diff --git "a/ReCogDrive_VLM_8B/trainer_state.json" "b/ReCogDrive_VLM_8B/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/ReCogDrive_VLM_8B/trainer_state.json"
@@ -0,0 +1,20496 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.999176141044653,
+  "eval_steps": 500,
+  "global_step": 3411,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1695906432748539e-07,
+      "loss": 1.2939,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3391812865497077e-07,
+      "loss": 1.339,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5087719298245616e-07,
+      "loss": 1.2919,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6783625730994155e-07,
+      "loss": 1.302,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.847953216374269e-07,
+      "loss": 1.3463,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.017543859649123e-07,
+      "loss": 1.315,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.187134502923977e-07,
+      "loss": 1.3027,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.356725146198831e-07,
+      "loss": 1.2802,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 1.2884,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1695906432748538e-06,
+      "loss": 1.249,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2865497076023392e-06,
+      "loss": 1.1605,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4035087719298246e-06,
+      "loss": 1.2293,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.52046783625731e-06,
+      "loss": 1.1543,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6374269005847953e-06,
+      "loss": 1.0892,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 1.0496,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8713450292397662e-06,
+      "loss": 1.012,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9883040935672516e-06,
+      "loss": 1.0774,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 0.9941,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.0029,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3391812865497075e-06,
+      "loss": 0.9502,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.456140350877193e-06,
+      "loss": 0.9425,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5730994152046784e-06,
+      "loss": 0.9205,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.690058479532164e-06,
+      "loss": 0.9358,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8070175438596493e-06,
+      "loss": 0.8788,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9239766081871347e-06,
+      "loss": 0.8904,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.04093567251462e-06,
+      "loss": 0.8605,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.8398,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2748538011695906e-06,
+      "loss": 0.8182,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.391812865497076e-06,
+      "loss": 0.8385,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 0.8121,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.625730994152047e-06,
+      "loss": 0.8401,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7426900584795324e-06,
+      "loss": 0.8244,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.859649122807018e-06,
+      "loss": 0.7866,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.976608187134503e-06,
+      "loss": 0.7787,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.093567251461989e-06,
+      "loss": 0.7769,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.7946,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3274853801169596e-06,
+      "loss": 0.7561,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.7661,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.56140350877193e-06,
+      "loss": 0.7763,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.678362573099415e-06,
+      "loss": 0.7473,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7953216374269005e-06,
+      "loss": 0.7347,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.912280701754386e-06,
+      "loss": 0.7731,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.029239766081871e-06,
+      "loss": 0.7501,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.146198830409357e-06,
+      "loss": 0.7497,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.7213,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.380116959064328e-06,
+      "loss": 0.719,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.497076023391813e-06,
+      "loss": 0.7205,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6140350877192985e-06,
+      "loss": 0.7191,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.730994152046784e-06,
+      "loss": 0.7462,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.847953216374269e-06,
+      "loss": 0.7601,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.964912280701755e-06,
+      "loss": 0.709,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.08187134502924e-06,
+      "loss": 0.7179,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.198830409356725e-06,
+      "loss": 0.6803,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.7,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.432748538011696e-06,
+      "loss": 0.7122,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.549707602339181e-06,
+      "loss": 0.7033,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.6824,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.783625730994152e-06,
+      "loss": 0.7297,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.9005847953216375e-06,
+      "loss": 0.6916,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 0.7115,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.134502923976608e-06,
+      "loss": 0.7309,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.251461988304094e-06,
+      "loss": 0.725,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.6703,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.485380116959065e-06,
+      "loss": 0.705,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.60233918128655e-06,
+      "loss": 0.7035,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.719298245614036e-06,
+      "loss": 0.6965,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.836257309941521e-06,
+      "loss": 0.7027,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.953216374269006e-06,
+      "loss": 0.7036,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.070175438596492e-06,
+      "loss": 0.6914,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.187134502923977e-06,
+      "loss": 0.6797,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.304093567251463e-06,
+      "loss": 0.6852,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.7129,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.538011695906434e-06,
+      "loss": 0.6725,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.654970760233919e-06,
+      "loss": 0.6734,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.771929824561405e-06,
+      "loss": 0.6877,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.6641,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.005847953216374e-06,
+      "loss": 0.7079,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.12280701754386e-06,
+      "loss": 0.677,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.239766081871345e-06,
+      "loss": 0.6801,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.35672514619883e-06,
+      "loss": 0.675,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.657,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.590643274853801e-06,
+      "loss": 0.6852,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.707602339181286e-06,
+      "loss": 0.6923,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.824561403508772e-06,
+      "loss": 0.6634,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.941520467836257e-06,
+      "loss": 0.6658,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0058479532163743e-05,
+      "loss": 0.6605,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.017543859649123e-05,
+      "loss": 0.6818,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0292397660818714e-05,
+      "loss": 0.6562,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.04093567251462e-05,
+      "loss": 0.6729,
+      "step": 89
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.6722,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0643274853801172e-05,
+      "loss": 0.6715,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0760233918128655e-05,
+      "loss": 0.6564,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0877192982456142e-05,
+      "loss": 0.6771,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0994152046783626e-05,
+      "loss": 0.6849,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.6888,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1228070175438597e-05,
+      "loss": 0.655,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1345029239766083e-05,
+      "loss": 0.6614,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1461988304093568e-05,
+      "loss": 0.6697,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.6886,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1695906432748539e-05,
+      "loss": 0.6621,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1812865497076024e-05,
+      "loss": 0.6502,
+      "step": 101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.192982456140351e-05,
+      "loss": 0.6592,
+      "step": 102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2046783625730995e-05,
+      "loss": 0.6577,
+      "step": 103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.216374269005848e-05,
+      "loss": 0.6809,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2280701754385966e-05,
+      "loss": 0.6685,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.239766081871345e-05,
+      "loss": 0.6616,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2514619883040937e-05,
+      "loss": 0.6717,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.665,
+      "step": 108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2748538011695908e-05,
+      "loss": 0.6807,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2865497076023392e-05,
+      "loss": 0.6595,
+      "step": 110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2982456140350879e-05,
+      "loss": 0.6646,
+      "step": 111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3099415204678362e-05,
+      "loss": 0.6494,
+      "step": 112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.321637426900585e-05,
+      "loss": 0.6401,
+      "step": 113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.6702,
+      "step": 114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.345029239766082e-05,
+      "loss": 0.6597,
+      "step": 115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3567251461988304e-05,
+      "loss": 0.6676,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.6677,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3801169590643275e-05,
+      "loss": 0.6594,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3918128654970762e-05,
+      "loss": 0.6708,
+      "step": 119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 0.661,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4152046783625733e-05,
+      "loss": 0.6427,
+      "step": 121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4269005847953217e-05,
+      "loss": 0.6533,
+      "step": 122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4385964912280704e-05,
+      "loss": 0.6518,
+      "step": 123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4502923976608188e-05,
+      "loss": 0.6147,
+      "step": 124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4619883040935675e-05,
+      "loss": 0.674,
+      "step": 125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.6592,
+      "step": 126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4853801169590644e-05,
+      "loss": 0.6397,
+      "step": 127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.497076023391813e-05,
+      "loss": 0.6624,
+      "step": 128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5087719298245615e-05,
+      "loss": 0.6781,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.52046783625731e-05,
+      "loss": 0.6554,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5321637426900587e-05,
+      "loss": 0.659,
+      "step": 131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.543859649122807e-05,
+      "loss": 0.646,
+      "step": 132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.6337,
+      "step": 133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5672514619883042e-05,
+      "loss": 0.6398,
+      "step": 134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.6665,
+      "step": 135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5906432748538013e-05,
+      "loss": 0.6442,
+      "step": 136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.60233918128655e-05,
+      "loss": 0.6234,
+      "step": 137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6140350877192984e-05,
+      "loss": 0.6533,
+      "step": 138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.625730994152047e-05,
+      "loss": 0.6479,
+      "step": 139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6374269005847955e-05,
