.5916,
+      "step": 2976
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2893557875541661e-05,
+      "loss": 0.5906,
+      "step": 2977
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2879217674512865e-05,
+      "loss": 0.5578,
+      "step": 2978
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2864881664502998e-05,
+      "loss": 0.5487,
+      "step": 2979
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2850549853949678e-05,
+      "loss": 0.5465,
+      "step": 2980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2836222251288061e-05,
+      "loss": 0.5935,
+      "step": 2981
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2821898864950826e-05,
+      "loss": 0.5264,
+      "step": 2982
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2807579703368162e-05,
+      "loss": 0.5943,
+      "step": 2983
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2793264774967784e-05,
+      "loss": 0.5569,
+      "step": 2984
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2778954088174905e-05,
+      "loss": 0.5459,
+      "step": 2985
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.276464765141225e-05,
+      "loss": 0.5591,
+      "step": 2986
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.275034547310003e-05,
+      "loss": 0.5631,
+      "step": 2987
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2736047561655962e-05,
+      "loss": 0.5783,
+      "step": 2988
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2721753925495248e-05,
+      "loss": 0.5767,
+      "step": 2989
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2707464573030572e-05,
+      "loss": 0.5664,
+      "step": 2990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.26931795126721e-05,
+      "loss": 0.5835,
+      "step": 2991
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.267889875282747e-05,
+      "loss": 0.5713,
+      "step": 2992
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2664622301901788e-05,
+      "loss": 0.5547,
+      "step": 2993
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2650350168297618e-05,
+      "loss": 0.5796,
+      "step": 2994
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2636082360414999e-05,
+      "loss": 0.571,
+      "step": 2995
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2621818886651408e-05,
+      "loss": 0.5774,
+      "step": 2996
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.260755975540178e-05,
+      "loss": 0.5572,
+      "step": 2997
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2593304975058497e-05,
+      "loss": 0.5687,
+      "step": 2998
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2579054554011371e-05,
+      "loss": 0.575,
+      "step": 2999
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.256480850064765e-05,
+      "loss": 0.5557,
+      "step": 3000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2550566823352015e-05,
+      "loss": 0.5346,
+      "step": 3001
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2536329530506569e-05,
+      "loss": 0.5791,
+      "step": 3002
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2522096630490838e-05,
+      "loss": 0.584,
+      "step": 3003
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.250786813168176e-05,
+      "loss": 0.5684,
+      "step": 3004
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2493644042453683e-05,
+      "loss": 0.5771,
+      "step": 3005
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2479424371178362e-05,
+      "loss": 0.5196,
+      "step": 3006
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2465209126224941e-05,
+      "loss": 0.5819,
+      "step": 3007
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2450998315959972e-05,
+      "loss": 0.5741,
+      "step": 3008
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2436791948747396e-05,
+      "loss": 0.5836,
+      "step": 3009
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2422590032948528e-05,
+      "loss": 0.5713,
+      "step": 3010
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2408392576922075e-05,
+      "loss": 0.5743,
+      "step": 3011
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2394199589024112e-05,
+      "loss": 0.5958,
+      "step": 3012
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.238001107760809e-05,
+      "loss": 0.597,
+      "step": 3013
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2365827051024812e-05,
+      "loss": 0.5694,
+      "step": 3014
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.235164751762246e-05,
+      "loss": 0.5911,
+      "step": 3015
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2337472485746561e-05,
+      "loss": 0.5968,
+      "step": 3016
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2323301963739995e-05,
+      "loss": 0.5761,
+      "step": 3017
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2309135959942986e-05,
+      "loss": 0.5889,
+      "step": 3018
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2294974482693101e-05,
+      "loss": 0.562,
+      "step": 3019
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2280817540325237e-05,
+      "loss": 0.5581,
+      "step": 3020
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2266665141171625e-05,
+      "loss": 0.5785,
+      "step": 3021
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2252517293561833e-05,
+      "loss": 0.5513,
+      "step": 3022
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2238374005822733e-05,
+      "loss": 0.5747,
+      "step": 3023
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2224235286278521e-05,
+      "loss": 0.5437,
+      "step": 3024
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2210101143250708e-05,
+      "loss": 0.5812,
+      "step": 3025
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.219597158505811e-05,
+      "loss": 0.5759,
+      "step": 3026
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2181846620016833e-05,
+      "loss": 0.5672,
+      "step": 3027
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2167726256440294e-05,
+      "loss": 0.5661,
+      "step": 3028
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2153610502639194e-05,
+      "loss": 0.5805,
+      "step": 3029
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.213949936692153e-05,
+      "loss": 0.5779,
+      "step": 3030
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2125392857592569e-05,
+      "loss": 0.5965,
+      "step": 3031
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.211129098295486e-05,
+      "loss": 0.5722,
+      "step": 3032
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2097193751308232e-05,
+      "loss": 0.5936,
+      "step": 3033
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.208310117094976e-05,
+      "loss": 0.5569,
+      "step": 3034
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2069013250173808e-05,
+      "loss": 0.5261,
+      "step": 3035
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.205492999727198e-05,
+      "loss": 0.5312,
+      "step": 3036
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2040851420533137e-05,
+      "loss": 0.5195,
+      "step": 3037
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.202677752824339e-05,
+      "loss": 0.5127,
+      "step": 3038
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2012708328686093e-05,
+      "loss": 0.5257,
+      "step": 3039
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1998643830141835e-05,
+      "loss": 0.5367,
+      "step": 3040
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1984584040888433e-05,
+      "loss": 0.5246,
+      "step": 3041
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.197052896920095e-05,
+      "loss": 0.5419,
+      "step": 3042
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1956478623351652e-05,
+      "loss": 0.541,
+      "step": 3043
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.194243301161004e-05,
+      "loss": 0.5218,
+      "step": 3044
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1928392142242818e-05,
+      "loss": 0.5195,
+      "step": 3045
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1914356023513904e-05,
+      "loss": 0.5224,
+      "step": 3046
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1900324663684417e-05,
+      "loss": 0.5171,
+      "step": 3047
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1886298071012674e-05,
+      "loss": 0.5332,
+      "step": 3048
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1872276253754189e-05,
+      "loss": 0.5159,
+      "step": 3049
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.185825922016167e-05,
+      "loss": 0.5326,
+      "step": 3050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1844246978484998e-05,
+      "loss": 0.5081,
+      "step": 3051
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1830239536971244e-05,
+      "loss": 0.5272,
+      "step": 3052
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1816236903864656e-05,
+      "loss": 0.5387,
+      "step": 3053
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1802239087406635e-05,
+      "loss": 0.4919,
+      "step": 3054
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1788246095835763e-05,
+      "loss": 0.5515,
+      "step": 3055
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1774257937387774e-05,
+      "loss": 0.5423,
+      "step": 3056
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1760274620295563e-05,
+      "loss": 0.5287,
+      "step": 3057
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1746296152789176e-05,
+      "loss": 0.5383,
+      "step": 3058
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1732322543095802e-05,
+      "loss": 0.5157,
+      "step": 3059
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1718353799439766e-05,
+      "loss": 0.5225,
+      "step": 3060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1704389930042534e-05,
+      "loss": 0.5417,
+      "step": 3061
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1690430943122709e-05,
+      "loss": 0.5458,
+      "step": 3062
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1676476846896013e-05,
+      "loss": 0.5231,
+      "step": 3063
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1662527649575286e-05,
+      "loss": 0.5244,
+      "step": 3064
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1648583359370491e-05,
+      "loss": 0.5021,
+      "step": 3065
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1634643984488703e-05,
+      "loss": 0.52,
+      "step": 3066
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1620709533134104e-05,
+      "loss": 0.5156,
+      "step": 3067
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1606780013507974e-05,
+      "loss": 0.5211,
+      "step": 3068
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1592855433808694e-05,
+      "loss": 0.5526,
+      "step": 3069
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1578935802231741e-05,
+      "loss": 0.5299,
+      "step": 3070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1565021126969667e-05,
+      "loss": 0.5302,
+      "step": 3071
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1551111416212124e-05,
+      "loss": 0.532,
+      "step": 3072
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1537206678145826e-05,
+      "loss": 0.5228,
+      "step": 3073
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1523306920954571e-05,
+      "loss": 0.5506,
+      "step": 3074
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1509412152819226e-05,
+      "loss": 0.5283,
+      "step": 3075
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1495522381917719e-05,
+      "loss": 0.5319,
+      "step": 3076
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1481637616425024e-05,
+      "loss": 0.5217,
+      "step": 3077
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1467757864513183e-05,
+      "loss": 0.5046,
+      "step": 3078
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1453883134351291e-05,
+      "loss": 0.5209,
+      "step": 3079
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1440013434105474e-05,
+      "loss": 0.5151,
+      "step": 3080
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1426148771938915e-05,
+      "loss": 0.5258,
+      "step": 3081
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1412289156011816e-05,
+      "loss": 0.499,
+      "step": 3082
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1398434594481416e-05,
+      "loss": 0.528,
+      "step": 3083
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1384585095501977e-05,
+      "loss": 0.5248,
+      "step": 3084
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1370740667224778e-05,
+      "loss": 0.5082,
+      "step": 3085
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1356901317798123e-05,
+      "loss": 0.4715,
+      "step": 3086
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1343067055367328e-05,
+      "loss": 0.521,
+      "step": 3087
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1329237888074691e-05,
+      "loss": 0.5135,
+      "step": 3088
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.131541382405954e-05,
+      "loss": 0.5295,
+      "step": 3089
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1301594871458183e-05,
+      "loss": 0.5214,
+      "step": 3090
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1287781038403929e-05,
+      "loss": 0.5237,
+      "step": 3091
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1273972333027066e-05,
+      "loss": 0.532,
+      "step": 3092
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1260168763454868e-05,
+      "loss": 0.5027,
+      "step": 3093
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.124637033781159e-05,
+      "loss": 0.509,
+      "step": 3094
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.123257706421845e-05,
+      "loss": 0.5084,
+      "step": 3095
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1218788950793642e-05,
+      "loss": 0.5157,
+      "step": 3096
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1205006005652319e-05,
+      "loss": 0.5267,
+      "step": 3097
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1191228236906593e-05,
+      "loss": 0.5189,
+      "step": 3098
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.117745565266553e-05,
+      "loss": 0.5092,
+      "step": 3099
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1163688261035151e-05,
+      "loss": 0.533,
+      "step": 3100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.11499260701184e-05,
+      "loss": 0.5294,
+      "step": 3101
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1136169088015177e-05,
+      "loss": 0.5157,
+      "step": 3102
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1122417322822319e-05,
+      "loss": 0.5189,
+      "step": 3103
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.110867078263358e-05,
+      "loss": 0.5164,
+      "step": 3104
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1094929475539654e-05,
+      "loss": 0.5214,
+      "step": 3105
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.108119340962814e-05,
+      "loss": 0.5174,
+      "step": 3106
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.106746259298356e-05,
+      "loss": 0.5264,
+      "step": 3107
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1053737033687346e-05,
+      "loss": 0.5074,
+      "step": 3108
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1040016739817836e-05,
+      "loss": 0.5271,
+      "step": 3109
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1026301719450267e-05,
+      "loss": 0.5253,
+      "step": 3110
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1012591980656771e-05,
+      "loss": 0.5422,
+      "step": 3111
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0998887531506376e-05,
+      "loss": 0.5383,
+      "step": 3112
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0985188380064998e-05,
+      "loss": 0.5127,
+      "step": 3113
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0971494534395428e-05,
+      "loss": 0.5322,
+      "step": 3114
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0957806002557331e-05,
+      "loss": 0.5078,
+      "step": 3115
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.094412279260726e-05,
+      "loss": 0.521,
+      "step": 3116
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0930444912598618e-05,
+      "loss": 0.5323,
+      "step": 3117
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0916772370581687e-05,
+      "loss": 0.527,
+      "step": 3118
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0903105174603598e-05,
+      "loss": 0.5074,
+      "step": 3119
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0889443332708338e-05,
+      "loss": 0.5326,
+      "step": 3120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.087578685293674e-05,
+      "loss": 0.5241,
+      "step": 3121
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0862135743326487e-05,
+      "loss": 0.5098,
+      "step": 3122
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0848490011912096e-05,
+      "loss": 0.5555,
+      "step": 3123
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0834849666724917e-05,
+      "loss": 0.525,
+      "step": 3124
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0821214715793138e-05,
+      "loss": 0.5111,
+      "step": 3125
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0807585167141768e-05,
+      "loss": 0.5187,
+      "step": 3126
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0793961028792639e-05,
+      "loss": 0.5291,
+      "step": 3127
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0780342308764383e-05,
+      "loss": 0.5344,
+      "step": 3128
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0766729015072459e-05,
+      "loss": 0.504,
+      "step": 3129
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0753121155729133e-05,
+      "loss": 0.5108,
+      "step": 3130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0739518738743468e-05,
+      "loss": 0.508,
+      "step": 3131
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0725921772121324e-05,
+      "loss": 0.5248,
+      "step": 3132
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0712330263865352e-05,
+      "loss": 0.5307,
+      "step": 3133
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0698744221974992e-05,
+      "loss": 0.4928,
+      "step": 3134
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0685163654446464e-05,
+      "loss": 0.5521,
+      "step": 3135
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0671588569272773e-05,
+      "loss": 0.5256,
+      "step": 3136
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0658018974443692e-05,
+      "loss": 0.5381,
+      "step": 3137
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.064445487794576e-05,
+      "loss": 0.5452,
+      "step": 3138
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0630896287762282e-05,
+      "loss": 0.5115,
+      "step": 3139
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0617343211873337e-05,
+      "loss": 0.5064,
+      "step": 3140
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0603795658255727e-05,
+      "loss": 0.5254,
+      "step": 3141
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0590253634883028e-05,
+      "loss": 0.5325,
+      "step": 3142
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0576717149725553e-05,
+      "loss": 0.5153,
+      "step": 3143
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.056318621075036e-05,
+      "loss": 0.5184,
+      "step": 3144
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.054966082592124e-05,
+      "loss": 0.5357,
+      "step": 3145
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0536141003198713e-05,
+      "loss": 0.516,
+      "step": 3146
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0522626750540029e-05,
+      "loss": 0.5307,
+      "step": 3147
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0509118075899156e-05,
+      "loss": 0.5287,
+      "step": 3148
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0495614987226785e-05,
+      "loss": 0.539,
+      "step": 3149
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0482117492470311e-05,
+      "loss": 0.5168,
+      "step": 3150
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0468625599573842e-05,
+      "loss": 0.5189,
+      "step": 3151
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0455139316478192e-05,
+      "loss": 0.5187,
+      "step": 3152
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0441658651120861e-05,
+      "loss": 0.5213,
+      "step": 3153
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0428183611436064e-05,
+      "loss": 0.5312,
+      "step": 3154
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0414714205354675e-05,
+      "loss": 0.523,
+      "step": 3155
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0401250440804279e-05,
+      "loss": 0.5228,
+      "step": 3156
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0387792325709127e-05,
+      "loss": 0.5159,
+      "step": 3157
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.037433986799015e-05,
+      "loss": 0.5161,
+      "step": 3158
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0360893075564946e-05,
+      "loss": 0.5283,
+      "step": 3159
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0347451956347787e-05,
+      "loss": 0.5417,
+      "step": 3160
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.033401651824959e-05,
+      "loss": 0.5543,
+      "step": 3161
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0320586769177945e-05,
+      "loss": 0.5208,
+      "step": 3162
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0307162717037083e-05,
+      "loss": 0.5155,
+      "step": 3163
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0293744369727886e-05,
+      "loss": 0.5084,
+      "step": 3164
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.028033173514788e-05,
+      "loss": 0.5164,
+      "step": 3165
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0266924821191223e-05,
+      "loss": 0.4916,
+      "step": 3166
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0253523635748717e-05,
+      "loss": 0.5259,
+      "step": 3167
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0240128186707775e-05,
+      "loss": 0.5032,
+      "step": 3168
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0226738481952444e-05,
+      "loss": 0.5283,
+      "step": 3169
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0213354529363393e-05,
+      "loss": 0.5238,
+      "step": 3170
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0199976336817905e-05,
+      "loss": 0.5054,
+      "step": 3171
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0186603912189867e-05,
+      "loss": 0.5081,
+      "step": 3172
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0173237263349776e-05,
+      "loss": 0.5074,
+      "step": 3173