+      "loss": 0.627,
+      "step": 140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.649122807017544e-05,
+      "loss": 0.647,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6608187134502926e-05,
+      "loss": 0.6514,
+      "step": 142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.672514619883041e-05,
+      "loss": 0.6251,
+      "step": 143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.6316,
+      "step": 144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.695906432748538e-05,
+      "loss": 0.641,
+      "step": 145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7076023391812867e-05,
+      "loss": 0.6433,
+      "step": 146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.719298245614035e-05,
+      "loss": 0.656,
+      "step": 147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7309941520467838e-05,
+      "loss": 0.6205,
+      "step": 148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7426900584795322e-05,
+      "loss": 0.6417,
+      "step": 149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.754385964912281e-05,
+      "loss": 0.6229,
+      "step": 150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7660818713450293e-05,
+      "loss": 0.6498,
+      "step": 151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.6402,
+      "step": 152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.6234,
+      "step": 153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8011695906432747e-05,
+      "loss": 0.6435,
+      "step": 154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8128654970760235e-05,
+      "loss": 0.6509,
+      "step": 155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.824561403508772e-05,
+      "loss": 0.6509,
+      "step": 156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8362573099415205e-05,
+      "loss": 0.6559,
+      "step": 157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.847953216374269e-05,
+      "loss": 0.6422,
+      "step": 158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8596491228070176e-05,
+      "loss": 0.666,
+      "step": 159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.871345029239766e-05,
+      "loss": 0.6609,
+      "step": 160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8830409356725147e-05,
+      "loss": 0.6421,
+      "step": 161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.6278,
+      "step": 162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9064327485380118e-05,
+      "loss": 0.6424,
+      "step": 163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9181286549707602e-05,
+      "loss": 0.652,
+      "step": 164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.929824561403509e-05,
+      "loss": 0.6341,
+      "step": 165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9415204678362573e-05,
+      "loss": 0.6293,
+      "step": 166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.953216374269006e-05,
+      "loss": 0.6218,
+      "step": 167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9649122807017544e-05,
+      "loss": 0.6458,
+      "step": 168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.976608187134503e-05,
+      "loss": 0.6258,
+      "step": 169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9883040935672515e-05,
+      "loss": 0.628,
+      "step": 170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.6522,
+      "step": 171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0116959064327485e-05,
+      "loss": 0.6439,
+      "step": 172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0233918128654976e-05,
+      "loss": 0.6474,
+      "step": 173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.035087719298246e-05,
+      "loss": 0.6368,
+      "step": 174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0467836257309943e-05,
+      "loss": 0.6388,
+      "step": 175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0584795321637427e-05,
+      "loss": 0.6464,
+      "step": 176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0701754385964918e-05,
+      "loss": 0.6182,
+      "step": 177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.08187134502924e-05,
+      "loss": 0.6703,
+      "step": 178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0935672514619885e-05,
+      "loss": 0.6385,
+      "step": 179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 0.6361,
+      "step": 180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1169590643274856e-05,
+      "loss": 0.6427,
+      "step": 181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1286549707602343e-05,
+      "loss": 0.6357,
+      "step": 182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1403508771929827e-05,
+      "loss": 0.6261,
+      "step": 183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.152046783625731e-05,
+      "loss": 0.6611,
+      "step": 184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1637426900584798e-05,
+      "loss": 0.6369,
+      "step": 185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1754385964912285e-05,
+      "loss": 0.6452,
+      "step": 186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.187134502923977e-05,
+      "loss": 0.6413,
+      "step": 187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1988304093567252e-05,
+      "loss": 0.6493,
+      "step": 188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.210526315789474e-05,
+      "loss": 0.6219,
+      "step": 189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2222222222222227e-05,
+      "loss": 0.6127,
+      "step": 190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.233918128654971e-05,
+      "loss": 0.624,
+      "step": 191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2456140350877194e-05,
+      "loss": 0.645,
+      "step": 192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.257309941520468e-05,
+      "loss": 0.6615,
+      "step": 193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2690058479532165e-05,
+      "loss": 0.6459,
+      "step": 194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2807017543859652e-05,
+      "loss": 0.6216,
+      "step": 195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2923976608187136e-05,
+      "loss": 0.6433,
+      "step": 196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3040935672514623e-05,
+      "loss": 0.6661,
+      "step": 197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3157894736842107e-05,
+      "loss": 0.6278,
+      "step": 198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3274853801169594e-05,
+      "loss": 0.6402,
+      "step": 199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3391812865497078e-05,
+      "loss": 0.6431,
+      "step": 200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3508771929824565e-05,
+      "loss": 0.6383,
+      "step": 201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.362573099415205e-05,
+      "loss": 0.6209,
+      "step": 202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3742690058479532e-05,
+      "loss": 0.6218,
+      "step": 203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.385964912280702e-05,
+      "loss": 0.6348,
+      "step": 204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3976608187134507e-05,
+      "loss": 0.6502,
+      "step": 205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.409356725146199e-05,
+      "loss": 0.6248,
+      "step": 206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4210526315789474e-05,
+      "loss": 0.6537,
+      "step": 207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.432748538011696e-05,
+      "loss": 0.6405,
+      "step": 208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.444444444444445e-05,
+      "loss": 0.6363,
+      "step": 209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4561403508771932e-05,
+      "loss": 0.6533,
+      "step": 210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4678362573099416e-05,
+      "loss": 0.6763,
+      "step": 211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.47953216374269e-05,
+      "loss": 0.6248,
+      "step": 212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.491228070175439e-05,
+      "loss": 0.6348,
+      "step": 213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5029239766081874e-05,
+      "loss": 0.6342,
+      "step": 214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5146198830409358e-05,
+      "loss": 0.6356,
+      "step": 215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.526315789473684e-05,
+      "loss": 0.6437,
+      "step": 216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5380116959064332e-05,
+      "loss": 0.6148,
+      "step": 217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5497076023391816e-05,
+      "loss": 0.6473,
+      "step": 218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.56140350877193e-05,
+      "loss": 0.6464,
+      "step": 219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5730994152046783e-05,
+      "loss": 0.6235,
+      "step": 220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5847953216374274e-05,
+      "loss": 0.6407,
+      "step": 221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5964912280701757e-05,
+      "loss": 0.6464,
+      "step": 222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.608187134502924e-05,
+      "loss": 0.6377,
+      "step": 223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6198830409356725e-05,
+      "loss": 0.62,
+      "step": 224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6315789473684215e-05,
+      "loss": 0.6412,
+      "step": 225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.64327485380117e-05,
+      "loss": 0.6464,
+      "step": 226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6549707602339183e-05,
+      "loss": 0.6426,
+      "step": 227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.6573,
+      "step": 228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6783625730994157e-05,
+      "loss": 0.6324,
+      "step": 229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.690058479532164e-05,
+      "loss": 0.6215,
+      "step": 230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7017543859649125e-05,
+      "loss": 0.6267,
+      "step": 231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.713450292397661e-05,
+      "loss": 0.6448,
+      "step": 232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.72514619883041e-05,
+      "loss": 0.6619,
+      "step": 233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7368421052631583e-05,
+      "loss": 0.6322,
+      "step": 234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7485380116959066e-05,
+      "loss": 0.6272,
+      "step": 235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.760233918128655e-05,
+      "loss": 0.6234,
+      "step": 236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.771929824561404e-05,
+      "loss": 0.645,
+      "step": 237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7836257309941524e-05,
+      "loss": 0.646,
+      "step": 238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7953216374269008e-05,
+      "loss": 0.6821,
+      "step": 239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8070175438596492e-05,
+      "loss": 0.6322,
+      "step": 240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.818713450292398e-05,
+      "loss": 0.6447,
+      "step": 241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8304093567251466e-05,
+      "loss": 0.635,
+      "step": 242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.842105263157895e-05,
+      "loss": 0.6094,
+      "step": 243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8538011695906434e-05,
+      "loss": 0.6521,
+      "step": 244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.865497076023392e-05,
+      "loss": 0.6489,
+      "step": 245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8771929824561408e-05,
+      "loss": 0.6299,
+      "step": 246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.888888888888889e-05,
+      "loss": 0.639,
+      "step": 247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9005847953216375e-05,
+      "loss": 0.6402,
+      "step": 248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9122807017543863e-05,
+      "loss": 0.625,
+      "step": 249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.923976608187135e-05,
+      "loss": 0.6378,
+      "step": 250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9356725146198833e-05,
+      "loss": 0.6321,
+      "step": 251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9473684210526317e-05,
+      "loss": 0.6409,
+      "step": 252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9590643274853804e-05,
+      "loss": 0.6207,
+      "step": 253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9707602339181288e-05,
+      "loss": 0.6358,
+      "step": 254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9824561403508775e-05,
+      "loss": 0.6224,
+      "step": 255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.994152046783626e-05,
+      "loss": 0.6468,
+      "step": 256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0058479532163746e-05,
+      "loss": 0.6273,
+      "step": 257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.017543859649123e-05,
+      "loss": 0.6445,
+      "step": 258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0292397660818717e-05,
+      "loss": 0.6231,
+      "step": 259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.04093567251462e-05,
+      "loss": 0.6347,
+      "step": 260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.052631578947369e-05,
+      "loss": 0.6275,
+      "step": 261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0643274853801175e-05,
+      "loss": 0.6216,
+      "step": 262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.076023391812866e-05,
+      "loss": 0.6207,
+      "step": 263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.087719298245614e-05,
+      "loss": 0.6117,
+      "step": 264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.099415204678363e-05,
+      "loss": 0.6794,
+      "step": 265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.111111111111112e-05,
+      "loss": 0.6344,
+      "step": 266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.12280701754386e-05,
+      "loss": 0.6182,
+      "step": 267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1345029239766084e-05,
+      "loss": 0.6187,
+      "step": 268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1461988304093575e-05,
+      "loss": 0.6394,
+      "step": 269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.157894736842106e-05,
+      "loss": 0.6318,
+      "step": 270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.169590643274854e-05,
+      "loss": 0.6062,
+      "step": 271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1812865497076026e-05,
+      "loss": 0.65,
+      "step": 272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.192982456140351e-05,
+      "loss": 0.6323,
+      "step": 273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.2046783625731e-05,
+      "loss": 0.6366,
+      "step": 274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.2163742690058484e-05,
+      "loss": 0.6379,
+      "step": 275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.228070175438597e-05,
+      "loss": 0.6496,
+      "step": 276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.239766081871345e-05,
+      "loss": 0.6322,
+      "step": 277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.251461988304094e-05,
+      "loss": 0.6764,
+      "step": 278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.2631578947368426e-05,
+      "loss": 0.6493,
+      "step": 279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.274853801169591e-05,
+      "loss": 0.6256,
+      "step": 280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.286549707602339e-05,
+      "loss": 0.6452,
+      "step": 281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.298245614035088e-05,
+      "loss": 0.6227,
+      "step": 282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.309941520467837e-05,
+      "loss": 0.612,
+      "step": 283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.321637426900585e-05,
+      "loss": 0.6227,
+      "step": 284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.6415,
+      "step": 285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.345029239766082e-05,
+      "loss": 0.6379,
+      "step": 286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.356725146198831e-05,
+      "loss": 0.6215,
+      "step": 287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.368421052631579e-05,
+      "loss": 0.6228,
+      "step": 288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.380116959064328e-05,
+      "loss": 0.6412,
+      "step": 289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.391812865497076e-05,
+      "loss": 0.6058,
+      "step": 290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.403508771929825e-05,
+      "loss": 0.6541,
+      "step": 291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4152046783625735e-05,
+      "loss": 0.6338,
+      "step": 292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.426900584795322e-05,
+      "loss": 0.6359,
+      "step": 293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.43859649122807e-05,
+      "loss": 0.628,
+      "step": 294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4502923976608186e-05,
+      "loss": 0.6141,
+      "step": 295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4619883040935676e-05,
+      "loss": 0.6143,
+      "step": 296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.473684210526316e-05,
+      "loss": 0.6439,
+      "step": 297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4853801169590644e-05,
+      "loss": 0.6226,
+      "step": 298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.497076023391813e-05,
+      "loss": 0.6104,
+      "step": 299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.508771929824562e-05,
+      "loss": 0.6474,
+      "step": 300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.52046783625731e-05,
+      "loss": 0.635,
+      "step": 301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.5321637426900586e-05,
+      "loss": 0.6322,
+      "step": 302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.543859649122807e-05,
+      "loss": 0.6486,
+      "step": 303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.555555555555555e-05,
+      "loss": 0.63,
+      "step": 304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.5672514619883044e-05,
+      "loss": 0.6306,
+      "step": 305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.578947368421053e-05,
+      "loss": 0.6442,
+      "step": 306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.590643274853801e-05,
+      "loss": 0.6519,
+      "step": 307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6023391812865495e-05,
+      "loss": 0.6439,
+      "step": 308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6140350877192985e-05,
+      "loss": 0.6416,
+      "step": 309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.625730994152047e-05,
+      "loss": 0.6221,
+      "step": 310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.637426900584795e-05,
+      "loss": 0.6533,
+      "step": 311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.649122807017544e-05,
+      "loss": 0.6107,
+      "step": 312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.660818713450293e-05,
+      "loss": 0.6376,
+      "step": 313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.672514619883041e-05,
+      "loss": 0.6265,
+      "step": 314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.6568,
+      "step": 315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.695906432748538e-05,
+      "loss": 0.6377,
+      "step": 316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.707602339181287e-05,
+      "loss": 0.6328,
+      "step": 317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.719298245614035e-05,
+      "loss": 0.6441,
+      "step": 318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7309941520467836e-05,
+      "loss": 0.6209,
+      "step": 319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.742690058479532e-05,
+      "loss": 0.6204,
+      "step": 320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.754385964912281e-05,
+      "loss": 0.5931,
+      "step": 321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7660818713450294e-05,
+      "loss": 0.6344,
+      "step": 322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 0.6255,
+      "step": 323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.789473684210526e-05,
+      "loss": 0.6423,
+      "step": 324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.801169590643275e-05,
+      "loss": 0.6256,
+      "step": 325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8128654970760236e-05,
+      "loss": 0.6346,
+      "step": 326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.824561403508772e-05,
+      "loss": 0.6328,
+      "step": 327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8362573099415204e-05,
+      "loss": 0.6301,
+      "step": 328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8479532163742694e-05,
+      "loss": 0.624,
+      "step": 329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.859649122807018e-05,
+      "loss": 0.6272,
+      "step": 330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.871345029239766e-05,
+      "loss": 0.6221,
+      "step": 331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8830409356725145e-05,
+      "loss": 0.6474,
+      "step": 332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8947368421052636e-05,
+      "loss": 0.6224,
+      "step": 333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.906432748538012e-05,
+      "loss": 0.6481,
+      "step": 334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9181286549707604e-05,
+      "loss": 0.6185,
+      "step": 335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.929824561403509e-05,
+      "loss": 0.6344,
+      "step": 336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.941520467836258e-05,
+      "loss": 0.61,
+      "step": 337
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.953216374269006e-05,
+      "loss": 0.6313,
+      "step": 338
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9649122807017545e-05,
+      "loss": 0.632,
+      "step": 339
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.976608187134503e-05,
+      "loss": 0.6443,
+      "step": 340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.988304093567252e-05,
+      "loss": 0.6458,
+      "step": 341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4e-05,
+      "loss": 0.6404,
+      "step": 342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999998952133562e-05,
+      "loss": 0.6207,
+      "step": 343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999958085353465e-05,
+      "loss": 0.6213,
+      "step": 344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999990569208647e-05,
+      "loss": 0.6327,
+      "step": 345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999983234158953e-05,
+      "loss": 0.574,
+      "step": 346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999973803393952e-05,
+      "loss": 0.6333,
+      "step": 347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999962276923525e-05,
+      "loss": 0.6634,
+      "step": 348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999486547597514e-05,
+      "loss": 0.6373,
+      "step": 349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999329369169045e-05,
+      "loss": 0.6051,
+      "step": 350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999151234114554e-05,
+      "loss": 0.6257,
+      "step": 351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9998952142620694e-05,
+      "loss": 0.6451,
+      "step": 352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9998732094896084e-05,
+      "loss": 0.6276,
+      "step": 353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9998491091171315e-05,
+      "loss": 0.6363,
+      "step": 354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999822913169892e-05,
+      "loss": 0.6042,
+      "step": 355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.99979462167534e-05,
+      "loss": 0.6217,
+      "step": 356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9997642346631205e-05,
+      "loss": 0.6262,
+      "step": 357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999731752165076e-05,
+      "loss": 0.6127,
+      "step": 358
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999697174215243e-05,
+      "loss": 0.6254,
+      "step": 359
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9996605008498556e-05,
+      "loss": 0.6435,
+      "step": 360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9996217321073424e-05,
+      "loss": 0.6131,
+      "step": 361
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9995808680283266e-05,
+      "loss": 0.6314,
+      "step": 362
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9995379086556294e-05,
+      "loss": 0.6486,
+      "step": 363
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999492854034266e-05,
+      "loss": 0.6398,
+      "step": 364
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9994457042114485e-05,
+      "loss": 0.6292,
+      "step": 365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999396459236583e-05,
+      "loss": 0.5955,
+      "step": 366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999345119161271e-05,
+      "loss": 0.6451,
+      "step": 367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9992916840393115e-05,
+      "loss": 0.6385,
+      "step": 368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999236153926696e-05,
+      "loss": 0.6333,
+      "step": 369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999178528881613e-05,
+      "loss": 0.602,
+      "step": 370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999118808964446e-05,
+      "loss": 0.647,
+      "step": 371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9990569942377745e-05,
+      "loss": 0.6488,
+      "step": 372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9989930847663706e-05,
+      "loss": 0.6226,
+      "step": 373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998927080617203e-05,
+      "loss": 0.6462,