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0159876398164727e-05,
+      "loss": 0.5349,
+      "step": 3174
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0146521324498413e-05,
+      "loss": 0.5304,
+      "step": 3175
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0133172050211115e-05,
+      "loss": 0.4865,
+      "step": 3176
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0119828583159705e-05,
+      "loss": 0.5287,
+      "step": 3177
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0106490931197639e-05,
+      "loss": 0.5274,
+      "step": 3178
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0093159102174938e-05,
+      "loss": 0.5409,
+      "step": 3179
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0079833103938208e-05,
+      "loss": 0.5516,
+      "step": 3180
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0066512944330631e-05,
+      "loss": 0.5018,
+      "step": 3181
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0053198631191922e-05,
+      "loss": 0.5122,
+      "step": 3182
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0039890172358384e-05,
+      "loss": 0.52,
+      "step": 3183
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.002658757566286e-05,
+      "loss": 0.5331,
+      "step": 3184
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0013290848934755e-05,
+      "loss": 0.5404,
+      "step": 3185
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.5329,
+      "step": 3186
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.986715036681095e-06,
+      "loss": 0.5404,
+      "step": 3187
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.973435966797047e-06,
+      "loss": 0.5277,
+      "step": 3188
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.960162798163406e-06,
+      "loss": 0.5109,
+      "step": 3189
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.946895538592257e-06,
+      "loss": 0.5032,
+      "step": 3190
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.93363419589219e-06,
+      "loss": 0.5251,
+      "step": 3191
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.920378777868327e-06,
+      "loss": 0.523,
+      "step": 3192
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.907129292322298e-06,
+      "loss": 0.516,
+      "step": 3193
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.893885747052247e-06,
+      "loss": 0.5185,
+      "step": 3194
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.880648149852803e-06,
+      "loss": 0.5334,
+      "step": 3195
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.867416508515114e-06,
+      "loss": 0.5233,
+      "step": 3196
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.854190830826814e-06,
+      "loss": 0.5161,
+      "step": 3197
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.84097112457203e-06,
+      "loss": 0.5209,
+      "step": 3198
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.827757397531373e-06,
+      "loss": 0.516,
+      "step": 3199
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.814549657481935e-06,
+      "loss": 0.5346,
+      "step": 3200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.801347912197288e-06,
+      "loss": 0.5007,
+      "step": 3201
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.788152169447467e-06,
+      "loss": 0.534,
+      "step": 3202
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.77496243699898e-06,
+      "loss": 0.5091,
+      "step": 3203
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.761778722614799e-06,
+      "loss": 0.5299,
+      "step": 3204
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.748601034054353e-06,
+      "loss": 0.528,
+      "step": 3205
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.735429379073518e-06,
+      "loss": 0.5234,
+      "step": 3206
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.722263765424628e-06,
+      "loss": 0.516,
+      "step": 3207
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.709104200856458e-06,
+      "loss": 0.5573,
+      "step": 3208
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.69595069311422e-06,
+      "loss": 0.5356,
+      "step": 3209
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.682803249939565e-06,
+      "loss": 0.5431,
+      "step": 3210
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.66966187907057e-06,
+      "loss": 0.5377,
+      "step": 3211
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.656526588241745e-06,
+      "loss": 0.5296,
+      "step": 3212
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.643397385184016e-06,
+      "loss": 0.5116,
+      "step": 3213
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.630274277624729e-06,
+      "loss": 0.5325,
+      "step": 3214
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.617157273287636e-06,
+      "loss": 0.5236,
+      "step": 3215
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.604046379892909e-06,
+      "loss": 0.5334,
+      "step": 3216
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.59094160515711e-06,
+      "loss": 0.5315,
+      "step": 3217
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.577842956793215e-06,
+      "loss": 0.5401,
+      "step": 3218
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.564750442510571e-06,
+      "loss": 0.5265,
+      "step": 3219
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.551664070014939e-06,
+      "loss": 0.5011,
+      "step": 3220
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.538583847008452e-06,
+      "loss": 0.4952,
+      "step": 3221
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.525509781189626e-06,
+      "loss": 0.5282,
+      "step": 3222
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.512441880253355e-06,
+      "loss": 0.5377,
+      "step": 3223
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.499380151890905e-06,
+      "loss": 0.5225,
+      "step": 3224
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.486324603789905e-06,
+      "loss": 0.5035,
+      "step": 3225
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.473275243634353e-06,
+      "loss": 0.5223,
+      "step": 3226
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.460232079104597e-06,
+      "loss": 0.5188,
+      "step": 3227
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.447195117877343e-06,
+      "loss": 0.5027,
+      "step": 3228
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.434164367625648e-06,
+      "loss": 0.5192,
+      "step": 3229
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.421139836018909e-06,
+      "loss": 0.5049,
+      "step": 3230
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.408121530722867e-06,
+      "loss": 0.509,
+      "step": 3231
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.3951094593996e-06,
+      "loss": 0.5231,
+      "step": 3232
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.382103629707504e-06,
+      "loss": 0.5291,
+      "step": 3233
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.369104049301311e-06,
+      "loss": 0.4997,
+      "step": 3234
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.356110725832081e-06,
+      "loss": 0.5144,
+      "step": 3235
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.343123666947183e-06,
+      "loss": 0.5233,
+      "step": 3236
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.3301428802903e-06,
+      "loss": 0.5073,
+      "step": 3237
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.317168373501426e-06,
+      "loss": 0.516,
+      "step": 3238
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.304200154216856e-06,
+      "loss": 0.5108,
+      "step": 3239
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.291238230069184e-06,
+      "loss": 0.5356,
+      "step": 3240
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.278282608687303e-06,
+      "loss": 0.5384,
+      "step": 3241
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.265333297696395e-06,
+      "loss": 0.515,
+      "step": 3242
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.252390304717927e-06,
+      "loss": 0.5157,
+      "step": 3243
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.239453637369641e-06,
+      "loss": 0.5273,
+      "step": 3244
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.22652330326558e-06,
+      "loss": 0.535,
+      "step": 3245
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.213599310016019e-06,
+      "loss": 0.5406,
+      "step": 3246
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.200681665227535e-06,
+      "loss": 0.5167,
+      "step": 3247
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.187770376502958e-06,
+      "loss": 0.5171,
+      "step": 3248
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.174865451441375e-06,
+      "loss": 0.519,
+      "step": 3249
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.161966897638128e-06,
+      "loss": 0.5379,
+      "step": 3250
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.149074722684815e-06,
+      "loss": 0.5158,
+      "step": 3251
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.136188934169267e-06,
+      "loss": 0.5292,
+      "step": 3252
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.123309539675571e-06,
+      "loss": 0.5253,
+      "step": 3253
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.110436546784042e-06,
+      "loss": 0.5166,
+      "step": 3254
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.097569963071229e-06,
+      "loss": 0.5107,
+      "step": 3255
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.084709796109907e-06,
+      "loss": 0.5485,
+      "step": 3256
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.071856053469078e-06,
+      "loss": 0.5395,
+      "step": 3257
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.05900874271397e-06,
+      "loss": 0.5096,
+      "step": 3258
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.046167871406e-06,
+      "loss": 0.5037,
+      "step": 3259
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.033333447102817e-06,
+      "loss": 0.5265,
+      "step": 3260
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.020505477358272e-06,
+      "loss": 0.5164,
+      "step": 3261
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.007683969722419e-06,
+      "loss": 0.5156,
+      "step": 3262
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.994868931741499e-06,
+      "loss": 0.5264,
+      "step": 3263
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.982060370957953e-06,
+      "loss": 0.5251,
+      "step": 3264
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.969258294910405e-06,
+      "loss": 0.5231,
+      "step": 3265
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.956462711133672e-06,
+      "loss": 0.5277,
+      "step": 3266
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.94367362715874e-06,
+      "loss": 0.5114,
+      "step": 3267
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.93089105051277e-06,
+      "loss": 0.5106,
+      "step": 3268
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.918114988719098e-06,
+      "loss": 0.5062,
+      "step": 3269
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.905345449297223e-06,
+      "loss": 0.5199,
+      "step": 3270
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.892582439762802e-06,
+      "loss": 0.5208,
+      "step": 3271
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.879825967627667e-06,
+      "loss": 0.5379,
+      "step": 3272
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.867076040399765e-06,
+      "loss": 0.5346,
+      "step": 3273
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.854332665583223e-06,
+      "loss": 0.5026,
+      "step": 3274
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.841595850678304e-06,
+      "loss": 0.5194,
+      "step": 3275
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.828865603181407e-06,
+      "loss": 0.5313,
+      "step": 3276
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.816141930585067e-06,
+      "loss": 0.505,
+      "step": 3277
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.803424840377945e-06,
+      "loss": 0.5225,
+      "step": 3278
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.79071434004484e-06,
+      "loss": 0.5234,
+      "step": 3279
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.778010437066655e-06,
+      "loss": 0.5198,
+      "step": 3280
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.765313138920423e-06,
+      "loss": 0.5418,
+      "step": 3281
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.752622453079289e-06,
+      "loss": 0.5297,
+      "step": 3282
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.7399383870125e-06,
+      "loss": 0.506,
+      "step": 3283
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.72726094818541e-06,
+      "loss": 0.5468,
+      "step": 3284
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.714590144059478e-06,
+      "loss": 0.5135,
+      "step": 3285
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.701925982092243e-06,
+      "loss": 0.5255,
+      "step": 3286
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.689268469737348e-06,
+      "loss": 0.5155,
+      "step": 3287
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.676617614444518e-06,
+      "loss": 0.5228,
+      "step": 3288
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.663973423659564e-06,
+      "loss": 0.5495,
+      "step": 3289
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.651335904824367e-06,
+      "loss": 0.5161,
+      "step": 3290
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.638705065376887e-06,
+      "loss": 0.523,
+      "step": 3291
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.626080912751151e-06,
+      "loss": 0.5199,
+      "step": 3292
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.613463454377249e-06,
+      "loss": 0.5097,
+      "step": 3293
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.600852697681335e-06,
+      "loss": 0.5297,
+      "step": 3294
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.588248650085613e-06,
+      "loss": 0.5029,
+      "step": 3295
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.575651319008342e-06,
+      "loss": 0.5265,
+      "step": 3296
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.563060711863826e-06,
+      "loss": 0.5253,
+      "step": 3297
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.550476836062419e-06,
+      "loss": 0.5283,
+      "step": 3298
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.537899699010503e-06,
+      "loss": 0.518,
+      "step": 3299
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.525329308110493e-06,
+      "loss": 0.5382,
+      "step": 3300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.512765670760843e-06,
+      "loss": 0.519,
+      "step": 3301
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.500208794356025e-06,
+      "loss": 0.5304,
+      "step": 3302
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.487658686286533e-06,
+      "loss": 0.5223,
+      "step": 3303
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.475115353938882e-06,
+      "loss": 0.4963,
+      "step": 3304
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.462578804695595e-06,
+      "loss": 0.5205,
+      "step": 3305
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.450049045935204e-06,
+      "loss": 0.5243,
+      "step": 3306
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.437526085032241e-06,
+      "loss": 0.5382,
+      "step": 3307
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.42500992935724e-06,
+      "loss": 0.5285,
+      "step": 3308
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.412500586276732e-06,
+      "loss": 0.5347,
+      "step": 3309
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.399998063153234e-06,
+      "loss": 0.5238,
+      "step": 3310
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.387502367345248e-06,
+      "loss": 0.5134,
+      "step": 3311
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.375013506207275e-06,
+      "loss": 0.5124,
+      "step": 3312
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.362531487089757e-06,
+      "loss": 0.5023,
+      "step": 3313
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.350056317339143e-06,
+      "loss": 0.5084,
+      "step": 3314
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.337588004297837e-06,
+      "loss": 0.5345,
+      "step": 3315
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.325126555304208e-06,
+      "loss": 0.5129,
+      "step": 3316
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.312671977692589e-06,
+      "loss": 0.5059,
+      "step": 3317
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.30022427879326e-06,
+      "loss": 0.5209,
+      "step": 3318
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.287783465932466e-06,
+      "loss": 0.5192,
+      "step": 3319
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.275349546432387e-06,
+      "loss": 0.4901,
+      "step": 3320
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.262922527611153e-06,
+      "loss": 0.5066,
+      "step": 3321
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.250502416782828e-06,
+      "loss": 0.5023,
+      "step": 3322
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.23808922125741e-06,
+      "loss": 0.4989,
+      "step": 3323
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.225682948340834e-06,
+      "loss": 0.5221,
+      "step": 3324
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.213283605334959e-06,
+      "loss": 0.5178,
+      "step": 3325
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.200891199537549e-06,
+      "loss": 0.5244,
+      "step": 3326
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.188505738242306e-06,
+      "loss": 0.5185,
+      "step": 3327
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.176127228738835e-06,
+      "loss": 0.5237,
+      "step": 3328
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.163755678312651e-06,
+      "loss": 0.5176,
+      "step": 3329
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.151391094245175e-06,
+      "loss": 0.5314,
+      "step": 3330
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.139033483813725e-06,
+      "loss": 0.5269,
+      "step": 3331
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.126682854291517e-06,
+      "loss": 0.5338,
+      "step": 3332
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.114339212947655e-06,
+      "loss": 0.4914,
+      "step": 3333
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.102002567047133e-06,
+      "loss": 0.5212,
+      "step": 3334
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.08967292385083e-06,
+      "loss": 0.5262,
+      "step": 3335
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.077350290615495e-06,
+      "loss": 0.5541,
+      "step": 3336
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.065034674593761e-06,
+      "loss": 0.5307,
+      "step": 3337
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.052726083034124e-06,
+      "loss": 0.5522,
+      "step": 3338
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.040424523180948e-06,
+      "loss": 0.5146,
+      "step": 3339
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.028130002274459e-06,
+      "loss": 0.5303,
+      "step": 3340
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.01584252755074e-06,
+      "loss": 0.5235,
+      "step": 3341
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.003562106241727e-06,
+      "loss": 0.5143,
+      "step": 3342
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.991288745575203e-06,
+      "loss": 0.484,
+      "step": 3343
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.979022452774795e-06,
+      "loss": 0.5086,
+      "step": 3344
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.96676323505998e-06,
+      "loss": 0.5126,
+      "step": 3345
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.954511099646052e-06,
+      "loss": 0.5282,
+      "step": 3346
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.942266053744155e-06,
+      "loss": 0.5082,
+      "step": 3347
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.930028104561247e-06,
+      "loss": 0.519,
+      "step": 3348
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.917797259300123e-06,
+      "loss": 0.5211,
+      "step": 3349
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.905573525159381e-06,
+      "loss": 0.5113,
+      "step": 3350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.893356909333438e-06,
+      "loss": 0.5165,
+      "step": 3351
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.881147419012534e-06,
+      "loss": 0.5149,
+      "step": 3352
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.8689450613827e-06,
+      "loss": 0.51,
+      "step": 3353
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.856749843625777e-06,
+      "loss": 0.5122,
+      "step": 3354
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.844561772919402e-06,
+      "loss": 0.5391,
+      "step": 3355
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.832380856437005e-06,
+      "loss": 0.5196,
+      "step": 3356
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.820207101347803e-06,
+      "loss": 0.517,
+      "step": 3357
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.808040514816806e-06,
+      "loss": 0.5128,
+      "step": 3358
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.795881104004794e-06,
+      "loss": 0.5179,
+      "step": 3359
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.783728876068333e-06,
+      "loss": 0.5311,
+      "step": 3360
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.771583838159756e-06,
+      "loss": 0.5197,
+      "step": 3361
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.759445997427164e-06,
+      "loss": 0.5567,
+      "step": 3362
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.747315361014433e-06,
+      "loss": 0.507,
+      "step": 3363
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.735191936061176e-06,
+      "loss": 0.5325,
+      "step": 3364
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.723075729702781e-06,
+      "loss": 0.5163,
+      "step": 3365
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.710966749070382e-06,
+      "loss": 0.5306,
+      "step": 3366
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.698865001290859e-06,
+      "loss": 0.5094,
+      "step": 3367
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.686770493486835e-06,
+      "loss": 0.5434,
+      "step": 3368
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.674683232776676e-06,
+      "loss": 0.5168,
+      "step": 3369
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.662603226274476e-06,
+      "loss": 0.5101,
+      "step": 3370
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.650530481090066e-06,
+      "loss": 0.5075,