+      "step": 374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998858981859436e-05,
+      "loss": 0.6339,
+      "step": 375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998788788564428e-05,
+      "loss": 0.656,
+      "step": 376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9987165008057316e-05,
+      "loss": 0.6277,
+      "step": 377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998642118659095e-05,
+      "loss": 0.6277,
+      "step": 378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.99856564220246e-05,
+      "loss": 0.627,
+      "step": 379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998487071515966e-05,
+      "loss": 0.6377,
+      "step": 380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998406406681942e-05,
+      "loss": 0.6056,
+      "step": 381
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.998323647784915e-05,
+      "loss": 0.6432,
+      "step": 382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.998238794911605e-05,
+      "loss": 0.6052,
+      "step": 383
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9981518481509275e-05,
+      "loss": 0.5941,
+      "step": 384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9980628075939897e-05,
+      "loss": 0.6379,
+      "step": 385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997971673334095e-05,
+      "loss": 0.6323,
+      "step": 386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.99787844546674e-05,
+      "loss": 0.6133,
+      "step": 387
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997783124089614e-05,
+      "loss": 0.6205,
+      "step": 388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997685709302603e-05,
+      "loss": 0.6374,
+      "step": 389
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997586201207783e-05,
+      "loss": 0.6521,
+      "step": 390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9974845999094266e-05,
+      "loss": 0.6112,
+      "step": 391
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9973809055139966e-05,
+      "loss": 0.6453,
+      "step": 392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9972751181301524e-05,
+      "loss": 0.6472,
+      "step": 393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9971672378687444e-05,
+      "loss": 0.61,
+      "step": 394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9970572648428166e-05,
+      "loss": 0.6307,
+      "step": 395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9969451991676064e-05,
+      "loss": 0.6215,
+      "step": 396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996831040960543e-05,
+      "loss": 0.6329,
+      "step": 397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9967147903412505e-05,
+      "loss": 0.637,
+      "step": 398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996596447431542e-05,
+      "loss": 0.5999,
+      "step": 399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996476012355427e-05,
+      "loss": 0.6309,
+      "step": 400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996353485239104e-05,
+      "loss": 0.6327,
+      "step": 401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996228866210965e-05,
+      "loss": 0.6593,
+      "step": 402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996102155401595e-05,
+      "loss": 0.6113,
+      "step": 403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995973352943769e-05,
+      "loss": 0.6218,
+      "step": 404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995842458972456e-05,
+      "loss": 0.6456,
+      "step": 405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995709473624815e-05,
+      "loss": 0.6196,
+      "step": 406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995574397040196e-05,
+      "loss": 0.6104,
+      "step": 407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9954372293601415e-05,
+      "loss": 0.6342,
+      "step": 408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995297970728386e-05,
+      "loss": 0.6182,
+      "step": 409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995156621290853e-05,
+      "loss": 0.6021,
+      "step": 410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995013181195658e-05,
+      "loss": 0.6045,
+      "step": 411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.994867650593107e-05,
+      "loss": 0.6463,
+      "step": 412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.994720029635697e-05,
+      "loss": 0.6105,
+      "step": 413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9945703184781146e-05,
+      "loss": 0.632,
+      "step": 414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.994418517277238e-05,
+      "loss": 0.6326,
+      "step": 415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.994264626192132e-05,
+      "loss": 0.6122,
+      "step": 416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9941086453840576e-05,
+      "loss": 0.6127,
+      "step": 417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993950575016459e-05,
+      "loss": 0.6196,
+      "step": 418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9937904152549746e-05,
+      "loss": 0.6296,
+      "step": 419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9936281662674296e-05,
+      "loss": 0.6558,
+      "step": 420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993463828223839e-05,
+      "loss": 0.6173,
+      "step": 421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993297401296407e-05,
+      "loss": 0.6299,
+      "step": 422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993128885659528e-05,
+      "loss": 0.6305,
+      "step": 423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9929582814897834e-05,
+      "loss": 0.6174,
+      "step": 424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9927855889659424e-05,
+      "loss": 0.6134,
+      "step": 425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.992610808268965e-05,
+      "loss": 0.6292,
+      "step": 426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.992433939581999e-05,
+      "loss": 0.6392,
+      "step": 427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9922549830903756e-05,
+      "loss": 0.6137,
+      "step": 428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.992073938981621e-05,
+      "loss": 0.6224,
+      "step": 429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991890807445443e-05,
+      "loss": 0.6156,
+      "step": 430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.99170558867374e-05,
+      "loss": 0.6221,
+      "step": 431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991518282860597e-05,
+      "loss": 0.6268,
+      "step": 432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991328890202284e-05,
+      "loss": 0.6469,
+      "step": 433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.99113741089726e-05,
+      "loss": 0.6216,
+      "step": 434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9909438451461695e-05,
+      "loss": 0.6362,
+      "step": 435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9907481931518444e-05,
+      "loss": 0.6113,
+      "step": 436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9905504551193006e-05,
+      "loss": 0.607,
+      "step": 437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.990350631255742e-05,
+      "loss": 0.6188,
+      "step": 438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.990148721770557e-05,
+      "loss": 0.6179,
+      "step": 439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.98994472687532e-05,
+      "loss": 0.6312,
+      "step": 440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.98973864678379e-05,
+      "loss": 0.6423,
+      "step": 441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9895304817119127e-05,
+      "loss": 0.6608,
+      "step": 442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.989320231877815e-05,
+      "loss": 0.6299,
+      "step": 443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.989107897501813e-05,
+      "loss": 0.627,
+      "step": 444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.988893478806403e-05,
+      "loss": 0.6267,
+      "step": 445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.988676976016268e-05,
+      "loss": 0.6135,
+      "step": 446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.988458389358274e-05,
+      "loss": 0.6305,
+      "step": 447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.98823771906147e-05,
+      "loss": 0.6168,
+      "step": 448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9880149653570905e-05,
+      "loss": 0.5975,
+      "step": 449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.98779012847855e-05,
+      "loss": 0.6192,
+      "step": 450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9875632086614475e-05,
+      "loss": 0.6356,
+      "step": 451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9873342061435664e-05,
+      "loss": 0.6157,
+      "step": 452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.987103121164869e-05,
+      "loss": 0.6392,
+      "step": 453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986869953967502e-05,
+      "loss": 0.6141,
+      "step": 454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986634704795794e-05,
+      "loss": 0.6282,
+      "step": 455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986397373896254e-05,
+      "loss": 0.6179,
+      "step": 456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986157961517573e-05,
+      "loss": 0.6139,
+      "step": 457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.985916467910624e-05,
+      "loss": 0.6368,
+      "step": 458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.98567289332846e-05,
+      "loss": 0.6256,
+      "step": 459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.985427238026314e-05,
+      "loss": 0.6149,
+      "step": 460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9851795022616e-05,
+      "loss": 0.6588,
+      "step": 461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.984929686293913e-05,
+      "loss": 0.6137,
+      "step": 462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.984677790385025e-05,
+      "loss": 0.6095,
+      "step": 463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.984423814798891e-05,
+      "loss": 0.6505,
+      "step": 464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.984167759801643e-05,
+      "loss": 0.6038,
+      "step": 465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.983909625661591e-05,
+      "loss": 0.6037,
+      "step": 466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.983649412649227e-05,
+      "loss": 0.5967,
+      "step": 467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9833871210372185e-05,
+      "loss": 0.6118,
+      "step": 468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.983122751100412e-05,
+      "loss": 0.6121,
+      "step": 469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.982856303115833e-05,
+      "loss": 0.6169,
+      "step": 470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.982587777362683e-05,
+      "loss": 0.5865,
+      "step": 471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9823171741223396e-05,
+      "loss": 0.6033,
+      "step": 472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9820444936783606e-05,
+      "loss": 0.6411,
+      "step": 473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.981769736316478e-05,
+      "loss": 0.6014,
+      "step": 474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.981492902324601e-05,
+      "loss": 0.6127,
+      "step": 475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9812139919928147e-05,
+      "loss": 0.611,
+      "step": 476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9809330056133796e-05,
+      "loss": 0.6095,
+      "step": 477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.980649943480732e-05,
+      "loss": 0.618,
+      "step": 478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9803648058914825e-05,
+      "loss": 0.6166,
+      "step": 479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.980077593144419e-05,
+      "loss": 0.6133,
+      "step": 480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.979788305540501e-05,
+      "loss": 0.6098,
+      "step": 481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.979496943382863e-05,
+      "loss": 0.6141,
+      "step": 482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9792035069768136e-05,
+      "loss": 0.6248,
+      "step": 483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9789079966298355e-05,
+      "loss": 0.6135,
+      "step": 484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.978610412651584e-05,
+      "loss": 0.6303,
+      "step": 485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.978310755353888e-05,
+      "loss": 0.6129,
+      "step": 486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.978009025050746e-05,
+      "loss": 0.6048,
+      "step": 487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.977705222058334e-05,
+      "loss": 0.5992,
+      "step": 488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.977399346694995e-05,
+      "loss": 0.6127,
+      "step": 489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9770913992812465e-05,
+      "loss": 0.6028,
+      "step": 490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.976781380139775e-05,
+      "loss": 0.5979,
+      "step": 491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.976469289595441e-05,
+      "loss": 0.6308,
+      "step": 492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.976155127975272e-05,
+      "loss": 0.6015,
+      "step": 493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.975838895608469e-05,
+      "loss": 0.6235,
+      "step": 494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.975520592826399e-05,
+      "loss": 0.6451,
+      "step": 495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9752002199626035e-05,
+      "loss": 0.6356,
+      "step": 496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.974877777352789e-05,
+      "loss": 0.6037,
+      "step": 497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.974553265334832e-05,
+      "loss": 0.6023,
+      "step": 498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.974226684248779e-05,
+      "loss": 0.6229,
+      "step": 499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9738980344368426e-05,
+      "loss": 0.6678,
+      "step": 500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.973567316243404e-05,
+      "loss": 0.6183,
+      "step": 501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9732345300150105e-05,
+      "loss": 0.6441,
+      "step": 502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9728996761003796e-05,
+      "loss": 0.6262,
+      "step": 503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.972562754850392e-05,
+      "loss": 0.6397,
+      "step": 504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.972223766618097e-05,
+      "loss": 0.6166,
+      "step": 505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.971882711758709e-05,
+      "loss": 0.6112,
+      "step": 506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.971539590629608e-05,
+      "loss": 0.6156,
+      "step": 507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9711944035903384e-05,
+      "loss": 0.6136,
+      "step": 508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9708471510026105e-05,
+      "loss": 0.6205,
+      "step": 509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.970497833230299e-05,
+      "loss": 0.6357,
+      "step": 510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.970146450639442e-05,
+      "loss": 0.6137,
+      "step": 511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9697930035982414e-05,
+      "loss": 0.6002,
+      "step": 512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.969437492477062e-05,
+      "loss": 0.6107,
+      "step": 513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.969079917648433e-05,
+      "loss": 0.5844,
+      "step": 514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.968720279487045e-05,
+      "loss": 0.5973,
+      "step": 515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.968358578369749e-05,
+      "loss": 0.6205,
+      "step": 516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.967994814675562e-05,
+      "loss": 0.596,
+      "step": 517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.967628988785658e-05,
+      "loss": 0.6072,
+      "step": 518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.967261101083375e-05,
+      "loss": 0.6047,
+      "step": 519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.966891151954209e-05,
+      "loss": 0.6222,
+      "step": 520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9665191417858175e-05,
+      "loss": 0.6079,
+      "step": 521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.966145070968018e-05,
+      "loss": 0.6196,
+      "step": 522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.965768939892786e-05,
+      "loss": 0.62,
+      "step": 523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.965390748954257e-05,
+      "loss": 0.6495,
+      "step": 524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.965010498548726e-05,
+      "loss": 0.6075,
+      "step": 525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.964628189074642e-05,
+      "loss": 0.633,
+      "step": 526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.964243820932616e-05,
+      "loss": 0.6455,
+      "step": 527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.963857394525413e-05,
+      "loss": 0.6117,
+      "step": 528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.963468910257959e-05,
+      "loss": 0.6348,
+      "step": 529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.963078368537331e-05,
+      "loss": 0.5959,
+      "step": 530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9626857697727657e-05,
+      "loss": 0.6217,
+      "step": 531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9622911143756534e-05,
+      "loss": 0.6279,
+      "step": 532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.961894402759541e-05,
+      "loss": 0.6028,
+      "step": 533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.96149563534013e-05,
+      "loss": 0.6094,
+      "step": 534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.961094812535274e-05,
+      "loss": 0.6394,
+      "step": 535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.960691934764982e-05,
+      "loss": 0.608,
+      "step": 536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.960287002451416e-05,
+      "loss": 0.6354,
+      "step": 537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9598800160188924e-05,
+      "loss": 0.6395,
+      "step": 538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.959470975893877e-05,
+      "loss": 0.6251,
+      "step": 539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9590598825049896e-05,
+      "loss": 0.6178,
+      "step": 540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.958646736283001e-05,
+      "loss": 0.6096,
+      "step": 541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.958231537660834e-05,
+      "loss": 0.5936,
+      "step": 542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.957814287073561e-05,
+      "loss": 0.6112,
+      "step": 543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9573949849584046e-05,
+      "loss": 0.6066,
+      "step": 544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.956973631754737e-05,
+      "loss": 0.606,
+      "step": 545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.956550227904082e-05,
+      "loss": 0.6079,
+      "step": 546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.956124773850108e-05,
+      "loss": 0.6208,
+      "step": 547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9556972700386354e-05,
+      "loss": 0.5972,
+      "step": 548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.955267716917631e-05,
+      "loss": 0.6015,
+      "step": 549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.954836114937208e-05,
+      "loss": 0.6236,
+      "step": 550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.954402464549628e-05,
+      "loss": 0.6075,
+      "step": 551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9539667662093e-05,
+      "loss": 0.5859,
+      "step": 552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9535290203727775e-05,
+      "loss": 0.6095,
+      "step": 553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9530892274987575e-05,
+      "loss": 0.6203,
+      "step": 554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.952647388048086e-05,
+      "loss": 0.6282,
+      "step": 555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9522035024837514e-05,
+      "loss": 0.6342,
+      "step": 556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9517575712708864e-05,
+      "loss": 0.6078,
+      "step": 557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.951309594876768e-05,
+      "loss": 0.6142,
+      "step": 558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.950859573770815e-05,
+      "loss": 0.6057,
+      "step": 559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.950407508424588e-05,
+      "loss": 0.6161,
+      "step": 560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.949953399311794e-05,
+      "loss": 0.5996,
+      "step": 561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9494972469082764e-05,
+      "loss": 0.6373,
+      "step": 562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9490390516920237e-05,
+      "loss": 0.6454,
+      "step": 563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.948578814143161e-05,
+      "loss": 0.6098,
+      "step": 564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.948116534743959e-05,
+      "loss": 0.6147,
+      "step": 565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.947652213978822e-05,
+      "loss": 0.615,
+      "step": 566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.947185852334297e-05,
+      "loss": 0.6271,
+      "step": 567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.946717450299069e-05,
+      "loss": 0.5899,
+      "step": 568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9462470083639606e-05,
+      "loss": 0.6066,
+      "step": 569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.945774527021932e-05,
+      "loss": 0.6159,
+      "step": 570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.945300006768081e-05,
+      "loss": 0.6037,
+      "step": 571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9448234480996406e-05,
+      "loss": 0.6227,
+      "step": 572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9443448515159815e-05,
+      "loss": 0.5969,
+      "step": 573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9438642175186085e-05,
+      "loss": 0.6184,
+      "step": 574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.943381546611162e-05,
+      "loss": 0.6256,
+      "step": 575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9428968392994174e-05,
+      "loss": 0.5803,
+      "step": 576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9424100960912814e-05,
+      "loss": 0.6105,
+      "step": 577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.941921317496797e-05,
+      "loss": 0.6101,
+      "step": 578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.941430504028139e-05,
+      "loss": 0.6135,
+      "step": 579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.940937656199614e-05,
+      "loss": 0.624,
+      "step": 580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.940442774527661e-05,
+      "loss": 0.6055,
+      "step": 581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9399458595308504e-05,
+      "loss": 0.6156,
+      "step": 582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.939446911729881e-05,
+      "loss": 0.6175,
+      "step": 583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.938945931647585e-05,
+      "loss": 0.5873,
+      "step": 584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.938442919808922e-05,
+      "loss": 0.6092,
+      "step": 585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9379378767409815e-05,
+      "loss": 0.6007,
+      "step": 586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9374308029729815e-05,
+      "loss": 0.5989,
+      "step": 587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.936921699036267e-05,
+      "loss": 0.6135,
+      "step": 588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9364105654643107e-05,
+      "loss": 0.6246,
+      "step": 589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.935897402792713e-05,
+      "loss": 0.616,
+      "step": 590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9353822115591994e-05,
+      "loss": 0.6148,
+      "step": 591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9348649923036216e-05,
+      "loss": 0.6075,
+      "step": 592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9343457455679564e-05,
+      "loss": 0.6058,
+      "step": 593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.933824471896305e-05,
+      "loss": 0.6076,
+      "step": 594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9333011718348925e-05,
+      "loss": 0.6268,
+      "step": 595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9327758459320676e-05,
+      "loss": 0.6067,
+      "step": 596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.932248494738302e-05,
+      "loss": 0.6099,
+      "step": 597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9317191188061886e-05,
+      "loss": 0.6223,
+      "step": 598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.931187718690442e-05,
+      "loss": 0.6287,
+      "step": 599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.930654294947901e-05,
+      "loss": 0.628,
+      "step": 600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.930118848137521e-05,
+      "loss": 0.637,
+      "step": 601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.929581378820378e-05,
+      "loss": 0.5907,
+      "step": 602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.929041887559668e-05,
+      "loss": 0.6223,
+      "step": 603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9285003749207075e-05,
+      "loss": 0.6375,
+      "step": 604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.927956841470929e-05,
+      "loss": 0.5604,
+      "step": 605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.927411287779882e-05,