+      "step": 3371
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.638465004328998e-06,
+      "loss": 0.5305,
+      "step": 3372
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.626406803092547e-06,
+      "loss": 0.5211,
+      "step": 3373
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.614355884477711e-06,
+      "loss": 0.538,
+      "step": 3374
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.602312255577193e-06,
+      "loss": 0.4931,
+      "step": 3375
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.59027592347942e-06,
+      "loss": 0.5265,
+      "step": 3376
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.578246895268502e-06,
+      "loss": 0.5106,
+      "step": 3377
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.566225178024264e-06,
+      "loss": 0.5408,
+      "step": 3378
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.554210778822236e-06,
+      "loss": 0.5236,
+      "step": 3379
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.5422037047336214e-06,
+      "loss": 0.5077,
+      "step": 3380
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.530203962825331e-06,
+      "loss": 0.4921,
+      "step": 3381
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.518211560159949e-06,
+      "loss": 0.5347,
+      "step": 3382
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.506226503795744e-06,
+      "loss": 0.5351,
+      "step": 3383
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.494248800786657e-06,
+      "loss": 0.5168,
+      "step": 3384
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.482278458182306e-06,
+      "loss": 0.5491,
+      "step": 3385
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.470315483027975e-06,
+      "loss": 0.5479,
+      "step": 3386
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.45835988236461e-06,
+      "loss": 0.5341,
+      "step": 3387
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.446411663228818e-06,
+      "loss": 0.5274,
+      "step": 3388
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.434470832652865e-06,
+      "loss": 0.5083,
+      "step": 3389
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.422537397664671e-06,
+      "loss": 0.5368,
+      "step": 3390
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.410611365287781e-06,
+      "loss": 0.5302,
+      "step": 3391
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.398692742541409e-06,
+      "loss": 0.5208,
+      "step": 3392
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.386781536440397e-06,
+      "loss": 0.5312,
+      "step": 3393
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.374877753995224e-06,
+      "loss": 0.5281,
+      "step": 3394
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.362981402211999e-06,
+      "loss": 0.5227,
+      "step": 3395
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.3510924880924575e-06,
+      "loss": 0.5184,
+      "step": 3396
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.339211018633958e-06,
+      "loss": 0.5329,
+      "step": 3397
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.327337000829475e-06,
+      "loss": 0.5138,
+      "step": 3398
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.315470441667603e-06,
+      "loss": 0.5328,
+      "step": 3399
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.30361134813254e-06,
+      "loss": 0.499,
+      "step": 3400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.2917597272040955e-06,
+      "loss": 0.53,
+      "step": 3401
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.2799155858576756e-06,
+      "loss": 0.5097,
+      "step": 3402
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.268078931064293e-06,
+      "loss": 0.5151,
+      "step": 3403
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.256249769790537e-06,
+      "loss": 0.5039,
+      "step": 3404
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.244428108998602e-06,
+      "loss": 0.5516,
+      "step": 3405
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.232613955646268e-06,
+      "loss": 0.52,
+      "step": 3406
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.220807316686886e-06,
+      "loss": 0.5018,
+      "step": 3407
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.209008199069392e-06,
+      "loss": 0.5279,
+      "step": 3408
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.197216609738296e-06,
+      "loss": 0.5276,
+      "step": 3409
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.185432555633672e-06,
+      "loss": 0.5207,
+      "step": 3410
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.17365604369116e-06,
+      "loss": 0.5106,
+      "step": 3411
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.161887080841967e-06,
+      "loss": 0.5572,
+      "step": 3412
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.150125674012849e-06,
+      "loss": 0.5493,
+      "step": 3413
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.138371830126119e-06,
+      "loss": 0.5244,
+      "step": 3414
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.126625556099638e-06,
+      "loss": 0.5146,
+      "step": 3415
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.1148868588468125e-06,
+      "loss": 0.5073,
+      "step": 3416
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.1031557452765934e-06,
+      "loss": 0.5329,
+      "step": 3417
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.091432222293453e-06,
+      "loss": 0.5218,
+      "step": 3418
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.0797162967974095e-06,
+      "loss": 0.5037,
+      "step": 3419
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.06800797568401e-06,
+      "loss": 0.5217,
+      "step": 3420
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.0563072658443225e-06,
+      "loss": 0.5027,
+      "step": 3421
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.044614174164932e-06,
+      "loss": 0.5395,
+      "step": 3422
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.032928707527948e-06,
+      "loss": 0.5384,
+      "step": 3423
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.021250872810983e-06,
+      "loss": 0.5188,
+      "step": 3424
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.009580676887162e-06,
+      "loss": 0.4872,
+      "step": 3425
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.9979181266251165e-06,
+      "loss": 0.5334,
+      "step": 3426
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.986263228888974e-06,
+      "loss": 0.5314,
+      "step": 3427
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.9746159905383605e-06,
+      "loss": 0.497,
+      "step": 3428
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.9629764184283936e-06,
+      "loss": 0.5055,
+      "step": 3429
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.951344519409682e-06,
+      "loss": 0.5326,
+      "step": 3430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.939720300328303e-06,
+      "loss": 0.5295,
+      "step": 3431
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.9281037680258336e-06,
+      "loss": 0.5219,
+      "step": 3432
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.916494929339315e-06,
+      "loss": 0.5262,
+      "step": 3433
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.9048937911012685e-06,
+      "loss": 0.5203,
+      "step": 3434
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.893300360139674e-06,
+      "loss": 0.5204,
+      "step": 3435
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.881714643277981e-06,
+      "loss": 0.5072,
+      "step": 3436
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.870136647335097e-06,
+      "loss": 0.5411,
+      "step": 3437
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.858566379125389e-06,
+      "loss": 0.5138,
+      "step": 3438
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.847003845458666e-06,
+      "loss": 0.519,
+      "step": 3439
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.835449053140197e-06,
+      "loss": 0.517,
+      "step": 3440
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.823902008970687e-06,
+      "loss": 0.5093,
+      "step": 3441
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.812362719746282e-06,
+      "loss": 0.5105,
+      "step": 3442
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.80083119225857e-06,
+      "loss": 0.5395,
+      "step": 3443
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.789307433294554e-06,
+      "loss": 0.5239,
+      "step": 3444
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.777791449636681e-06,
+      "loss": 0.5373,
+      "step": 3445
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.766283248062817e-06,
+      "loss": 0.5102,
+      "step": 3446
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.754782835346247e-06,
+      "loss": 0.5241,
+      "step": 3447
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.743290218255674e-06,
+      "loss": 0.5205,
+      "step": 3448
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.73180540355521e-06,
+      "loss": 0.5043,
+      "step": 3449
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.720328398004372e-06,
+      "loss": 0.5078,
+      "step": 3450
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.708859208358091e-06,
+      "loss": 0.4946,
+      "step": 3451
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.697397841366686e-06,
+      "loss": 0.5309,
+      "step": 3452
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.685944303775882e-06,
+      "loss": 0.5374,
+      "step": 3453
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.674498602326786e-06,
+      "loss": 0.5095,
+      "step": 3454
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.6630607437559026e-06,
+      "loss": 0.5035,
+      "step": 3455
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.651630734795114e-06,
+      "loss": 0.5118,
+      "step": 3456
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.64020858217169e-06,
+      "loss": 0.521,
+      "step": 3457
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.62879429260826e-06,
+      "loss": 0.522,
+      "step": 3458
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.617387872822842e-06,
+      "loss": 0.5167,
+      "step": 3459
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.605989329528817e-06,
+      "loss": 0.5391,
+      "step": 3460
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.5945986694349265e-06,
+      "loss": 0.5321,
+      "step": 3461
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.583215899245281e-06,
+      "loss": 0.5134,
+      "step": 3462
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.571841025659336e-06,
+      "loss": 0.5299,
+      "step": 3463
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.560474055371904e-06,
+      "loss": 0.505,
+      "step": 3464
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.549114995073156e-06,
+      "loss": 0.5444,
+      "step": 3465
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.537763851448593e-06,
+      "loss": 0.5021,
+      "step": 3466
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.5264206311790686e-06,
+      "loss": 0.5175,
+      "step": 3467
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.515085340940754e-06,
+      "loss": 0.4946,
+      "step": 3468
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.503757987405173e-06,
+      "loss": 0.5238,
+      "step": 3469
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.49243857723917e-06,
+      "loss": 0.4855,
+      "step": 3470
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.481127117104913e-06,
+      "loss": 0.5179,
+      "step": 3471
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.469823613659896e-06,
+      "loss": 0.5342,
+      "step": 3472
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.458528073556925e-06,
+      "loss": 0.5161,
+      "step": 3473
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.447240503444121e-06,
+      "loss": 0.5279,
+      "step": 3474
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.435960909964916e-06,
+      "loss": 0.5174,
+      "step": 3475
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.424689299758043e-06,
+      "loss": 0.5301,
+      "step": 3476
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.413425679457539e-06,
+      "loss": 0.5042,
+      "step": 3477
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.402170055692741e-06,
+      "loss": 0.5156,
+      "step": 3478
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.390922435088274e-06,
+      "loss": 0.54,
+      "step": 3479
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.379682824264055e-06,
+      "loss": 0.5291,
+      "step": 3480
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.368451229835295e-06,
+      "loss": 0.5184,
+      "step": 3481
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.357227658412466e-06,
+      "loss": 0.5156,
+      "step": 3482
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.346012116601337e-06,
+      "loss": 0.5291,
+      "step": 3483
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.334804611002945e-06,
+      "loss": 0.5255,
+      "step": 3484
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.323605148213598e-06,
+      "loss": 0.5325,
+      "step": 3485
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.3124137348248694e-06,
+      "loss": 0.5181,
+      "step": 3486
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.301230377423595e-06,
+      "loss": 0.517,
+      "step": 3487
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.290055082591868e-06,
+      "loss": 0.5035,
+      "step": 3488
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.278887856907041e-06,
+      "loss": 0.5195,
+      "step": 3489
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.267728706941709e-06,
+      "loss": 0.5102,
+      "step": 3490
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.256577639263723e-06,
+      "loss": 0.5415,
+      "step": 3491
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.245434660436174e-06,
+      "loss": 0.51,
+      "step": 3492
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.234299777017387e-06,
+      "loss": 0.5296,
+      "step": 3493
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.223172995560935e-06,
+      "loss": 0.506,
+      "step": 3494
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.2120543226156035e-06,
+      "loss": 0.5385,
+      "step": 3495
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.20094376472542e-06,
+      "loss": 0.4943,
+      "step": 3496
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.189841328429631e-06,
+      "loss": 0.5155,
+      "step": 3497
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.178747020262708e-06,
+      "loss": 0.4956,
+      "step": 3498
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.167660846754331e-06,
+      "loss": 0.5194,
+      "step": 3499
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.156582814429399e-06,
+      "loss": 0.5048,
+      "step": 3500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.145512929808013e-06,
+      "loss": 0.5342,
+      "step": 3501
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.134451199405482e-06,
+      "loss": 0.5269,
+      "step": 3502
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.1233976297323196e-06,
+      "loss": 0.5139,
+      "step": 3503
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.112352227294227e-06,
+      "loss": 0.511,
+      "step": 3504
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.101314998592107e-06,
+      "loss": 0.5377,
+      "step": 3505
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.09028595012205e-06,
+      "loss": 0.5127,
+      "step": 3506
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.079265088375325e-06,
+      "loss": 0.5397,
+      "step": 3507
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.068252419838399e-06,
+      "loss": 0.5409,
+      "step": 3508
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.05724795099289e-06,
+      "loss": 0.5312,
+      "step": 3509
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.046251688315612e-06,
+      "loss": 0.5127,
+      "step": 3510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.035263638278546e-06,
+      "loss": 0.5386,
+      "step": 3511
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.024283807348832e-06,
+      "loss": 0.4967,
+      "step": 3512
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.0133122019887816e-06,
+      "loss": 0.5079,
+      "step": 3513
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.002348828655853e-06,
+      "loss": 0.5338,
+      "step": 3514
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.991393693802674e-06,
+      "loss": 0.5118,
+      "step": 3515
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.980446803877011e-06,
+      "loss": 0.5294,
+      "step": 3516
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.969508165321787e-06,
+      "loss": 0.5259,
+      "step": 3517
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.9585777845750616e-06,
+      "loss": 0.5174,
+      "step": 3518
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.947655668070036e-06,
+      "loss": 0.5096,
+      "step": 3519
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.9367418222350505e-06,
+      "loss": 0.5134,
+      "step": 3520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.925836253493582e-06,
+      "loss": 0.5026,
+      "step": 3521
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.9149389682642165e-06,
+      "loss": 0.5191,
+      "step": 3522
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.904049972960684e-06,
+      "loss": 0.521,
+      "step": 3523
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.893169273991825e-06,
+      "loss": 0.494,
+      "step": 3524
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.882296877761606e-06,
+      "loss": 0.5337,
+      "step": 3525
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.8714327906691e-06,
+      "loss": 0.5423,
+      "step": 3526
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.860577019108493e-06,
+      "loss": 0.4834,
+      "step": 3527
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.849729569469069e-06,
+      "loss": 0.5144,
+      "step": 3528
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.838890448135228e-06,
+      "loss": 0.5099,
+      "step": 3529
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.828059661486456e-06,
+      "loss": 0.511,
+      "step": 3530
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.817237215897338e-06,
+      "loss": 0.5218,
+      "step": 3531
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.80642311773755e-06,
+      "loss": 0.5361,
+      "step": 3532
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.795617373371855e-06,
+      "loss": 0.5202,
+      "step": 3533
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.784819989160107e-06,
+      "loss": 0.5097,
+      "step": 3534
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7740309714572164e-06,
+      "loss": 0.5128,
+      "step": 3535
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7632503266131925e-06,
+      "loss": 0.5017,
+      "step": 3536
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.752478060973108e-06,
+      "loss": 0.5273,
+      "step": 3537
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.741714180877105e-06,
+      "loss": 0.5075,
+      "step": 3538
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.730958692660389e-06,
+      "loss": 0.5388,
+      "step": 3539
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720211602653227e-06,
+      "loss": 0.5388,
+      "step": 3540
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.709472917180945e-06,
+      "loss": 0.5223,
+      "step": 3541
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.698742642563921e-06,
+      "loss": 0.5327,
+      "step": 3542
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.688020785117581e-06,
+      "loss": 0.5099,
+      "step": 3543
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.677307351152397e-06,
+      "loss": 0.5106,
+      "step": 3544
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.666602346973886e-06,
+      "loss": 0.5479,
+      "step": 3545
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.655905778882604e-06,
+      "loss": 0.5063,
+      "step": 3546
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.6452176531741374e-06,
+      "loss": 0.5207,
+      "step": 3547
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.634537976139116e-06,
+      "loss": 0.5275,
+      "step": 3548
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.62386675406317e-06,
+      "loss": 0.5353,
+      "step": 3549
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.613203993226981e-06,
+      "loss": 0.5098,
+      "step": 3550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.602549699906239e-06,
+      "loss": 0.5334,
+      "step": 3551
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.591903880371654e-06,
+      "loss": 0.5179,
+      "step": 3552
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.581266540888941e-06,
+      "loss": 0.5095,
+      "step": 3553
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.570637687718836e-06,
+      "loss": 0.4867,
+      "step": 3554
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.560017327117068e-06,
+      "loss": 0.5265,
+      "step": 3555
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.549405465334377e-06,
+      "loss": 0.5056,
+      "step": 3556
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.538802108616494e-06,
+      "loss": 0.5156,
+      "step": 3557
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.528207263204151e-06,
+      "loss": 0.5318,
+      "step": 3558
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.517620935333063e-06,
+      "loss": 0.5106,
+      "step": 3559
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.507043131233943e-06,
+      "loss": 0.5229,
+      "step": 3560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.496473857132476e-06,
+      "loss": 0.5173,
+      "step": 3561
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.485913119249327e-06,
+      "loss": 0.527,
+      "step": 3562
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.475360923800141e-06,
+      "loss": 0.5283,
+      "step": 3563
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.46481727699554e-06,
+      "loss": 0.538,
+      "step": 3564
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.454282185041106e-06,
+      "loss": 0.5305,
+      "step": 3565
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.44375565413739e-06,
+      "loss": 0.5154,
+      "step": 3566
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.433237690479907e-06,
+      "loss": 0.5359,
+      "step": 3567
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.422728300259119e-06,
+      "loss": 0.5357,
+      "step": 3568