+      "loss": 0.6152,
+      "step": 606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.926863714419234e-05,
+      "loss": 0.6037,
+      "step": 607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9263141219627694e-05,
+      "loss": 0.6223,
+      "step": 608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9257625109863877e-05,
+      "loss": 0.6106,
+      "step": 609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.925208882068103e-05,
+      "loss": 0.6099,
+      "step": 610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9246532357880445e-05,
+      "loss": 0.6119,
+      "step": 611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9240955727284555e-05,
+      "loss": 0.6233,
+      "step": 612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9235358934736926e-05,
+      "loss": 0.5865,
+      "step": 613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9229741986102246e-05,
+      "loss": 0.6122,
+      "step": 614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.922410488726633e-05,
+      "loss": 0.6034,
+      "step": 615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.921844764413609e-05,
+      "loss": 0.6275,
+      "step": 616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.921277026263959e-05,
+      "loss": 0.6133,
+      "step": 617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.920707274872595e-05,
+      "loss": 0.6021,
+      "step": 618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.920135510836539e-05,
+      "loss": 0.6178,
+      "step": 619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.919561734754926e-05,
+      "loss": 0.6142,
+      "step": 620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.6005,
+      "step": 621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9184081488620955e-05,
+      "loss": 0.5873,
+      "step": 622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9178283402596824e-05,
+      "loss": 0.5962,
+      "step": 623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.917246522029317e-05,
+      "loss": 0.6125,
+      "step": 624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.916662694780668e-05,
+      "loss": 0.6332,
+      "step": 625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.916076859125509e-05,
+      "loss": 0.6207,
+      "step": 626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9154890156777165e-05,
+      "loss": 0.6332,
+      "step": 627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.914899165053272e-05,
+      "loss": 0.6363,
+      "step": 628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9143073078702596e-05,
+      "loss": 0.6103,
+      "step": 629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.913713444748868e-05,
+      "loss": 0.6092,
+      "step": 630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9131175763113857e-05,
+      "loss": 0.5963,
+      "step": 631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.912519703182203e-05,
+      "loss": 0.5786,
+      "step": 632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.911919825987812e-05,
+      "loss": 0.6072,
+      "step": 633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.911317945356802e-05,
+      "loss": 0.6264,
+      "step": 634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.910714061919866e-05,
+      "loss": 0.6134,
+      "step": 635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9101081763097924e-05,
+      "loss": 0.6022,
+      "step": 636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.909500289161466e-05,
+      "loss": 0.621,
+      "step": 637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9088904011118746e-05,
+      "loss": 0.5764,
+      "step": 638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.908278512800098e-05,
+      "loss": 0.5921,
+      "step": 639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9076646248673136e-05,
+      "loss": 0.6068,
+      "step": 640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9070487379567935e-05,
+      "loss": 0.5985,
+      "step": 641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9064308527139056e-05,
+      "loss": 0.6266,
+      "step": 642
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9058109697861103e-05,
+      "loss": 0.6002,
+      "step": 643
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.905189089822963e-05,
+      "loss": 0.5963,
+      "step": 644
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9045652134761105e-05,
+      "loss": 0.6281,
+      "step": 645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9039393413992915e-05,
+      "loss": 0.6115,
+      "step": 646
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9033114742483375e-05,
+      "loss": 0.6103,
+      "step": 647
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9026816126811675e-05,
+      "loss": 0.6051,
+      "step": 648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9020497573577944e-05,
+      "loss": 0.6378,
+      "step": 649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9014159089403167e-05,
+      "loss": 0.6029,
+      "step": 650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.900780068092924e-05,
+      "loss": 0.6134,
+      "step": 651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.900142235481891e-05,
+      "loss": 0.5938,
+      "step": 652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.899502411775583e-05,
+      "loss": 0.6466,
+      "step": 653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.898860597644448e-05,
+      "loss": 0.6286,
+      "step": 654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.898216793761023e-05,
+      "loss": 0.6002,
+      "step": 655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8975710007999265e-05,
+      "loss": 0.5993,
+      "step": 656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.896923219437866e-05,
+      "loss": 0.5992,
+      "step": 657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8962734503536285e-05,
+      "loss": 0.6045,
+      "step": 658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.895621694228084e-05,
+      "loss": 0.5933,
+      "step": 659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8949679517441884e-05,
+      "loss": 0.5871,
+      "step": 660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.894312223586974e-05,
+      "loss": 0.617,
+      "step": 661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8936545104435576e-05,
+      "loss": 0.609,
+      "step": 662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.892994813003135e-05,
+      "loss": 0.5752,
+      "step": 663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89233313195698e-05,
+      "loss": 0.6128,
+      "step": 664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8916694679984474e-05,
+      "loss": 0.595,
+      "step": 665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8910038218229666e-05,
+      "loss": 0.6309,
+      "step": 666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.890336194128048e-05,
+      "loss": 0.6106,
+      "step": 667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.889666585613274e-05,
+      "loss": 0.6063,
+      "step": 668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.888994996980306e-05,
+      "loss": 0.6157,
+      "step": 669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8883214289328795e-05,
+      "loss": 0.598,
+      "step": 670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.887645882176803e-05,
+      "loss": 0.6026,
+      "step": 671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.886968357419961e-05,
+      "loss": 0.609,
+      "step": 672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8862888553723074e-05,
+      "loss": 0.6103,
+      "step": 673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.885607376745869e-05,
+      "loss": 0.6108,
+      "step": 674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.884923922254745e-05,
+      "loss": 0.6084,
+      "step": 675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8842384926151065e-05,
+      "loss": 0.5945,
+      "step": 676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8835510885451894e-05,
+      "loss": 0.6155,
+      "step": 677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.882861710765302e-05,
+      "loss": 0.5982,
+      "step": 678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8821703599978213e-05,
+      "loss": 0.6103,
+      "step": 679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.881477036967189e-05,
+      "loss": 0.6164,
+      "step": 680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8807817423999165e-05,
+      "loss": 0.5863,
+      "step": 681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.880084477024579e-05,
+      "loss": 0.6107,
+      "step": 682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.6207,
+      "step": 683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.878684036774337e-05,
+      "loss": 0.6139,
+      "step": 684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.877980863366906e-05,
+      "loss": 0.6046,
+      "step": 685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.877275722086358e-05,
+      "loss": 0.5881,
+      "step": 686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.876568613671586e-05,
+      "loss": 0.6302,
+      "step": 687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.875859538863545e-05,
+      "loss": 0.5874,
+      "step": 688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.875148498405252e-05,
+      "loss": 0.6014,
+      "step": 689
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.87443549304178e-05,
+      "loss": 0.6051,
+      "step": 690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.873720523520265e-05,
+      "loss": 0.6273,
+      "step": 691
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8730035905898985e-05,
+      "loss": 0.5985,
+      "step": 692
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.872284695001933e-05,
+      "loss": 0.6238,
+      "step": 693
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.871563837509672e-05,
+      "loss": 0.6088,
+      "step": 694
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.870841018868479e-05,
+      "loss": 0.5912,
+      "step": 695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8701162398357715e-05,
+      "loss": 0.6062,
+      "step": 696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8693895011710216e-05,
+      "loss": 0.6125,
+      "step": 697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.868660803635754e-05,
+      "loss": 0.6135,
+      "step": 698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.867930147993546e-05,
+      "loss": 0.6012,
+      "step": 699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8671975350100276e-05,
+      "loss": 0.617,
+      "step": 700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8664629654528794e-05,
+      "loss": 0.5881,
+      "step": 701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8657264400918314e-05,
+      "loss": 0.6085,
+      "step": 702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.864987959698665e-05,
+      "loss": 0.6257,
+      "step": 703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8642475250472074e-05,
+      "loss": 0.5957,
+      "step": 704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.863505136913337e-05,
+      "loss": 0.5977,
+      "step": 705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8627607960749765e-05,
+      "loss": 0.6045,
+      "step": 706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.862014503312096e-05,
+      "loss": 0.6134,
+      "step": 707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8612662594067106e-05,
+      "loss": 0.5847,
+      "step": 708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.860516065142879e-05,
+      "loss": 0.6275,
+      "step": 709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8597639213067056e-05,
+      "loss": 0.6264,
+      "step": 710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.859009828686337e-05,
+      "loss": 0.6258,
+      "step": 711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.858253788071961e-05,
+      "loss": 0.6017,
+      "step": 712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.857495800255807e-05,
+      "loss": 0.5859,
+      "step": 713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.856735866032145e-05,
+      "loss": 0.6154,
+      "step": 714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8559739861972856e-05,
+      "loss": 0.6082,
+      "step": 715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8552101615495755e-05,
+      "loss": 0.6113,
+      "step": 716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.854444392889402e-05,
+      "loss": 0.6067,
+      "step": 717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.853676681019188e-05,
+      "loss": 0.5829,
+      "step": 718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.852907026743394e-05,
+      "loss": 0.5979,
+      "step": 719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8521354308685126e-05,
+      "loss": 0.6017,
+      "step": 720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8513618942030754e-05,
+      "loss": 0.5936,