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.412227489660455e-06,
+      "loss": 0.5271,
+      "step": 3569
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.40173526486429e-06,
+      "loss": 0.5089,
+      "step": 3570
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.39125163204594e-06,
+      "loss": 0.5088,
+      "step": 3571
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.38077659737567e-06,
+      "loss": 0.5145,
+      "step": 3572
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.370310167018684e-06,
+      "loss": 0.5008,
+      "step": 3573
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.359852347135121e-06,
+      "loss": 0.5252,
+      "step": 3574
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.349403143880056e-06,
+      "loss": 0.5268,
+      "step": 3575
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.338962563403478e-06,
+      "loss": 0.519,
+      "step": 3576
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.32853061185032e-06,
+      "loss": 0.5294,
+      "step": 3577
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.318107295360424e-06,
+      "loss": 0.5163,
+      "step": 3578
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.3076926200685545e-06,
+      "loss": 0.5328,
+      "step": 3579
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.297286592104389e-06,
+      "loss": 0.5073,
+      "step": 3580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.286889217592519e-06,
+      "loss": 0.5142,
+      "step": 3581
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.276500502652435e-06,
+      "loss": 0.5134,
+      "step": 3582
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.26612045339854e-06,
+      "loss": 0.5216,
+      "step": 3583
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.255749075940128e-06,
+      "loss": 0.522,
+      "step": 3584
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.245386376381398e-06,
+      "loss": 0.5147,
+      "step": 3585
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.235032360821435e-06,
+      "loss": 0.5047,
+      "step": 3586
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.224687035354212e-06,
+      "loss": 0.517,
+      "step": 3587
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.214350406068598e-06,
+      "loss": 0.5392,
+      "step": 3588
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.204022479048325e-06,
+      "loss": 0.5309,
+      "step": 3589
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.1937032603720206e-06,
+      "loss": 0.5209,
+      "step": 3590
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.1833927561131795e-06,
+      "loss": 0.4901,
+      "step": 3591
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.17309097234016e-06,
+      "loss": 0.5213,
+      "step": 3592
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.162797915116211e-06,
+      "loss": 0.5184,
+      "step": 3593
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.152513590499424e-06,
+      "loss": 0.4992,
+      "step": 3594
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.142238004542759e-06,
+      "loss": 0.4933,
+      "step": 3595
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.131971163294027e-06,
+      "loss": 0.519,
+      "step": 3596
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.1217130727959e-06,
+      "loss": 0.5173,
+      "step": 3597
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.111463739085896e-06,
+      "loss": 0.5182,
+      "step": 3598
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.101223168196381e-06,
+      "loss": 0.5146,
+      "step": 3599
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.090991366154554e-06,
+      "loss": 0.5038,
+      "step": 3600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.0807683389824666e-06,
+      "loss": 0.5347,
+      "step": 3601
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.070554092696997e-06,
+      "loss": 0.516,
+      "step": 3602
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.060348633309862e-06,
+      "loss": 0.5052,
+      "step": 3603
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.050151966827597e-06,
+      "loss": 0.5213,
+      "step": 3604
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.039964099251573e-06,
+      "loss": 0.5004,
+      "step": 3605
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.029785036577976e-06,
+      "loss": 0.5213,
+      "step": 3606
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.019614784797812e-06,
+      "loss": 0.5251,
+      "step": 3607
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.009453349896902e-06,
+      "loss": 0.5129,
+      "step": 3608
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.999300737855872e-06,
+      "loss": 0.5081,
+      "step": 3609
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.989156954650163e-06,
+      "loss": 0.5332,
+      "step": 3610
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.979022006250012e-06,
+      "loss": 0.4948,
+      "step": 3611
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.968895898620471e-06,
+      "loss": 0.5307,
+      "step": 3612
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.958778637721364e-06,
+      "loss": 0.4863,
+      "step": 3613
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.948670229507326e-06,
+      "loss": 0.5152,
+      "step": 3614
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.938570679927784e-06,
+      "loss": 0.5198,
+      "step": 3615
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.928479994926938e-06,
+      "loss": 0.5028,
+      "step": 3616
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.91839818044378e-06,
+      "loss": 0.5015,
+      "step": 3617
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.90832524241208e-06,
+      "loss": 0.5215,
+      "step": 3618
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.89826118676038e-06,
+      "loss": 0.5211,
+      "step": 3619
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8882060194119985e-06,
+      "loss": 0.5391,
+      "step": 3620
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.87815974628502e-06,
+      "loss": 0.5139,
+      "step": 3621
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8681223732922964e-06,
+      "loss": 0.4894,
+      "step": 3622
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.858093906341437e-06,
+      "loss": 0.5114,
+      "step": 3623
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.848074351334813e-06,
+      "loss": 0.4918,
+      "step": 3624
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.838063714169552e-06,
+      "loss": 0.5255,
+      "step": 3625
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.828062000737534e-06,
+      "loss": 0.5161,
+      "step": 3626
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8180692169253714e-06,
+      "loss": 0.5151,
+      "step": 3627
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.808085368614441e-06,
+      "loss": 0.5067,
+      "step": 3628
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.79811046168085e-06,
+      "loss": 0.5199,
+      "step": 3629
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.7881445019954445e-06,
+      "loss": 0.5354,
+      "step": 3630
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.778187495423809e-06,
+      "loss": 0.527,
+      "step": 3631
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.768239447826253e-06,
+      "loss": 0.5219,
+      "step": 3632
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.758300365057818e-06,
+      "loss": 0.491,
+      "step": 3633
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.74837025296826e-06,
+      "loss": 0.5011,
+      "step": 3634
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.738449117402067e-06,
+      "loss": 0.4966,
+      "step": 3635
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.728536964198436e-06,
+      "loss": 0.4988,
+      "step": 3636
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.7186337991912816e-06,
+      "loss": 0.5055,
+      "step": 3637
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.708739628209223e-06,
+      "loss": 0.4964,
+      "step": 3638
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.698854457075597e-06,
+      "loss": 0.5421,
+      "step": 3639
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.688978291608423e-06,
+      "loss": 0.5144,
+      "step": 3640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.5113,
+      "step": 3641
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.669253000919078e-06,
+      "loss": 0.5023,
+      "step": 3642
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.659403887306451e-06,
+      "loss": 0.5042,
+      "step": 3643
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.649563802579373e-06,
+      "loss": 0.5163,
+      "step": 3644
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.63973275252934e-06,
+      "loss": 0.4929,
+      "step": 3645
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.62991074294253e-06,
+      "loss": 0.521,
+      "step": 3646
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.620097779599799e-06,
+      "loss": 0.5015,
+      "step": 3647
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.610293868276681e-06,
+      "loss": 0.5248,
+      "step": 3648
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.600499014743383e-06,
+      "loss": 0.5396,
+      "step": 3649
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.590713224764778e-06,
+      "loss": 0.5133,
+      "step": 3650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.580936504100408e-06,
+      "loss": 0.5056,
+      "step": 3651
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5711688585044774e-06,
+      "loss": 0.4908,
+      "step": 3652
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.561410293725839e-06,
+      "loss": 0.5146,
+      "step": 3653
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.551660815508012e-06,
+      "loss": 0.5185,
+      "step": 3654
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.541920429589168e-06,
+      "loss": 0.5364,
+      "step": 3655
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.532189141702123e-06,
+      "loss": 0.5361,
+      "step": 3656
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.522466957574338e-06,
+      "loss": 0.5154,
+      "step": 3657
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.512753882927916e-06,
+      "loss": 0.5331,
+      "step": 3658
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5030499234796035e-06,
+      "loss": 0.5411,
+      "step": 3659
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.493355084940776e-06,
+      "loss": 0.5245,
+      "step": 3660
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.483669373017441e-06,
+      "loss": 0.4992,
+      "step": 3661
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.47399279341024e-06,
+      "loss": 0.5283,
+      "step": 3662
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.464325351814434e-06,
+      "loss": 0.5242,
+      "step": 3663
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.45466705391991e-06,
+      "loss": 0.514,
+      "step": 3664
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.4450179054111666e-06,
+      "loss": 0.4997,
+      "step": 3665
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.435377911967331e-06,
+      "loss": 0.5287,
+      "step": 3666
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.425747079262121e-06,
+      "loss": 0.5176,
+      "step": 3667
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.416125412963883e-06,
+      "loss": 0.5178,
+      "step": 3668
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.406512918735555e-06,
+      "loss": 0.5312,
+      "step": 3669
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.396909602234685e-06,
+      "loss": 0.5091,
+      "step": 3670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.387315469113416e-06,
+      "loss": 0.5268,
+      "step": 3671
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.377730525018487e-06,
+      "loss": 0.5119,
+      "step": 3672
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.368154775591225e-06,
+      "loss": 0.5387,
+      "step": 3673
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.358588226467551e-06,
+      "loss": 0.4843,
+      "step": 3674
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.3490308832779646e-06,
+      "loss": 0.5328,
+      "step": 3675
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.339482751647557e-06,
+      "loss": 0.5274,
+      "step": 3676
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.329943837195987e-06,
+      "loss": 0.5083,
+      "step": 3677
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.3204141455374925e-06,
+      "loss": 0.528,
+      "step": 3678
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.3108936822808925e-06,
+      "loss": 0.5371,
+      "step": 3679
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.3013824530295545e-06,
+      "loss": 0.5316,
+      "step": 3680
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.291880463381426e-06,
+      "loss": 0.5018,
+      "step": 3681
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.282387718929015e-06,
+      "loss": 0.5243,
+      "step": 3682
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.272904225259387e-06,
+      "loss": 0.544,
+      "step": 3683
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.26342998795416e-06,
+      "loss": 0.51,
+      "step": 3684
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.253965012589507e-06,
+      "loss": 0.5344,
+      "step": 3685
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.244509304736148e-06,
+      "loss": 0.534,
+      "step": 3686
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.23506286995935e-06,
+      "loss": 0.5068,
+      "step": 3687
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.225625713818922e-06,
+      "loss": 0.4998,
+      "step": 3688
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.216197841869212e-06,
+      "loss": 0.5238,
+      "step": 3689
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.206779259659102e-06,
+      "loss": 0.5106,
+      "step": 3690
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.197369972732008e-06,
+      "loss": 0.5035,
+      "step": 3691
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.187969986625875e-06,
+      "loss": 0.5353,
+      "step": 3692
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.17857930687318e-06,
+      "loss": 0.5305,
+      "step": 3693
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.169197939000904e-06,
+      "loss": 0.5157,
+      "step": 3694
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.1598258885305665e-06,
+      "loss": 0.5243,
+      "step": 3695
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.150463160978196e-06,
+      "loss": 0.5166,
+      "step": 3696
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.141109761854332e-06,
+      "loss": 0.5245,
+      "step": 3697
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.13176569666403e-06,
+      "loss": 0.5335,
+      "step": 3698
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.122430970906845e-06,
+      "loss": 0.5354,
+      "step": 3699
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.1131055900768375e-06,
+      "loss": 0.5195,
+      "step": 3700
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.103789559662572e-06,
+      "loss": 0.5039,
+      "step": 3701
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.094482885147102e-06,
+      "loss": 0.5247,
+      "step": 3702
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.085185572007979e-06,
+      "loss": 0.532,
+      "step": 3703
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.075897625717249e-06,
+      "loss": 0.5109,
+      "step": 3704
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.066619051741434e-06,
+      "loss": 0.5114,
+      "step": 3705
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.057349855541557e-06,
+      "loss": 0.4962,
+      "step": 3706
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.048090042573096e-06,
+      "loss": 0.4891,
+      "step": 3707
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.0388396182860275e-06,
+      "loss": 0.5168,
+      "step": 3708
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.029598588124799e-06,
+      "loss": 0.5167,
+      "step": 3709
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.020366957528321e-06,
+      "loss": 0.5099,
+      "step": 3710
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.011144731929981e-06,
+      "loss": 0.5051,
+      "step": 3711
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.001931916757624e-06,
+      "loss": 0.5153,
+      "step": 3712
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.99272851743356e-06,
+      "loss": 0.501,
+      "step": 3713
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.983534539374556e-06,
+      "loss": 0.5323,
+      "step": 3714
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.9743499879918345e-06,
+      "loss": 0.5298,
+      "step": 3715
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.965174868691068e-06,
+      "loss": 0.5193,
+      "step": 3716
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.956009186872383e-06,
+      "loss": 0.5295,
+      "step": 3717
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.9468529479303445e-06,
+      "loss": 0.5239,
+      "step": 3718
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.937706157253971e-06,
+      "loss": 0.496,
+      "step": 3719
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.928568820226692e-06,
+      "loss": 0.5343,
+      "step": 3720
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.919440942226413e-06,
+      "loss": 0.4831,
+      "step": 3721
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.910322528625446e-06,
+      "loss": 0.4994,
+      "step": 3722
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.901213584790535e-06,
+      "loss": 0.507,
+      "step": 3723
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.892114116082861e-06,
+      "loss": 0.5021,
+      "step": 3724
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.883024127858017e-06,
+      "loss": 0.5047,
+      "step": 3725
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.873943625466024e-06,
+      "loss": 0.5163,
+      "step": 3726
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.864872614251316e-06,
+      "loss": 0.5149,
+      "step": 3727
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.85581109955274e-06,
+      "loss": 0.5152,
+      "step": 3728
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.846759086703555e-06,
+      "loss": 0.5457,
+      "step": 3729
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.837716581031436e-06,
+      "loss": 0.4915,
+      "step": 3730
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.828683587858442e-06,
+      "loss": 0.5272,
+      "step": 3731
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.819660112501053e-06,
+      "loss": 0.5136,
+      "step": 3732
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.810646160270137e-06,
+      "loss": 0.527,
+      "step": 3733
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8016417364709647e-06,
+      "loss": 0.52,
+      "step": 3734
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7926468464031875e-06,
+      "loss": 0.5082,
+      "step": 3735
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7836614953608573e-06,
+      "loss": 0.5054,
+      "step": 3736
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.774685688632407e-06,
+      "loss": 0.5286,
+      "step": 3737
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7657194315006474e-06,
+      "loss": 0.5357,
+      "step": 3738
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.756762729242773e-06,
+      "loss": 0.4988,
+      "step": 3739
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.747815587130361e-06,
+      "loss": 0.5212,
+      "step": 3740
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7388780104293474e-06,
+      "loss": 0.5013,
+      "step": 3741
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7299500044000534e-06,
+      "loss": 0.5099,
+      "step": 3742
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7210315742971625e-06,
+      "loss": 0.5343,
+      "step": 3743
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.712122725369709e-06,
+      "loss": 0.5184,
+      "step": 3744
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7032234628611053e-06,
+      "loss": 0.5239,
+      "step": 3745
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.694333792009115e-06,
+      "loss": 0.521,
+      "step": 3746
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6854537180458572e-06,
+      "loss": 0.5014,
+      "step": 3747
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6765832461978e-06,
+      "loss": 0.5265,
+      "step": 3748
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.667722381685768e-06,
+      "loss": 0.542,
+      "step": 3749
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6588711297249213e-06,
+      "loss": 0.5116,
+      "step": 3750
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.650029495524767e-06,
+      "loss": 0.5354,
+      "step": 3751
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.641197484289152e-06,
+      "loss": 0.5121,
+      "step": 3752
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.632375101216259e-06,
+      "loss": 0.5092,
+      "step": 3753
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.623562351498604e-06,
+      "loss": 0.5223,
+      "step": 3754
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6147592403230334e-06,
+      "loss": 0.5114,
+      "step": 3755
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6059657728707166e-06,
+      "loss": 0.5184,
+      "step": 3756
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5971819543171613e-06,
+      "loss": 0.4949,
+      "step": 3757
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5884077898321713e-06,
+      "loss": 0.5031,
+      "step": 3758
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.57964328457989e-06,
+      "loss": 0.4877,
+      "step": 3759
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5708884437187673e-06,
+      "loss": 0.5163,
+      "step": 3760
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.562143272401568e-06,
+      "loss": 0.5231,
+      "step": 3761
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5534077757753618e-06,
+      "loss": 0.5144,
+      "step": 3762
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5446819589815283e-06,
+      "loss": 0.5193,
+      "step": 3763
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.535965827155745e-06,
+      "loss": 0.542,
+      "step": 3764
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.527259385427997e-06,
+      "loss": 0.5115,
+      "step": 3765
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.518562638922556e-06,
+      "loss": 0.5148,
+      "step": 3766
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.509875592757999e-06,
+      "loss": 0.5106,
+      "step": 3767
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.501198252047182e-06,
+      "loss": 0.5108,
+      "step": 3768
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4925306218972565e-06,
+      "loss": 0.5291,
+      "step": 3769
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4838727074096633e-06,
+      "loss": 0.499,
+      "step": 3770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4752245136801065e-06,
+      "loss": 0.5234,
+      "step": 3771
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4665860457985857e-06,
+      "loss": 0.5376,
+      "step": 3772
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.457957308849371e-06,
+      "loss": 0.5145,
+      "step": 3773
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4493383079110054e-06,
+      "loss": 0.5168,
+      "step": 3774
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.440729048056304e-06,
+      "loss": 0.5078,
+      "step": 3775
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.432129534352344e-06,
+      "loss": 0.5122,
+      "step": 3776
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4235397718604714e-06,
+      "loss": 0.5323,
+      "step": 3777
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4149597656362877e-06,
+      "loss": 0.5309,
+      "step": 3778
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.406389520729656e-06,
+      "loss": 0.5335,
+      "step": 3779
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3978290421846926e-06,
+      "loss": 0.5032,
+      "step": 3780
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3892783350397675e-06,
+      "loss": 0.5137,
+      "step": 3781
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3807374043274966e-06,
+      "loss": 0.5405,
+      "step": 3782
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.372206255074746e-06,
+      "loss": 0.5046,
+      "step": 3783
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3636848923026257e-06,
+      "loss": 0.5231,
+      "step": 3784
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3551733210264725e-06,
+      "loss": 0.5274,
+      "step": 3785
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.346671546255875e-06,
+      "loss": 0.5243,
+      "step": 3786
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.338179572994653e-06,
+      "loss": 0.5063,
+      "step": 3787
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.329697406240855e-06,
+      "loss": 0.5034,
+      "step": 3788
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3212250509867583e-06,
+      "loss": 0.482,
+      "step": 3789
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3127625122188655e-06,
+      "loss": 0.5116,
+      "step": 3790
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3043097949179017e-06,
+      "loss": 0.5161,
+      "step": 3791
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.295866904058813e-06,
+      "loss": 0.4975,
+      "step": 3792
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.287433844610759e-06,
+      "loss": 0.5258,
+      "step": 3793
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.279010621537113e-06,
+      "loss": 0.5335,
+      "step": 3794
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.2705972397954655e-06,
+      "loss": 0.5187,
+      "step": 3795
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.262193704337604e-06,
+      "loss": 0.5359,
+      "step": 3796
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.2538000201095363e-06,
+      "loss": 0.5268,
+      "step": 3797
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.2454161920514494e-06,
+      "loss": 0.5264,
+      "step": 3798
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.237042225097751e-06,
+      "loss": 0.5005,
+      "step": 3799
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.2286781241770315e-06,
+      "loss": 0.5103,
+      "step": 3800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.220323894212083e-06,
+      "loss": 0.5192,
+      "step": 3801
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.211979540119883e-06,
+      "loss": 0.5255,
+      "step": 3802
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.203645066811598e-06,
+      "loss": 0.5116,
+      "step": 3803
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1953204791925784e-06,
+      "loss": 0.5173,
+      "step": 3804
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1870057821623577e-06,
+      "loss": 0.5043,
+      "step": 3805
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1787009806146464e-06,
+      "loss": 0.5103,
+      "step": 3806
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1704060794373315e-06,
+      "loss": 0.5296,
+      "step": 3807
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.162121083512473e-06,
+      "loss": 0.5167,
+      "step": 3808
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.153845997716303e-06,
+      "loss": 0.4989,
+      "step": 3809
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.145580826919217e-06,
+      "loss": 0.5201,
+      "step": 3810
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1373255759857834e-06,
+      "loss": 0.5385,
+      "step": 3811
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.129080249774714e-06,
+      "loss": 0.5243,
+      "step": 3812
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.120844853138898e-06,
+      "loss": 0.5357,
+      "step": 3813
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.112619390925371e-06,
+      "loss": 0.5138,
+      "step": 3814
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1044038679753276e-06,
+      "loss": 0.4932,
+      "step": 3815
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0961982891241083e-06,
+      "loss": 0.521,
+      "step": 3816
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.088002659201199e-06,
+      "loss": 0.5119,
+      "step": 3817
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0798169830302327e-06,
+      "loss": 0.5235,
+      "step": 3818
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0716412654289883e-06,
+      "loss": 0.501,
+      "step": 3819
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0634755112093728e-06,
+      "loss": 0.5285,
+      "step": 3820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.055319725177441e-06,
+      "loss": 0.5271,
+      "step": 3821
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.047173912133372e-06,
+      "loss": 0.5132,
+      "step": 3822
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.039038076871481e-06,
+      "loss": 0.5225,
+      "step": 3823
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.030912224180209e-06,
+      "loss": 0.5149,
+      "step": 3824
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.022796358842115e-06,
+      "loss": 0.522,
+      "step": 3825
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0146904856338888e-06,
+      "loss": 0.499,
+      "step": 3826
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0065946093263364e-06,
+      "loss": 0.5099,
+      "step": 3827
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.998508734684378e-06,
+      "loss": 0.5122,
+      "step": 3828
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.99043286646705e-06,
+      "loss": 0.5268,
+      "step": 3829
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.9823670094275e-06,
+      "loss": 0.5081,
+      "step": 3830
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9743111683129777e-06,
+      "loss": 0.5119,
+      "step": 3831
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9662653478648453e-06,
+      "loss": 0.5343,
+      "step": 3832
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.958229552818561e-06,
+      "loss": 0.5245,
+      "step": 3833
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.950203787903687e-06,
+      "loss": 0.5134,
+      "step": 3834
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.942188057843878e-06,
+      "loss": 0.513,
+      "step": 3835
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.934182367356888e-06,
+      "loss": 0.5311,
+      "step": 3836
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9261867211545603e-06,
+      "loss": 0.4993,
+      "step": 3837
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9182011239428165e-06,
+      "loss": 0.5241,
+      "step": 3838
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9102255804216794e-06,
+      "loss": 0.5158,
+      "step": 3839
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9022600952852455e-06,
+      "loss": 0.5281,
+      "step": 3840
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.894304673221693e-06,
+      "loss": 0.4853,
+      "step": 3841
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.886359318913279e-06,
+      "loss": 0.5452,
+      "step": 3842
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.8784240370363315e-06,
+      "loss": 0.5183,
+      "step": 3843
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.870498832261257e-06,
+      "loss": 0.5079,
+      "step": 3844
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.862583709252522e-06,
+      "loss": 0.5215,
+      "step": 3845
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.854678672668665e-06,
+      "loss": 0.5192,
+      "step": 3846
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.84678372716229e-06,
+      "loss": 0.4875,
+      "step": 3847
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.838898877380054e-06,
+      "loss": 0.5052,
+      "step": 3848
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8310241279626784e-06,
+      "loss": 0.5216,
+      "step": 3849
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8231594835449395e-06,
+      "loss": 0.4998,
+      "step": 3850
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.815304948755664e-06,
+      "loss": 0.5156,
+      "step": 3851
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.807460528217729e-06,
+      "loss": 0.482,
+      "step": 3852
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.79962622654806e-06,
+      "loss": 0.4919,
+      "step": 3853
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7918020483576257e-06,
+      "loss": 0.5232,
+      "step": 3854
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.783987998251436e-06,
+      "loss": 0.5291,
+      "step": 3855
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.776184080828541e-06,
+      "loss": 0.5118,
+      "step": 3856
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7683903006820267e-06,
+      "loss": 0.5357,
+      "step": 3857
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7606066623990145e-06,
+      "loss": 0.4977,
+      "step": 3858
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7528331705606515e-06,
+      "loss": 0.5149,
+      "step": 3859
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7450698297421172e-06,
+      "loss": 0.4994,
+      "step": 3860
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.737316644512622e-06,
+      "loss": 0.5297,
+      "step": 3861
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.729573619435384e-06,
+      "loss": 0.5312,
+      "step": 3862
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.721840759067653e-06,
+      "loss": 0.5057,
+      "step": 3863
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.7141180679606937e-06,
+      "loss": 0.5148,
+      "step": 3864
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.7064055506597875e-06,
+      "loss": 0.5165,
+      "step": 3865
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.698703211704223e-06,
+      "loss": 0.5213,
+      "step": 3866
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.691011055627304e-06,
+      "loss": 0.5425,
+      "step": 3867
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6833290869563346e-06,
+      "loss": 0.5382,
+      "step": 3868
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6756573102126294e-06,
+      "loss": 0.5159,
+      "step": 3869
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6679957299115012e-06,
+      "loss": 0.5049,
+      "step": 3870
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6603443505622583e-06,
+      "loss": 0.5295,
+      "step": 3871
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6527031766682142e-06,
+      "loss": 0.5099,
+      "step": 3872
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6450722127266648e-06,
+      "loss": 0.5111,
+      "step": 3873
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6374514632289085e-06,
+      "loss": 0.5043,
+      "step": 3874
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6298409326602237e-06,
+      "loss": 0.5107,
+      "step": 3875
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.62224062549987e-06,
+      "loss": 0.5406,
+      "step": 3876
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.6146505462211024e-06,
+      "loss": 0.489,
+      "step": 3877
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.60707069929115e-06,
+      "loss": 0.514,
+      "step": 3878
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.599501089171217e-06,
+      "loss": 0.5287,
+      "step": 3879
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5919417203164844e-06,
+      "loss": 0.5133,
+      "step": 3880
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5843925971761108e-06,
+      "loss": 0.5339,
+      "step": 3881
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5768537241932155e-06,
+      "loss": 0.5242,
+      "step": 3882
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.569325105804892e-06,
+      "loss": 0.512,
+      "step": 3883
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.561806746442197e-06,
+      "loss": 0.5067,
+      "step": 3884
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5542986505301446e-06,
+      "loss": 0.5294,
+      "step": 3885
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.546800822487714e-06,
+      "loss": 0.5214,
+      "step": 3886
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5393132667278364e-06,
+      "loss": 0.5047,
+      "step": 3887
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.531835987657407e-06,
+      "loss": 0.512,
+      "step": 3888
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.524368989677255e-06,
+      "loss": 0.5267,
+      "step": 3889
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.516912277182173e-06,
+      "loss": 0.5161,
+      "step": 3890
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5094658545608954e-06,
+      "loss": 0.5156,
+      "step": 3891
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5020297261961004e-06,
+      "loss": 0.5402,
+      "step": 3892
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.494603896464405e-06,
+      "loss": 0.5148,
+      "step": 3893
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.487188369736373e-06,
+      "loss": 0.5163,
+      "step": 3894
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4797831503764935e-06,
+      "loss": 0.5096,
+      "step": 3895
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4723882427431958e-06,
+      "loss": 0.5076,
+      "step": 3896
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.46500365118884e-06,
+      "loss": 0.5298,
+      "step": 3897
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4576293800597097e-06,
+      "loss": 0.5201,
+      "step": 3898
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4502654336960217e-06,
+      "loss": 0.5069,
+      "step": 3899
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4429118164319076e-06,
+      "loss": 0.5096,
+      "step": 3900
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.435568532595427e-06,
+      "loss": 0.5035,
+      "step": 3901
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4282355865085603e-06,
+      "loss": 0.5273,
+      "step": 3902
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.420912982487185e-06,
+      "loss": 0.4922,
+      "step": 3903
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.41360072484111e-06,
+      "loss": 0.4976,
+      "step": 3904
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4062988178740507e-06,
+      "loss": 0.508,
+      "step": 3905
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.3990072658836305e-06,
+      "loss": 0.5316,
+      "step": 3906
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3917260731613733e-06,
+      "loss": 0.4989,
+      "step": 3907
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3844552439927113e-06,
+      "loss": 0.5422,
+      "step": 3908
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.377194782656973e-06,
+      "loss": 0.5329,
+      "step": 3909
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3699446934273906e-06,
+      "loss": 0.4992,
+      "step": 3910
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3627049805710867e-06,
+      "loss": 0.5331,
+      "step": 3911
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.355475648349077e-06,
+      "loss": 0.497,
+      "step": 3912
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3482567010162694e-06,
+      "loss": 0.4933,
+      "step": 3913
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3410481428214602e-06,
+      "loss": 0.5319,
+      "step": 3914
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.333849978007332e-06,
+      "loss": 0.5031,
+      "step": 3915
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.326662210810442e-06,
+      "loss": 0.5285,
+      "step": 3916
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3194848454612352e-06,
+      "loss": 0.5124,
+      "step": 3917
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3123178861840346e-06,
+      "loss": 0.5048,
+      "step": 3918
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3051613371970396e-06,
+      "loss": 0.512,
+      "step": 3919
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2980152027123138e-06,
+      "loss": 0.5236,
+      "step": 3920
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2908794869358044e-06,
+      "loss": 0.5162,
+      "step": 3921
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2837541940673134e-06,
+      "loss": 0.5178,
+      "step": 3922
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2766393283005204e-06,
+      "loss": 0.5185,
+      "step": 3923
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2695348938229576e-06,
+      "loss": 0.5115,
+      "step": 3924
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2624408948160227e-06,
+      "loss": 0.5265,
+      "step": 3925
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.255357335454973e-06,
+      "loss": 0.5057,
+      "step": 3926
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.248284219908918e-06,
+      "loss": 0.5105,
+      "step": 3927
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2412215523408266e-06,
+      "loss": 0.5114,
+      "step": 3928
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.234169336907506e-06,
+      "loss": 0.5177,
+      "step": 3929
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2271275777596202e-06,
+      "loss": 0.537,
+      "step": 3930
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2200962790416836e-06,
+      "loss": 0.5062,
+      "step": 3931
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2130754448920433e-06,
+      "loss": 0.5305,
+      "step": 3932
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.206065079442894e-06,
+      "loss": 0.5187,
+      "step": 3933
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.1990651868202684e-06,
+      "loss": 0.4968,
+      "step": 3934
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.1920757711440354e-06,
+      "loss": 0.5188,
+      "step": 3935
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.1850968365278934e-06,
+      "loss": 0.5146,
+      "step": 3936
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.178128387079379e-06,
+      "loss": 0.5079,
+      "step": 3937
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.171170426899851e-06,
+      "loss": 0.5186,
+      "step": 3938
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1642229600844964e-06,
+      "loss": 0.5074,
+      "step": 3939
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.157285990722333e-06,
+      "loss": 0.5289,
+      "step": 3940
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.150359522896188e-06,
+      "loss": 0.5245,
+      "step": 3941
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.143443560682721e-06,
+      "loss": 0.5323,
+      "step": 3942
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.136538108152395e-06,
+      "loss": 0.5118,
+      "step": 3943
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1296431693694954e-06,
+      "loss": 0.4887,
+      "step": 3944
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1227587483921217e-06,
+      "loss": 0.5562,
+      "step": 3945
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.115884849272176e-06,
+      "loss": 0.518,
+      "step": 3946
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1090214760553727e-06,
+      "loss": 0.5028,
+      "step": 3947
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1021686327812296e-06,
+      "loss": 0.5238,
+      "step": 3948
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.0953263234830667e-06,
+      "loss": 0.518,
+      "step": 3949
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.088494552188007e-06,
+      "loss": 0.5228,
+      "step": 3950
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.081673322916964e-06,
+      "loss": 0.5347,
+      "step": 3951
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.0748626396846537e-06,
+      "loss": 0.4933,
+      "step": 3952
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.068062506499584e-06,
+      "loss": 0.5099,
+      "step": 3953
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0612729273640507e-06,
+      "loss": 0.5091,
+      "step": 3954
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.054493906274142e-06,
+      "loss": 0.5243,
+      "step": 3955
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0477254472197237e-06,
+      "loss": 0.5399,
+      "step": 3956
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.040967554184452e-06,
+      "loss": 0.52,
+      "step": 3957
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.034220231145765e-06,
+      "loss": 0.5069,
+      "step": 3958
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.027483482074879e-06,
+      "loss": 0.5004,
+      "step": 3959
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.020757310936783e-06,
+      "loss": 0.5323,
+      "step": 3960
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.014041721690243e-06,
+      "loss": 0.5006,
+      "step": 3961
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0073367182878e-06,
+      "loss": 0.496,
+      "step": 3962
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0006423046757596e-06,
+      "loss": 0.5288,