+      "step": 721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.850586417557645e-05,
+      "loss": 0.6145,
+      "step": 722
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8498090017448156e-05,
+      "loss": 0.6161,
+      "step": 723
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.849029647579218e-05,
+      "loss": 0.634,
+      "step": 724
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.848248355877509e-05,
+      "loss": 0.5958,
+      "step": 725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8474651274583794e-05,
+      "loss": 0.5982,
+      "step": 726
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8466799631425474e-05,
+      "loss": 0.6212,
+      "step": 727
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.84589286375276e-05,
+      "loss": 0.5741,
+      "step": 728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.845103830113793e-05,
+      "loss": 0.5927,
+      "step": 729
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.844312863052447e-05,
+      "loss": 0.6216,
+      "step": 730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8435199633975516e-05,
+      "loss": 0.584,
+      "step": 731
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.842725131979958e-05,
+      "loss": 0.606,
+      "step": 732
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.841928369632545e-05,
+      "loss": 0.593,
+      "step": 733
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.841129677190212e-05,
+      "loss": 0.6055,
+      "step": 734
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.840329055489882e-05,
+      "loss": 0.5987,
+      "step": 735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8395265053705005e-05,
+      "loss": 0.6219,
+      "step": 736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.838722027673032e-05,
+      "loss": 0.6146,
+      "step": 737
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.837915623240462e-05,
+      "loss": 0.5985,
+      "step": 738
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.837107292917794e-05,
+      "loss": 0.6057,
+      "step": 739
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8362970375520514e-05,
+      "loss": 0.5851,
+      "step": 740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8354848579922726e-05,
+      "loss": 0.6022,
+      "step": 741
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.834670755089514e-05,
+      "loss": 0.6012,
+      "step": 742
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.833854729696846e-05,
+      "loss": 0.6102,
+      "step": 743
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.833036782669355e-05,
+      "loss": 0.613,
+      "step": 744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.832216914864139e-05,
+      "loss": 0.602,
+      "step": 745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.831395127140312e-05,
+      "loss": 0.5837,
+      "step": 746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.830571420358996e-05,
+      "loss": 0.5918,
+      "step": 747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8297457953833264e-05,
+      "loss": 0.6105,
+      "step": 748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.828918253078448e-05,
+      "loss": 0.6213,
+      "step": 749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.828088794311514e-05,
+      "loss": 0.5688,
+      "step": 750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8272574199516875e-05,
+      "loss": 0.5682,
+      "step": 751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8264241308701365e-05,
+      "loss": 0.6327,
+      "step": 752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.825588927940037e-05,
+      "loss": 0.6008,
+      "step": 753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8247518120365705e-05,
+      "loss": 0.6419,
+      "step": 754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.823912784036923e-05,
+      "loss": 0.5878,
+      "step": 755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.823071844820283e-05,
+      "loss": 0.6025,
+      "step": 756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8222289952678425e-05,
+      "loss": 0.5966,
+      "step": 757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.821384236262796e-05,
+      "loss": 0.5961,
+      "step": 758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.820537568690338e-05,
+      "loss": 0.5881,
+      "step": 759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8196889934376617e-05,
+      "loss": 0.5934,
+      "step": 760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8188385113939613e-05,
+      "loss": 0.615,
+      "step": 761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.81798612345043e-05,
+      "loss": 0.5991,
+      "step": 762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8171318305002545e-05,
+      "loss": 0.5863,
+      "step": 763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.816275633438621e-05,
+      "loss": 0.5948,
+      "step": 764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.815417533162709e-05,
+      "loss": 0.6172,
+      "step": 765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8145575305716926e-05,
+      "loss": 0.617,
+      "step": 766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.81369562656674e-05,
+      "loss": 0.6125,
+      "step": 767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.812831822051013e-05,
+      "loss": 0.5964,
+      "step": 768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.81196611792966e-05,
+      "loss": 0.629,
+      "step": 769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8110985151098264e-05,
+      "loss": 0.6121,
+      "step": 770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.810229014500643e-05,
+      "loss": 0.6258,
+      "step": 771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.80935761701323e-05,
+      "loss": 0.635,
+      "step": 772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.808484323560696e-05,
+      "loss": 0.6078,
+      "step": 773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.807609135058135e-05,
+      "loss": 0.5895,
+      "step": 774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.806732052422629e-05,
+      "loss": 0.5912,
+      "step": 775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.805853076573243e-05,
+      "loss": 0.584,
+      "step": 776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.804972208431026e-05,
+      "loss": 0.591,
+      "step": 777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.80408944891901e-05,
+      "loss": 0.6038,
+      "step": 778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8032047989622094e-05,
+      "loss": 0.6062,
+      "step": 779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8023182594876195e-05,
+      "loss": 0.6202,
+      "step": 780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.801429831424215e-05,
+      "loss": 0.6107,
+      "step": 781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.800539515702949e-05,
+      "loss": 0.6097,
+      "step": 782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.799647313256755e-05,
+      "loss": 0.5888,
+      "step": 783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.798753225020541e-05,
+      "loss": 0.5838,
+      "step": 784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.797857251931192e-05,
+      "loss": 0.5963,
+      "step": 785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.796959394927569e-05,
+      "loss": 0.6271,
+      "step": 786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.7960596549505055e-05,
+      "loss": 0.6433,
+      "step": 787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.7951580329428084e-05,
+      "loss": 0.5978,
+      "step": 788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.794254529849259e-05,
+      "loss": 0.5942,
+      "step": 789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.7933491466166054e-05,
+      "loss": 0.6043,
+      "step": 790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.7924418841935694e-05,
+      "loss": 0.5926,
+      "step": 791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.791532743530842e-05,
+      "loss": 0.5956,
+      "step": 792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.790621725581079e-05,
+      "loss": 0.573,
+      "step": 793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.789708831298907e-05,
+      "loss": 0.5887,
+      "step": 794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.788794061640917e-05,
+      "loss": 0.5858,
+      "step": 795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.787877417565665e-05,
+      "loss": 0.6169,
+      "step": 796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.786958900033672e-05,
+      "loss": 0.6175,
+      "step": 797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.7860385100074225e-05,
+      "loss": 0.6158,
+      "step": 798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.7851162484513604e-05,
+      "loss": 0.5847,
+      "step": 799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.784192116331895e-05,
+      "loss": 0.6182,
+      "step": 800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.783266114617391e-05,
+      "loss": 0.6033,
+      "step": 801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.782338244278175e-05,
+      "loss": 0.6243,
+      "step": 802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.781408506286532e-05,
+      "loss": 0.6344,
+      "step": 803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7804769016167036e-05,
+      "loss": 0.6063,
+      "step": 804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.779543431244886e-05,
+      "loss": 0.6091,
+      "step": 805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.778608096149232e-05,
+      "loss": 0.6014,
+      "step": 806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7776708973098476e-05,
+      "loss": 0.5934,
+      "step": 807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.776731835708792e-05,
+      "loss": 0.6073,
+      "step": 808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.775790912330077e-05,
+      "loss": 0.5947,
+      "step": 809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.774848128159663e-05,
+      "loss": 0.5729,
+      "step": 810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.773903484185464e-05,
+      "loss": 0.621,
+      "step": 811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.772956981397339e-05,
+      "loss": 0.6048,
+      "step": 812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.772008620787098e-05,
+      "loss": 0.6151,
+      "step": 813
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.771058403348495e-05,
+      "loss": 0.5792,
+      "step": 814
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.770106330077231e-05,
+      "loss": 0.6091,
+      "step": 815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769152401970953e-05,
+      "loss": 0.5848,
+      "step": 816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7681966200292483e-05,
+      "loss": 0.6007,
+      "step": 817
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7672389852536505e-05,
+      "loss": 0.5848,
+      "step": 818
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.766279498647632e-05,
+      "loss": 0.6219,
+      "step": 819
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.765318161216606e-05,
+      "loss": 0.583,
+      "step": 820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.764354973967928e-05,
+      "loss": 0.6067,
+      "step": 821
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.763389937910887e-05,
+      "loss": 0.5901,
+      "step": 822
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.762423054056713e-05,
+      "loss": 0.6007,
+      "step": 823
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.761454323418572e-05,
+      "loss": 0.5934,
+      "step": 824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.760483747011563e-05,
+      "loss": 0.6097,
+      "step": 825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7595113258527206e-05,
+      "loss": 0.6109,
+      "step": 826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7585370609610126e-05,
+      "loss": 0.5772,
+      "step": 827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.757560953357339e-05,
+      "loss": 0.6084,
+      "step": 828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.756583004064529e-05,
+      "loss": 0.5984,
+      "step": 829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.755603214107344e-05,
+      "loss": 0.5834,
+      "step": 830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7546215845124715e-05,
+      "loss": 0.5901,
+      "step": 831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.753638116308531e-05,
+      "loss": 0.6142,
+      "step": 832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7526528105260626e-05,
+      "loss": 0.6093,
+      "step": 833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7516656681975376e-05,
+      "loss": 0.6051,
+      "step": 834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.750676690357348e-05,
+      "loss": 0.6193,
+      "step": 835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7496858780418104e-05,
+      "loss": 0.5917,
+      "step": 836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7486932322891646e-05,