+      "step": 3963
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.993958484794194e-06,
+      "loss": 0.5057,
+      "step": 3964
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9872852625769477e-06,
+      "loss": 0.5146,
+      "step": 3965
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9806226419516195e-06,
+      "loss": 0.5049,
+      "step": 3966
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.973970626839572e-06,
+      "loss": 0.5219,
+      "step": 3967
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.967329221155927e-06,
+      "loss": 0.4948,
+      "step": 3968
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9606984288095644e-06,
+      "loss": 0.5305,
+      "step": 3969
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9540782537031045e-06,
+      "loss": 0.5143,
+      "step": 3970
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.947468699732935e-06,
+      "loss": 0.495,
+      "step": 3971
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.940869770789182e-06,
+      "loss": 0.5047,
+      "step": 3972
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9342814707557255e-06,
+      "loss": 0.5257,
+      "step": 3973
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.927703803510181e-06,
+      "loss": 0.5476,
+      "step": 3974
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.921136772923915e-06,
+      "loss": 0.5115,
+      "step": 3975
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.914580382862026e-06,
+      "loss": 0.5166,
+      "step": 3976
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.908034637183356e-06,
+      "loss": 0.5038,
+      "step": 3977
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.901499539740479e-06,
+      "loss": 0.5093,
+      "step": 3978
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8949750943797051e-06,
+      "loss": 0.5075,
+      "step": 3979
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8884613049410672e-06,
+      "loss": 0.5109,
+      "step": 3980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8819581752583317e-06,
+      "loss": 0.5239,
+      "step": 3981
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8754657091589945e-06,
+      "loss": 0.5201,
+      "step": 3982
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8689839104642682e-06,
+      "loss": 0.5234,
+      "step": 3983
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8625127829890922e-06,
+      "loss": 0.5021,
+      "step": 3984
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8560523305421218e-06,
+      "loss": 0.5154,
+      "step": 3985
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8496025569257359e-06,
+      "loss": 0.5298,
+      "step": 3986
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8431634659360175e-06,
+      "loss": 0.5053,
+      "step": 3987
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.836735061362771e-06,
+      "loss": 0.5309,
+      "step": 3988
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8303173469895097e-06,
+      "loss": 0.5193,
+      "step": 3989
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.823910326593452e-06,
+      "loss": 0.4851,
+      "step": 3990
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.817514003945524e-06,
+      "loss": 0.5216,
+      "step": 3991
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8111283828103566e-06,
+      "loss": 0.5137,
+      "step": 3992
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8047534669462851e-06,
+      "loss": 0.4855,
+      "step": 3993
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7983892601053288e-06,
+      "loss": 0.5044,
+      "step": 3994
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7920357660332243e-06,
+      "loss": 0.5132,
+      "step": 3995
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7856929884693898e-06,
+      "loss": 0.5272,
+      "step": 3996
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.779360931146945e-06,
+      "loss": 0.5237,
+      "step": 3997
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7730395977926917e-06,
+      "loss": 0.5482,
+      "step": 3998
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7667289921271247e-06,
+      "loss": 0.5431,
+      "step": 3999
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.760429117864424e-06,
+      "loss": 0.5064,
+      "step": 4000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7541399787124547e-06,
+      "loss": 0.5053,
+      "step": 4001
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7478615783727627e-06,
+      "loss": 0.5318,
+      "step": 4002
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.741593920540572e-06,
+      "loss": 0.5495,
+      "step": 4003
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.735337008904785e-06,
+      "loss": 0.5448,
+      "step": 4004
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7290908471479805e-06,
+      "loss": 0.5106,
+      "step": 4005
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7228554389464136e-06,
+      "loss": 0.5047,
+      "step": 4006
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7166307879699973e-06,
+      "loss": 0.5075,
+      "step": 4007
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7104168978823254e-06,
+      "loss": 0.5335,
+      "step": 4008
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.704213772340657e-06,
+      "loss": 0.5461,
+      "step": 4009
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6980214149959117e-06,
+      "loss": 0.5049,
+      "step": 4010
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6918398294926742e-06,
+      "loss": 0.5277,
+      "step": 4011
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6856690194691872e-06,
+      "loss": 0.5183,
+      "step": 4012
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6795089885573523e-06,
+      "loss": 0.5078,
+      "step": 4013
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6733597403827295e-06,
+      "loss": 0.5264,
+      "step": 4014
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6672212785645282e-06,
+      "loss": 0.5032,
+      "step": 4015
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6610936067156092e-06,
+      "loss": 0.5206,
+      "step": 4016
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6549767284424878e-06,
+      "loss": 0.503,
+      "step": 4017
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.648870647345322e-06,
+      "loss": 0.5294,
+      "step": 4018
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6427753670179214e-06,
+      "loss": 0.521,
+      "step": 4019
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6366908910477276e-06,
+      "loss": 0.521,
+      "step": 4020
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.630617223015829e-06,
+      "loss": 0.4938,
+      "step": 4021
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6245543664969575e-06,
+      "loss": 0.5187,
+      "step": 4022
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6185023250594745e-06,
+      "loss": 0.5183,
+      "step": 4023
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6124611022653813e-06,
+      "loss": 0.5181,
+      "step": 4024
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6064307016703074e-06,
+      "loss": 0.5149,
+      "step": 4025
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6004111268235156e-06,
+      "loss": 0.5195,
+      "step": 4026
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5944023812678987e-06,
+      "loss": 0.5068,
+      "step": 4027
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5884044685399702e-06,
+      "loss": 0.5146,
+      "step": 4028
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5824173921698749e-06,
+      "loss": 0.502,
+      "step": 4029
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5764411556813719e-06,
+      "loss": 0.5167,
+      "step": 4030
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5704757625918454e-06,
+      "loss": 0.5134,
+      "step": 4031
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5645212164123002e-06,
+      "loss": 0.5035,
+      "step": 4032
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5585775206473508e-06,
+      "loss": 0.5423,
+      "step": 4033
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5526446787952276e-06,
+      "loss": 0.5036,
+      "step": 4034
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5467226943477709e-06,
+      "loss": 0.5405,
+      "step": 4035
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5408115707904393e-06,
+      "loss": 0.5198,
+      "step": 4036
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5349113116022874e-06,
+      "loss": 0.519,
+      "step": 4037
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5290219202559841e-06,
+      "loss": 0.5073,
+      "step": 4038
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5231434002177992e-06,
+      "loss": 0.5164,
+      "step": 4039
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5172757549476024e-06,
+      "loss": 0.5007,
+      "step": 4040
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.511418987898865e-06,
+      "loss": 0.4949,
+      "step": 4041
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.505573102518656e-06,
+      "loss": 0.5044,
+      "step": 4042
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4997381022476366e-06,
+      "loss": 0.543,
+      "step": 4043
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.493913990520066e-06,
+      "loss": 0.5074,
+      "step": 4044
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4881007707637919e-06,
+      "loss": 0.5317,
+      "step": 4045
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4822984464002543e-06,
+      "loss": 0.5191,
+      "step": 4046
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4765070208444732e-06,
+      "loss": 0.4948,
+      "step": 4047
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4707264975050616e-06,
+      "loss": 0.5322,
+      "step": 4048
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.464956879784214e-06,
+      "loss": 0.5306,
+      "step": 4049
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.459198171077707e-06,
+      "loss": 0.5284,
+      "step": 4050
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.453450374774892e-06,
+      "loss": 0.5289,
+      "step": 4051
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4477134942587046e-06,
+      "loss": 0.5133,
+      "step": 4052
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.441987532905651e-06,
+      "loss": 0.5077,
+      "step": 4053
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4362724940858109e-06,
+      "loss": 0.5319,
+      "step": 4054
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.430568381162838e-06,
+      "loss": 0.5208,
+      "step": 4055
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4248751974939534e-06,
+      "loss": 0.5036,
+      "step": 4056
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4191929464299481e-06,
+      "loss": 0.5004,
+      "step": 4057
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4135216313151735e-06,
+      "loss": 0.5237,
+      "step": 4058
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4078612554875503e-06,
+      "loss": 0.5182,
+      "step": 4059
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4022118222785608e-06,
+      "loss": 0.5312,
+      "step": 4060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.396573335013236e-06,
+      "loss": 0.5145,
+      "step": 4061
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3909457970101791e-06,
+      "loss": 0.5164,
+      "step": 4062
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.385329211581543e-06,
+      "loss": 0.5411,
+      "step": 4063
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3797235820330323e-06,
+      "loss": 0.5179,
+      "step": 4064
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3741289116639055e-06,
+      "loss": 0.5252,
+      "step": 4065
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.368545203766969e-06,
+      "loss": 0.5046,
+      "step": 4066
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3629724616285822e-06,
+      "loss": 0.5098,
+      "step": 4067
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3574106885286465e-06,
+      "loss": 0.5267,
+      "step": 4068
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3518598877406054e-06,
+      "loss": 0.5134,
+      "step": 4069
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3463200625314476e-06,
+      "loss": 0.5193,
+      "step": 4070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.340791216161703e-06,
+      "loss": 0.5402,
+      "step": 4071
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3352733518854378e-06,
+      "loss": 0.5292,
+      "step": 4072
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.329766472950258e-06,
+      "loss": 0.5062,
+      "step": 4073
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3242705825972935e-06,
+      "loss": 0.5007,
+      "step": 4074
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3187856840612167e-06,
+      "loss": 0.5275,
+      "step": 4075
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3133117805702323e-06,
+      "loss": 0.4952,
+      "step": 4076
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3078488753460671e-06,
+      "loss": 0.4931,
+      "step": 4077
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3023969716039764e-06,
+      "loss": 0.5173,
+      "step": 4078
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2969560725527419e-06,
+      "loss": 0.5095,
+      "step": 4079
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2915261813946666e-06,
+      "loss": 0.4981,
+      "step": 4080
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2861073013255787e-06,
+      "loss": 0.5193,
+      "step": 4081
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2806994355348224e-06,
+      "loss": 0.5204,
+      "step": 4082
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.275302587205256e-06,
+      "loss": 0.5397,
+      "step": 4083
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2699167595132611e-06,
+      "loss": 0.5174,
+      "step": 4084
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2645419556287286e-06,
+      "loss": 0.5067,
+      "step": 4085
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2591781787150593e-06,
+      "loss": 0.5161,
+      "step": 4086
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.253825431929172e-06,
+      "loss": 0.5136,
+      "step": 4087
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2484837184214804e-06,
+      "loss": 0.5102,
+      "step": 4088
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2431530413359138e-06,
+      "loss": 0.523,
+      "step": 4089
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.237833403809905e-06,
+      "loss": 0.5314,
+      "step": 4090
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.232524808974389e-06,
+      "loss": 0.5181,
+      "step": 4091
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2272272599537982e-06,
+      "loss": 0.5043,
+      "step": 4092
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2219407598660692e-06,
+      "loss": 0.5199,
+      "step": 4093
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2166653118226335e-06,
+      "loss": 0.5171,
+      "step": 4094
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.211400918928416e-06,
+      "loss": 0.5226,
+      "step": 4095
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.5046,
+      "step": 4096
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2009053109748002e-06,
+      "loss": 0.5207,
+      "step": 4097
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1956741020927166e-06,
+      "loss": 0.5278,
+      "step": 4098
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1904539607144682e-06,
+      "loss": 0.5365,
+      "step": 4099
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.185244889912438e-06,
+      "loss": 0.522,
+      "step": 4100
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1800468927524733e-06,
+      "loss": 0.5251,
+      "step": 4101
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1748599722939224e-06,
+      "loss": 0.5055,
+      "step": 4102
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.169684131589608e-06,
+      "loss": 0.5,
+      "step": 4103
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.164519373685824e-06,
+      "loss": 0.5061,
+      "step": 4104
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1593657016223569e-06,
+      "loss": 0.4725,
+      "step": 4105
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.154223118432458e-06,
+      "loss": 0.5213,
+      "step": 4106
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1490916271428531e-06,
+      "loss": 0.5053,
+      "step": 4107
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1439712307737394e-06,
+      "loss": 0.5128,
+      "step": 4108
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1388619323387884e-06,
+      "loss": 0.5136,
+      "step": 4109
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1337637348451369e-06,
+      "loss": 0.5066,
+      "step": 4110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.128676641293387e-06,
+      "loss": 0.5053,
+      "step": 4111
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1236006546776058e-06,
+      "loss": 0.5224,
+      "step": 4112
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1185357779853234e-06,
+      "loss": 0.5032,
+      "step": 4113
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.113482014197531e-06,
+      "loss": 0.5207,
+      "step": 4114
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1084393662886827e-06,
+      "loss": 0.5331,
+      "step": 4115
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1034078372266866e-06,
+      "loss": 0.5101,
+      "step": 4116
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0983874299729092e-06,
+      "loss": 0.5231,
+      "step": 4117
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0933781474821648e-06,
+      "loss": 0.5148,
+      "step": 4118
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0883799927027261e-06,
+      "loss": 0.5232,
+      "step": 4119
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0833929685763178e-06,
+      "loss": 0.4978,
+      "step": 4120
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0784170780381076e-06,
+      "loss": 0.5184,
+      "step": 4121
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.073452324016715e-06,
+      "loss": 0.5373,
+      "step": 4122
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0684987094342025e-06,
+      "loss": 0.5222,
+      "step": 4123
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0635562372060825e-06,
+      "loss": 0.4948,
+      "step": 4124
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.058624910241297e-06,
+      "loss": 0.5045,
+      "step": 4125
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0537047314422377e-06,
+      "loss": 0.5134,
+      "step": 4126
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0487957037047325e-06,
+      "loss": 0.5322,
+      "step": 4127
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0438978299180502e-06,
+      "loss": 0.5082,
+      "step": 4128
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0390111129648872e-06,
+      "loss": 0.5048,
+      "step": 4129
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0341355557213785e-06,
+      "loss": 0.5365,
+      "step": 4130
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0292711610570904e-06,
+      "loss": 0.5243,
+      "step": 4131
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0244179318350178e-06,
+      "loss": 0.4901,
+      "step": 4132
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0195758709115867e-06,
+      "loss": 0.5265,
+      "step": 4133
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0147449811366461e-06,
+      "loss": 0.4958,
+      "step": 4134
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0099252653534731e-06,
+      "loss": 0.4993,
+      "step": 4135
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.005116726398767e-06,
+      "loss": 0.5258,
+      "step": 4136
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0003193671026511e-06,
+      "loss": 0.5132,
+      "step": 4137
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.955331902886645e-07,
+      "loss": 0.508,
+      "step": 4138
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.907581987737668e-07,
+      "loss": 0.5263,
+      "step": 4139
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.859943953683349e-07,
+      "loss": 0.5413,
+      "step": 4140
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.812417828761611e-07,
+      "loss": 0.5316,
+      "step": 4141
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.765003640944505e-07,
+      "loss": 0.5217,
+      "step": 4142
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.717701418138193e-07,
+      "loss": 0.5121,
+      "step": 4143
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.670511188182941e-07,
+      "loss": 0.4966,
+      "step": 4144
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.62343297885313e-07,
+      "loss": 0.5218,
+      "step": 4145
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.576466817857177e-07,
+      "loss": 0.4969,
+      "step": 4146
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.529612732837545e-07,
+      "loss": 0.5133,
+      "step": 4147
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.482870751370755e-07,
+      "loss": 0.5084,
+      "step": 4148
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.436240900967353e-07,
+      "loss": 0.5168,
+      "step": 4149
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.389723209071877e-07,
+      "loss": 0.5095,
+      "step": 4150
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.343317703062848e-07,
+      "loss": 0.5276,
+      "step": 4151
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.297024410252753e-07,
+      "loss": 0.5207,
+      "step": 4152
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.250843357888062e-07,
+      "loss": 0.5128,
+      "step": 4153
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.204774573149144e-07,
+      "loss": 0.5147,
+      "step": 4154
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.158818083150333e-07,
+      "loss": 0.5318,
+      "step": 4155
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.11297391493986e-07,
+      "loss": 0.5407,
+      "step": 4156
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.067242095499828e-07,
+      "loss": 0.4983,
+      "step": 4157
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.021622651746242e-07,
+      "loss": 0.5111,
+      "step": 4158
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.976115610528957e-07,
+      "loss": 0.5299,
+      "step": 4159
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.930720998631637e-07,
+      "loss": 0.5275,
+      "step": 4160
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.885438842771843e-07,
+      "loss": 0.4829,
+      "step": 4161
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.840269169600901e-07,
+      "loss": 0.5312,
+      "step": 4162
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.795212005703946e-07,
+      "loss": 0.5033,
+      "step": 4163
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.750267377599919e-07,
+      "loss": 0.5272,
+      "step": 4164
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.70543531174144e-07,
+      "loss": 0.5118,
+      "step": 4165
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.660715834514977e-07,
+      "loss": 0.5128,
+      "step": 4166
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.616108972240699e-07,
+      "loss": 0.5102,
+      "step": 4167
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.571614751172475e-07,
+      "loss": 0.5185,
+      "step": 4168
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.527233197497908e-07,
+      "loss": 0.5232,
+      "step": 4169
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.482964337338284e-07,
+      "loss": 0.4932,
+      "step": 4170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.438808196748538e-07,
+      "loss": 0.5101,
+      "step": 4171
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.394764801717281e-07,
+      "loss": 0.519,
+      "step": 4172
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.350834178166755e-07,
+      "loss": 0.5191,
+      "step": 4173
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.307016351952857e-07,
+      "loss": 0.5061,
+      "step": 4174
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.263311348865066e-07,
+      "loss": 0.5476,
+      "step": 4175
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.219719194626474e-07,
+      "loss": 0.5312,
+      "step": 4176
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.176239914893735e-07,
+      "loss": 0.5137,
+      "step": 4177
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.132873535257135e-07,
+      "loss": 0.4978,
+      "step": 4178
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.089620081240368e-07,
+      "loss": 0.5259,
+      "step": 4179
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.046479578300803e-07,
+      "loss": 0.5219,
+      "step": 4180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.003452051829286e-07,
+      "loss": 0.535,
+      "step": 4181
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.960537527150158e-07,
+      "loss": 0.5052,
+      "step": 4182
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.917736029521262e-07,
+      "loss": 0.5174,
+      "step": 4183
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.875047584133932e-07,
+      "loss": 0.5137,
+      "step": 4184
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.832472216112919e-07,
+      "loss": 0.5134,
+      "step": 4185
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.790009950516464e-07,
+      "loss": 0.5348,
+      "step": 4186
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.747660812336221e-07,
+      "loss": 0.5303,
+      "step": 4187
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.705424826497277e-07,
+      "loss": 0.5187,
+      "step": 4188
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.663302017858099e-07,
+      "loss": 0.5125,
+      "step": 4189
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.621292411210568e-07,
+      "loss": 0.5145,
+      "step": 4190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.579396031279907e-07,
+      "loss": 0.5142,
+      "step": 4191
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.537612902724723e-07,
+      "loss": 0.5291,
+      "step": 4192
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.495943050136945e-07,
+      "loss": 0.4996,
+      "step": 4193
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.454386498041865e-07,
+      "loss": 0.5201,
+      "step": 4194
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.412943270898077e-07,
+      "loss": 0.513,
+      "step": 4195
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.371613393097466e-07,
+      "loss": 0.5173,
+      "step": 4196
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.330396888965219e-07,
+      "loss": 0.4932,
+      "step": 4197
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.289293782759755e-07,
+      "loss": 0.5079,
+      "step": 4198
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.24830409867281e-07,
+      "loss": 0.5097,
+      "step": 4199
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.207427860829352e-07,
+      "loss": 0.5137,
+      "step": 4200
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.166665093287539e-07,
+      "loss": 0.5304,
+      "step": 4201
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.126015820038757e-07,
+      "loss": 0.5347,
+      "step": 4202
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.085480065007643e-07,
+      "loss": 0.5077,
+      "step": 4203
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.045057852051984e-07,
+      "loss": 0.5016,
+      "step": 4204
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.004749204962725e-07,
+      "loss": 0.4917,
+      "step": 4205
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.964554147463976e-07,
+      "loss": 0.5112,
+      "step": 4206
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.924472703212992e-07,
+      "loss": 0.5335,
+      "step": 4207
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.884504895800237e-07,
+      "loss": 0.5164,
+      "step": 4208
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.84465074874916e-07,
+      "loss": 0.5089,
+      "step": 4209
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.804910285516397e-07,
+      "loss": 0.5146,
+      "step": 4210
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.765283529491684e-07,
+      "loss": 0.5236,
+      "step": 4211
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.725770503997786e-07,
+      "loss": 0.4895,
+      "step": 4212
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.686371232290567e-07,
+      "loss": 0.5336,
+      "step": 4213
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.6470857375589e-07,
+      "loss": 0.5,
+      "step": 4214
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.607914042924756e-07,
+      "loss": 0.516,
+      "step": 4215
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.56885617144305e-07,
+      "loss": 0.5088,
+      "step": 4216
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.529912146101747e-07,
+      "loss": 0.5071,
+      "step": 4217
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.491081989821868e-07,
+      "loss": 0.5054,
+      "step": 4218
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.452365725457266e-07,
+      "loss": 0.5421,
+      "step": 4219
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.413763375794868e-07,
+      "loss": 0.5309,
+      "step": 4220
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.375274963554568e-07,
+      "loss": 0.5005,
+      "step": 4221
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.336900511389133e-07,
+      "loss": 0.4991,
+      "step": 4222
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.298640041884319e-07,
+      "loss": 0.5231,
+      "step": 4223
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.260493577558757e-07,
+      "loss": 0.5293,
+      "step": 4224
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.222461140863978e-07,
+      "loss": 0.528,
+      "step": 4225
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.184542754184431e-07,
+      "loss": 0.5145,
+      "step": 4226
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.146738439837396e-07,
+      "loss": 0.5123,
+      "step": 4227
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.109048220073077e-07,
+      "loss": 0.4736,
+      "step": 4228
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.071472117074462e-07,
+      "loss": 0.5189,
+      "step": 4229
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.034010152957414e-07,
+      "loss": 0.5138,
+      "step": 4230
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.996662349770633e-07,
+      "loss": 0.5196,
+      "step": 4231
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.959428729495509e-07,
+      "loss": 0.4973,
+      "step": 4232
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.922309314046426e-07,
+      "loss": 0.501,
+      "step": 4233
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.885304125270397e-07,
+      "loss": 0.5147,
+      "step": 4234
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.848413184947244e-07,
+      "loss": 0.5258,
+      "step": 4235
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.811636514789598e-07,
+      "loss": 0.5105,
+      "step": 4236
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.774974136442746e-07,
+      "loss": 0.5205,
+      "step": 4237
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.738426071484782e-07,
+      "loss": 0.4947,
+      "step": 4238
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.701992341426499e-07,
+      "loss": 0.4967,
+      "step": 4239
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.665672967711389e-07,
+      "loss": 0.5345,
+      "step": 4240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.629467971715618e-07,
+      "loss": 0.5266,
+      "step": 4241
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.593377374748099e-07,
+      "loss": 0.5051,
+      "step": 4242
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.557401198050327e-07,
+      "loss": 0.5109,
+      "step": 4243
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.521539462796499e-07,
+      "loss": 0.5145,
+      "step": 4244
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.485792190093464e-07,
+      "loss": 0.5176,
+      "step": 4245
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.450159400980725e-07,
+      "loss": 0.4947,
+      "step": 4246
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.414641116430331e-07,
+      "loss": 0.4859,
+      "step": 4247
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.379237357346978e-07,
+      "loss": 0.533,
+      "step": 4248
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.343948144568001e-07,
+      "loss": 0.5381,
+      "step": 4249
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.308773498863251e-07,
+      "loss": 0.537,
+      "step": 4250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.273713440935146e-07,
+      "loss": 0.4914,
+      "step": 4251
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.238767991418737e-07,
+      "loss": 0.5236,
+      "step": 4252
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.203937170881545e-07,
+      "loss": 0.502,
+      "step": 4253
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.169220999823666e-07,
+      "loss": 0.5458,
+      "step": 4254
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.134619498677706e-07,
+      "loss": 0.5194,
+      "step": 4255
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.100132687808757e-07,
+      "loss": 0.5121,
+      "step": 4256
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.065760587514446e-07,
+      "loss": 0.5374,
+      "step": 4257
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.031503218024858e-07,
+      "loss": 0.5223,
+      "step": 4258
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.997360599502577e-07,
+      "loss": 0.5153,
+      "step": 4259
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.96333275204266e-07,
+      "loss": 0.5083,
+      "step": 4260
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.929419695672533e-07,
+      "loss": 0.5073,
+      "step": 4261
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.895621450352184e-07,
+      "loss": 0.5407,
+      "step": 4262
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.861938035973901e-07,
+      "loss": 0.4855,
+      "step": 4263
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.828369472362493e-07,
+      "loss": 0.4839,
+      "step": 4264
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.794915779275111e-07,
+      "loss": 0.4943,
+      "step": 4265
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.7615769764013166e-07,
+      "loss": 0.5261,
+      "step": 4266
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.7283530833630354e-07,
+      "loss": 0.5102,
+      "step": 4267
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.6952441197145817e-07,
+      "loss": 0.5141,
+      "step": 4268
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.662250104942656e-07,
+      "loss": 0.5137,
+      "step": 4269
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.629371058466192e-07,
+      "loss": 0.5069,
+      "step": 4270
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.5966069996365993e-07,
+      "loss": 0.524,
+      "step": 4271
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.5639579477374964e-07,
+      "loss": 0.5132,
+      "step": 4272
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.531423921984912e-07,
+      "loss": 0.5108,
+      "step": 4273
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.499004941527085e-07,
+      "loss": 0.4983,
+      "step": 4274
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.4667010254445755e-07,
+      "loss": 0.5279,
+      "step": 4275
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.434512192750262e-07,
+      "loss": 0.5044,
+      "step": 4276
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.402438462389236e-07,
+      "loss": 0.532,
+      "step": 4277
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.3704798532388624e-07,
+      "loss": 0.4854,
+      "step": 4278
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.338636384108763e-07,
+      "loss": 0.5212,
+      "step": 4279
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.3069080737407677e-07,
+      "loss": 0.5363,
+      "step": 4280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.2752949408089385e-07,
+      "loss": 0.5222,
+      "step": 4281
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.243797003919592e-07,
+      "loss": 0.5186,
+      "step": 4282
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.2124142816111215e-07,
+      "loss": 0.5118,
+      "step": 4283
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.1811467923542625e-07,
+      "loss": 0.5214,
+      "step": 4284
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.1499945545518283e-07,
+      "loss": 0.4915,
+      "step": 4285
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.1189575865388407e-07,
+      "loss": 0.5189,
+      "step": 4286
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.088035906582466e-07,
+      "loss": 0.4924,
+      "step": 4287
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.05722953288199e-07,
+      "loss": 0.5514,
+      "step": 4288
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.0265384835689093e-07,
+      "loss": 0.5206,
+      "step": 4289
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.9959627767067523e-07,
+      "loss": 0.5044,
+      "step": 4290
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.965502430291235e-07,
+      "loss": 0.4817,
+      "step": 4291
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.935157462250128e-07,
+      "loss": 0.499,
+      "step": 4292
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.9049278904433217e-07,
+      "loss": 0.5269,
+      "step": 4293
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.8748137326627855e-07,
+      "loss": 0.5329,
+      "step": 4294
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.844815006632563e-07,
+      "loss": 0.5201,
+      "step": 4295
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.814931730008753e-07,
+      "loss": 0.5395,
+      "step": 4296
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7851639203794646e-07,
+      "loss": 0.5146,
+      "step": 4297
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7555115952649045e-07,
+      "loss": 0.5255,
+      "step": 4298
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7259747721173134e-07,
+      "loss": 0.5124,
+      "step": 4299
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.696553468320918e-07,
+      "loss": 0.5316,
+      "step": 4300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6672477011919785e-07,
+      "loss": 0.5121,
+      "step": 4301
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.638057487978719e-07,
+      "loss": 0.507,
+      "step": 4302
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6089828458613974e-07,
+      "loss": 0.5107,
+      "step": 4303
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5800237919522363e-07,
+      "loss": 0.513,
+      "step": 4304
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5511803432954016e-07,
+      "loss": 0.5057,
+      "step": 4305
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.522452516867048e-07,
+      "loss": 0.5307,
+      "step": 4306
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.493840329575271e-07,
+      "loss": 0.5192,
+      "step": 4307
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.4653437982601125e-07,
+      "loss": 0.5352,
+      "step": 4308
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.4369629396935333e-07,
+      "loss": 0.5493,
+      "step": 4309
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.4086977705793943e-07,
+      "loss": 0.5112,
+      "step": 4310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3805483075535215e-07,
+      "loss": 0.5206,
+      "step": 4311
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3525145671835957e-07,
+      "loss": 0.5269,
+      "step": 4312
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.324596565969174e-07,
+      "loss": 0.5064,
+      "step": 4313
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.2967943203417564e-07,
+      "loss": 0.5289,
+      "step": 4314
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.2691078466646764e-07,
+      "loss": 0.5053,
+      "step": 4315
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.241537161233077e-07,
+      "loss": 0.5074,
+      "step": 4316
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.214082280274067e-07,
+      "loss": 0.5158,
+      "step": 4317
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.18674321994652e-07,
+      "loss": 0.5141,
+      "step": 4318
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.1595199963411207e-07,
+      "loss": 0.5072,
+      "step": 4319
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.1324126254804524e-07,
+      "loss": 0.5189,
+      "step": 4320
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.1054211233188414e-07,
+      "loss": 0.5109,
+      "step": 4321
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.078545505742514e-07,
+      "loss": 0.5045,
+      "step": 4322
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.051785788569328e-07,
+      "loss": 0.4981,
+      "step": 4323
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.025141987549085e-07,
+      "loss": 0.5078,
+      "step": 4324
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.9986141183633076e-07,
+      "loss": 0.5236,
+      "step": 4325
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.9722021966252404e-07,
+      "loss": 0.5185,
+      "step": 4326
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.9459062378799806e-07,
+      "loss": 0.5176,
+      "step": 4327
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.919726257604261e-07,
+      "loss": 0.5063,
+      "step": 4328
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.893662271206643e-07,
+      "loss": 0.4931,
+      "step": 4329
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8677142940274126e-07,
+      "loss": 0.5034,
+      "step": 4330
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.841882341338509e-07,
+      "loss": 0.5185,
+      "step": 4331
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.816166428343636e-07,
+      "loss": 0.5117,
+      "step": 4332
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.7905665701781994e-07,
+      "loss": 0.5043,
+      "step": 4333
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.7650827819093005e-07,
+      "loss": 0.5014,
+      "step": 4334
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.7397150785356984e-07,
+      "loss": 0.5234,
+      "step": 4335
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.714463474987894e-07,
+      "loss": 0.501,
+      "step": 4336
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.689327986127954e-07,
+      "loss": 0.5133,
+      "step": 4337
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.66430862674969e-07,
+      "loss": 0.5121,
+      "step": 4338
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.639405411578522e-07,
+      "loss": 0.5193,
+      "step": 4339
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6146183552715254e-07,
+      "loss": 0.4935,
+      "step": 4340
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5899474724174313e-07,
+      "loss": 0.516,
+      "step": 4341
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5653927775365573e-07,
+      "loss": 0.5189,
+      "step": 4342
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5409542850808765e-07,
+      "loss": 0.5077,
+      "step": 4343
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.516632009433906e-07,
+      "loss": 0.5003,
+      "step": 4344
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.492425964910816e-07,
+      "loss": 0.5032,
+      "step": 4345
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.468336165758389e-07,
+      "loss": 0.5088,
+      "step": 4346
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.444362626154906e-07,
+      "loss": 0.504,
+      "step": 4347
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.4205053602103015e-07,
+      "loss": 0.5068,
+      "step": 4348
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.396764381966055e-07,
+      "loss": 0.4853,
+      "step": 4349
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3731397053951666e-07,
+      "loss": 0.5276,
+      "step": 4350
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3496313444022255e-07,
+      "loss": 0.5172,
+      "step": 4351
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3262393128233418e-07,
+      "loss": 0.5088,
+      "step": 4352
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3029636244261688e-07,
+      "loss": 0.53,
+      "step": 4353
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.279804292909882e-07,
+      "loss": 0.5182,
+      "step": 4354
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2567613319051997e-07,
+      "loss": 0.4918,
+      "step": 4355
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2338347549742956e-07,
+      "loss": 0.5236,
+      "step": 4356
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.211024575610887e-07,
+      "loss": 0.4912,
+      "step": 4357
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1883308072401466e-07,
+      "loss": 0.5065,
+      "step": 4358
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1657534632187893e-07,
+      "loss": 0.5311,
+      "step": 4359
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1432925568349416e-07,
+      "loss": 0.5141,
+      "step": 4360
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.120948101308251e-07,
+      "loss": 0.5139,
+      "step": 4361
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0987201097897757e-07,
+      "loss": 0.5168,
+      "step": 4362
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.076608595362095e-07,
+      "loss": 0.5286,
+      "step": 4363
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0546135710391767e-07,
+      "loss": 0.5008,
+      "step": 4364
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0327350497664656e-07,
+      "loss": 0.5113,
+      "step": 4365
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0109730444207942e-07,
+      "loss": 0.509,
+      "step": 4366
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9893275678104507e-07,
+      "loss": 0.514,
+      "step": 4367
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9677986326751108e-07,
+      "loss": 0.5121,
+      "step": 4368
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9463862516859277e-07,
+      "loss": 0.5019,
+      "step": 4369
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.925090437445376e-07,
+      "loss": 0.4951,
+      "step": 4370
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.903911202487363e-07,
+      "loss": 0.492,
+      "step": 4371
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8828485592771617e-07,
+      "loss": 0.4925,
+      "step": 4372
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.861902520211456e-07,
+      "loss": 0.5047,
+      "step": 4373
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.841073097618229e-07,
+      "loss": 0.5088,
+      "step": 4374
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8203603037569406e-07,
+      "loss": 0.5122,
+      "step": 4375
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.799764150818306e-07,
+      "loss": 0.5085,
+      "step": 4376
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.779284650924451e-07,
+      "loss": 0.5318,
+      "step": 4377
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7589218161288003e-07,
+      "loss": 0.4949,
+      "step": 4378
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.738675658416167e-07,
+      "loss": 0.5225,
+      "step": 4379
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.718546189702641e-07,
+      "loss": 0.5073,
+      "step": 4380
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6985334218356575e-07,
+      "loss": 0.496,
+      "step": 4381
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6786373665939492e-07,
+      "loss": 0.5061,
+      "step": 4382
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.658858035687594e-07,
+      "loss": 0.53,
+      "step": 4383
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.639195440757946e-07,
+      "loss": 0.5296,
+      "step": 4384
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6196495933776147e-07,
+      "loss": 0.5415,
+      "step": 4385
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.600220505050598e-07,
+      "loss": 0.4951,
+      "step": 4386
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5809081872120603e-07,
+      "loss": 0.4787,
+      "step": 4387
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5617126512285086e-07,
+      "loss": 0.5088,
+      "step": 4388
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.542633908397706e-07,
+      "loss": 0.5394,
+      "step": 4389
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5236719699486256e-07,
+      "loss": 0.5204,
+      "step": 4390
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5048268470416073e-07,
+      "loss": 0.5118,
+      "step": 4391
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4860985507680893e-07,
+      "loss": 0.4998,
+      "step": 4392
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4674870921508987e-07,
+      "loss": 0.4938,
+      "step": 4393
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4489924821439627e-07,
+      "loss": 0.5054,
+      "step": 4394
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4306147316325293e-07,
+      "loss": 0.5103,
+      "step": 4395
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.412353851433057e-07,
+      "loss": 0.5157,
+      "step": 4396
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3942098522931491e-07,
+      "loss": 0.5253,
+      "step": 4397
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3761827448917076e-07,
+      "loss": 0.5235,
+      "step": 4398
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3582725398387785e-07,
+      "loss": 0.5136,
+      "step": 4399
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.340479247675619e-07,
+      "loss": 0.5215,
+      "step": 4400
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.322802878874696e-07,
+      "loss": 0.5408,
+      "step": 4401
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.305243443839599e-07,
+      "loss": 0.5232,
+      "step": 4402
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2878009529051938e-07,
+      "loss": 0.5245,
+      "step": 4403
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2704754163374022e-07,
+      "loss": 0.5175,
+      "step": 4404
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2532668443334006e-07,
+      "loss": 0.5362,
+      "step": 4405
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2361752470215094e-07,
+      "loss": 0.5238,
+      "step": 4406
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2192006344611485e-07,
+      "loss": 0.516,
+      "step": 4407
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.2023430166429485e-07,
+      "loss": 0.5235,
+      "step": 4408
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1856024034886616e-07,
+      "loss": 0.5342,
+      "step": 4409
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1689788048511397e-07,
+      "loss": 0.5144,
+      "step": 4410
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1524722305144231e-07,
+      "loss": 0.5459,
+      "step": 4411
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1360826901936073e-07,
+      "loss": 0.4982,
+      "step": 4412
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1198101935350203e-07,
+      "loss": 0.5342,
+      "step": 4413
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1036547501159567e-07,
+      "loss": 0.5176,
+      "step": 4414
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0876163694448993e-07,
+      "loss": 0.5002,
+      "step": 4415
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0716950609614751e-07,
+      "loss": 0.4967,
+      "step": 4416
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0558908340362994e-07,
+      "loss": 0.5188,
+      "step": 4417
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0402036979711317e-07,
+      "loss": 0.5283,
+      "step": 4418
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0246336619988751e-07,
+      "loss": 0.501,
+      "step": 4419
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0091807352833993e-07,
+      "loss": 0.5008,
+      "step": 4420
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.938449269197181e-08,
+      "loss": 0.4944,
+      "step": 4421
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.786262459339002e-08,
+      "loss": 0.5196,
+      "step": 4422
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.635247012830917e-08,
+      "loss": 0.4979,
+      "step": 4423
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.485403018554717e-08,
+      "loss": 0.4872,
+      "step": 4424
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.336730564702745e-08,
+      "loss": 0.5107,
+      "step": 4425
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.189229738777894e-08,
+      "loss": 0.5104,
+      "step": 4426
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.042900627593609e-08,
+      "loss": 0.5013,
+      "step": 4427
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.897743317273666e-08,
+      "loss": 0.5094,
+      "step": 4428
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.753757893251724e-08,
+      "loss": 0.5215,
+      "step": 4429
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.610944440272218e-08,
+      "loss": 0.5166,
+      "step": 4430
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.469303042389688e-08,
+      "loss": 0.5175,
+      "step": 4431
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.328833782969003e-08,
+      "loss": 0.5228,
+      "step": 4432
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.1895367446847e-08,
+      "loss": 0.5108,
+      "step": 4433
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.051412009521864e-08,
+      "loss": 0.5166,
+      "step": 4434
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.914459658775464e-08,
+      "loss": 0.5126,
+      "step": 4435
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.778679773050136e-08,
+      "loss": 0.5034,
+      "step": 4436
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.644072432260619e-08,
+      "loss": 0.4942,
+      "step": 4437
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.510637715631985e-08,
+      "loss": 0.5149,
+      "step": 4438
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.378375701698748e-08,
+      "loss": 0.5415,
+      "step": 4439
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.247286468305081e-08,
+      "loss": 0.5048,
+      "step": 4440
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.117370092605048e-08,
+      "loss": 0.5289,
+      "step": 4441
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.98862665106259e-08,
+      "loss": 0.5211,
+      "step": 4442
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.861056219450879e-08,
+      "loss": 0.5019,
+      "step": 4443
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.734658872853406e-08,
+      "loss": 0.5243,
+      "step": 4444
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.609434685662219e-08,
+      "loss": 0.5096,
+      "step": 4445
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.485383731580142e-08,
+      "loss": 0.5308,
+      "step": 4446
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.362506083618103e-08,
+      "loss": 0.4933,
+      "step": 4447
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.240801814097808e-08,
+      "loss": 0.5135,
+      "step": 4448
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.12027099464929e-08,
+      "loss": 0.5119,
+      "step": 4449
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.000913696212474e-08,
+      "loss": 0.5273,
+      "step": 4450
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.882729989036495e-08,
+      "loss": 0.4979,
+      "step": 4451
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.765719942679715e-08,
+      "loss": 0.502,
+      "step": 4452
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.649883626009933e-08,
+      "loss": 0.5061,
+      "step": 4453
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.5352211072037256e-08,
+      "loss": 0.5036,
+      "step": 4454
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.4217324537471084e-08,
+      "loss": 0.5278,
+      "step": 4455
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.309417732435096e-08,
+      "loss": 0.5152,
+      "step": 4456
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.1982770093719215e-08,
+      "loss": 0.5298,
+      "step": 4457
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.088310349970593e-08,
+      "loss": 0.517,
+      "step": 4458
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.979517818953561e-08,
+      "loss": 0.5143,
+      "step": 4459
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.871899480351605e-08,
+      "loss": 0.5231,
+      "step": 4460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.7654553975049476e-08,
+      "loss": 0.5155,
+      "step": 4461
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.660185633062364e-08,
+      "loss": 0.5258,
+      "step": 4462
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.556090248981626e-08,
+      "loss": 0.5276,
+      "step": 4463
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4531693065292816e-08,
+      "loss": 0.4816,
+      "step": 4464
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.3514228662806525e-08,
+      "loss": 0.4953,
+      "step": 4465
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2508509881198365e-08,
+      "loss": 0.494,
+      "step": 4466
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.151453731239707e-08,
+      "loss": 0.518,
+      "step": 4467
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.053231154141246e-08,
+      "loss": 0.4993,
+      "step": 4468
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.956183314634876e-08,
+      "loss": 0.4986,
+      "step": 4469
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.8603102698391295e-08,
+      "loss": 0.5161,
+      "step": 4470
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.765612076181091e-08,
+      "loss": 0.495,
+      "step": 4471
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.672088789396622e-08,
+      "loss": 0.5106,
+      "step": 4472
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5797404645296906e-08,
+      "loss": 0.513,
+      "step": 4473
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4885671559332645e-08,
+      "loss": 0.5045,
+      "step": 4474
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.398568917268419e-08,
+      "loss": 0.5182,
+      "step": 4475
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.309745801504338e-08,
+      "loss": 0.5228,
+      "step": 4476
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2220978609192045e-08,
+      "loss": 0.5107,
+      "step": 4477
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1356251470993084e-08,
+      "loss": 0.5237,
+      "step": 4478
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0503277109388274e-08,
+      "loss": 0.4775,
+      "step": 4479
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9662056026409368e-08,
+      "loss": 0.5069,
+      "step": 4480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8832588717164766e-08,
+      "loss": 0.4828,
+      "step": 4481
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8014875669846174e-08,
+      "loss": 0.506,
+      "step": 4482
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.7208917365728616e-08,
+      "loss": 0.4928,
+      "step": 4483
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.6414714279168196e-08,
+      "loss": 0.5283,
+      "step": 4484
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.5632266877604338e-08,
+      "loss": 0.5111,
+      "step": 4485
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4861575621553112e-08,
+      "loss": 0.5117,
+      "step": 4486
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.41026409646139e-08,
+      "loss": 0.5226,
+      "step": 4487
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.3355463353467168e-08,
+      "loss": 0.5316,
+      "step": 4488
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.262004322787448e-08,
+      "loss": 0.5367,
+      "step": 4489
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.189638102067404e-08,
+      "loss": 0.518,
+      "step": 4490
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1184477157785155e-08,
+      "loss": 0.5228,
+      "step": 4491
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0484332058208213e-08,
+      "loss": 0.5217,
+      "step": 4492
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.979594613402247e-08,
+      "loss": 0.5246,
+      "step": 4493
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.911931979038384e-08,
+      "loss": 0.5049,
+      "step": 4494
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8454453425527098e-08,
+      "loss": 0.494,
+      "step": 4495
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.780134743077033e-08,
+      "loss": 0.5197,
+      "step": 4496
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7160002190506064e-08,
+      "loss": 0.5226,
+      "step": 4497
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.653041808220568e-08,
+      "loss": 0.5216,
+      "step": 4498
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5912595476414993e-08,
+      "loss": 0.5018,
+      "step": 4499
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5306534736763134e-08,
+      "loss": 0.517,
+      "step": 4500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4712236219955878e-08,
+      "loss": 0.5005,
+      "step": 4501
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4129700275771208e-08,
+      "loss": 0.5209,
+      "step": 4502
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3558927247070419e-08,
+      "loss": 0.5055,
+      "step": 4503
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2999917469784795e-08,
+      "loss": 0.533,
+      "step": 4504
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.245267127292893e-08,
+      "loss": 0.5115,
+      "step": 4505
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1917188978591842e-08,
+      "loss": 0.5052,
+      "step": 4506
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1393470901934767e-08,
+      "loss": 0.5272,
+      "step": 4507
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.08815173511978e-08,
+      "loss": 0.5093,
+      "step": 4508
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0381328627702136e-08,
+      "loss": 0.5061,
+      "step": 4509
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.892905025834509e-09,
+      "loss": 0.4893,
+      "step": 4510
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.416246833062747e-09,
+      "loss": 0.4935,
+      "step": 4511
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.951354329933548e-09,
+      "loss": 0.5191,
+      "step": 4512
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.498227790059154e-09,
+      "loss": 0.5256,
+      "step": 4513
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.056867480137342e-09,
+      "loss": 0.5275,
+      "step": 4514
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.627273659933653e-09,
+      "loss": 0.525,
+      "step": 4515
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.209446582292501e-09,
+      "loss": 0.4935,
+      "step": 4516
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.803386493126063e-09,
+      "loss": 0.5192,
+      "step": 4517
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.4090936314298306e-09,
+      "loss": 0.5086,
+      "step": 4518
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.026568229269281e-09,
+      "loss": 0.5048,
+      "step": 4519
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.6558105117821e-09,
+      "loss": 0.5165,
+      "step": 4520
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.296820697184846e-09,
+      "loss": 0.5226,
+      "step": 4521
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.949598996761839e-09,
+      "loss": 0.5194,
+      "step": 4522
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.614145614876275e-09,
+      "loss": 0.4958,
+      "step": 4523
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.290460748963554e-09,
+      "loss": 0.5153,
+      "step": 4524
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.9785445895290656e-09,
+      "loss": 0.513,
+      "step": 4525
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.678397320159288e-09,
+      "loss": 0.5315,
+      "step": 4526
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3900191175084696e-09,
+      "loss": 0.4898,
+      "step": 4527
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.113410151300844e-09,
+      "loss": 0.5164,
+      "step": 4528
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8485705843417365e-09,
+      "loss": 0.5321,
+      "step": 4529
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5955005725064597e-09,
+      "loss": 0.4899,
+      "step": 4530
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3542002647380934e-09,
+      "loss": 0.5199,
+      "step": 4531
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1246698030585878e-09,
+      "loss": 0.5229,
+      "step": 4532
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9069093225621004e-09,
+      "loss": 0.5214,
+      "step": 4533
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7009189514127778e-09,
+      "loss": 0.5145,
+      "step": 4534
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5066988108491942e-09,
+      "loss": 0.5002,
+      "step": 4535
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.324249015182133e-09,
+      "loss": 0.5624,
+      "step": 4536
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1535696717945855e-09,
+      "loss": 0.5329,
+      "step": 4537
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.946608811395308e-10,
+      "loss": 0.5241,
+      "step": 4538
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.475227367465977e-10,
+      "loss": 0.5175,
+      "step": 4539
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.121553252154023e-10,
+      "loss": 0.5233,
+      "step": 4540
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.885587262177694e-10,
+      "loss": 0.4927,
+      "step": 4541
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.767330124999525e-10,
+      "loss": 0.5116,
+      "step": 4542
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7667824987597243e-10,
+      "loss": 0.5066,
+      "step": 4543
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8839449723205847e-10,
+      "loss": 0.5185,
+      "step": 4544
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1188180653330947e-10,
+      "loss": 0.5039,
+      "step": 4545
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4714022281037133e-10,
+      "loss": 0.5175,
+      "step": 4546
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.416978416831868e-11,
+      "loss": 0.5343,
+      "step": 4547
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.2970521779993625e-11,
+      "loss": 0.5162,
+      "step": 4548
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.3542459897107907e-11,
+      "loss": 0.5256,
+      "step": 4549
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.8856158413611585e-12,
+      "loss": 0.5272,
+      "step": 4550
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.4997,
+      "step": 4551
+    },
+    {
+      "epoch": 3.0,
+      "step": 4551,
+      "total_flos": 1.9580865713219143e+23,
+      "train_loss": 0.5922896525828923,
+      "train_runtime": 35636.7623,
+      "train_samples_per_second": 49.045,
+      "train_steps_per_second": 0.128
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4551,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 1.9580865713219143e+23,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}