+      "loss": 0.6129,
+      "step": 837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.74769875413957e-05,
+      "loss": 0.5941,
+      "step": 838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.746702444635108e-05,
+      "loss": 0.5997,
+      "step": 839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.745704304819776e-05,
+      "loss": 0.6302,
+      "step": 840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.744704335739493e-05,
+      "loss": 0.6253,
+      "step": 841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.743702538442093e-05,
+      "loss": 0.6083,
+      "step": 842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.742698913977324e-05,
+      "loss": 0.601,
+      "step": 843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.741693463396852e-05,
+      "loss": 0.5886,
+      "step": 844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.740686187754255e-05,
+      "loss": 0.5809,
+      "step": 845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.739677088105022e-05,
+      "loss": 0.6081,
+      "step": 846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.738666165506556e-05,
+      "loss": 0.5597,
+      "step": 847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.737653421018168e-05,
+      "loss": 0.6348,
+      "step": 848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.736638855701079e-05,
+      "loss": 0.6089,
+      "step": 849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.735622470618419e-05,
+      "loss": 0.5916,
+      "step": 850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.734604266835223e-05,
+      "loss": 0.6012,
+      "step": 851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7335842454184315e-05,
+      "loss": 0.62,
+      "step": 852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.732562407436893e-05,
+      "loss": 0.6025,
+      "step": 853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.731538753961356e-05,
+      "loss": 0.6183,
+      "step": 854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.730513286064471e-05,
+      "loss": 0.62,
+      "step": 855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.729486004820795e-05,
+      "loss": 0.617,
+      "step": 856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7284569113067786e-05,
+      "loss": 0.6153,
+      "step": 857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.727426006600776e-05,
+      "loss": 0.6319,
+      "step": 858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.726393291783036e-05,
+      "loss": 0.6162,
+      "step": 859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.725358767935708e-05,
+      "loss": 0.6186,
+      "step": 860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.724322436142833e-05,
+      "loss": 0.5912,
+      "step": 861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.723284297490349e-05,
+      "loss": 0.5944,
+      "step": 862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.722244353066086e-05,
+      "loss": 0.5795,
+      "step": 863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.721202603959768e-05,
+      "loss": 0.6114,
+      "step": 864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.720159051263008e-05,
+      "loss": 0.6056,
+      "step": 865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.719113696069311e-05,
+      "loss": 0.6029,
+      "step": 866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7180665394740676e-05,
+      "loss": 0.5943,
+      "step": 867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7170175825745603e-05,
+      "loss": 0.6137,
+      "step": 868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7159668264699546e-05,
+      "loss": 0.6182,
+      "step": 869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.714914272261302e-05,
+      "loss": 0.5927,
+      "step": 870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7138599210515403e-05,
+      "loss": 0.5837,
+      "step": 871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.712803773945488e-05,
+      "loss": 0.6372,
+      "step": 872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711745832049846e-05,
+      "loss": 0.6036,
+      "step": 873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7106860964731956e-05,
+      "loss": 0.601,
+      "step": 874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7096245683259996e-05,
+      "loss": 0.6079,
+      "step": 875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.708561248720597e-05,
+      "loss": 0.582,
+      "step": 876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.707496138771204e-05,
+      "loss": 0.5702,
+      "step": 877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7064292395939136e-05,
+      "loss": 0.5962,
+      "step": 878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7053605523066956e-05,
+      "loss": 0.5836,
+      "step": 879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.704290078029389e-05,
+      "loss": 0.6191,
+      "step": 880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.70321781788371e-05,
+      "loss": 0.5644,
+      "step": 881
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702143772993242e-05,
+      "loss": 0.6111,
+      "step": 882
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7010679444834427e-05,
+      "loss": 0.589,
+      "step": 883
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699990333481636e-05,
+      "loss": 0.6234,
+      "step": 884
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6989109411170134e-05,
+      "loss": 0.5881,
+      "step": 885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6978297685206344e-05,
+      "loss": 0.601,
+      "step": 886
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696746816825424e-05,
+      "loss": 0.605,
+      "step": 887
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695662087166169e-05,
+      "loss": 0.6068,
+      "step": 888
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694575580679524e-05,
+      "loss": 0.5906,
+      "step": 889
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.693487298504001e-05,
+      "loss": 0.6184,
+      "step": 890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.692397241779976e-05,
+      "loss": 0.5907,
+      "step": 891
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6913054116496797e-05,
+      "loss": 0.5851,
+      "step": 892
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.690211809257207e-05,
+      "loss": 0.6256,
+      "step": 893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689116435748506e-05,
+      "loss": 0.6107,
+      "step": 894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688019292271383e-05,
+      "loss": 0.5862,
+      "step": 895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686920379975497e-05,
+      "loss": 0.6035,
+      "step": 896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6858197000123596e-05,
+      "loss": 0.6087,
+      "step": 897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68471725353534e-05,
+      "loss": 0.5926,
+      "step": 898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6836130416996515e-05,
+      "loss": 0.6222,
+      "step": 899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.6041,
+      "step": 900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681399326582388e-05,
+      "loss": 0.606,
+      "step": 901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68028982562049e-05,
+      "loss": 0.584,
+      "step": 902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679178563939278e-05,
+      "loss": 0.6131,
+      "step": 903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6780655427032055e-05,
+      "loss": 0.621,
+      "step": 904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67695076307857e-05,
+      "loss": 0.5945,
+      "step": 905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6758342262335114e-05,
+      "loss": 0.5942,
+      "step": 906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6747159333380114e-05,
+      "loss": 0.5838,
+      "step": 907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6735958855638924e-05,
+      "loss": 0.5953,
+      "step": 908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672474084084814e-05,
+      "loss": 0.5944,
+      "step": 909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6713505300762745e-05,
+      "loss": 0.6278,
+      "step": 910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6702252247156074e-05,
+      "loss": 0.5981,
+      "step": 911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6690981691819845e-05,
+      "loss": 0.5876,
+      "step": 912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667969364656408e-05,
+      "loss": 0.602,
+      "step": 913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6668388123217154e-05,
+      "loss": 0.6082,
+      "step": 914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665706513362573e-05,
+      "loss": 0.6166,
+      "step": 915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664572468965481e-05,
+      "loss": 0.5988,
+      "step": 916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.663436680318763e-05,
+      "loss": 0.6137,
+      "step": 917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6622991486125775e-05,
+      "loss": 0.6093,
+      "step": 918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6611598750389044e-05,
+      "loss": 0.6046,
+      "step": 919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6600188607915495e-05,
+      "loss": 0.5808,
+      "step": 920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6588761070661444e-05,
+      "loss": 0.6129,
+      "step": 921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657731615060142e-05,
+      "loss": 0.5929,
+      "step": 922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656585385972816e-05,
+      "loss": 0.6168,
+      "step": 923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655437421005264e-05,
+      "loss": 0.6121,
+      "step": 924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654287721360398e-05,
+      "loss": 0.5916,
+      "step": 925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653136288242949e-05,
+      "loss": 0.6096,
+      "step": 926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6519831228594664e-05,
+      "loss": 0.6079,
+      "step": 927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6508282264183134e-05,
+      "loss": 0.6167,
+      "step": 928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6496716001296675e-05,
+      "loss": 0.6164,
+      "step": 929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648513245205518e-05,
+      "loss": 0.595,
+      "step": 930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6473531628596656e-05,
+      "loss": 0.599,
+      "step": 931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.646191354307723e-05,
+      "loss": 0.5929,
+      "step": 932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.645027820767109e-05,
+      "loss": 0.6023,
+      "step": 933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643862563457053e-05,
+      "loss": 0.5813,
+      "step": 934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642695583598588e-05,
+      "loss": 0.6027,
+      "step": 935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641526882414553e-05,
+      "loss": 0.5903,
+      "step": 936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6403564611295905e-05,
+      "loss": 0.6141,
+      "step": 937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639184320970147e-05,
+      "loss": 0.6325,
+      "step": 938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6380104631644664e-05,
+      "loss": 0.5667,
+      "step": 939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.636834888942598e-05,
+      "loss": 0.6023,
+      "step": 940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6356575995363834e-05,
+      "loss": 0.591,
+      "step": 941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6344785961794674e-05,
+      "loss": 0.5856,
+      "step": 942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6332978801072864e-05,
+      "loss": 0.6256,
+      "step": 943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.632115452557074e-05,
+      "loss": 0.6166,
+      "step": 944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.630931314767855e-05,
+      "loss": 0.5768,
+      "step": 945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.629745467980449e-05,
+      "loss": 0.5893,
+      "step": 946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6285579134374655e-05,
+      "loss": 0.5886,
+      "step": 947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627368652383302e-05,
+      "loss": 0.5797,
+      "step": 948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6261776860641446e-05,
+      "loss": 0.6372,
+      "step": 949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624985015727968e-05,
+      "loss": 0.5988,
+      "step": 950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.623790642624531e-05,
+      "loss": 0.603,
+      "step": 951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.622594568005378e-05,
+      "loss": 0.6034,
+      "step": 952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.621396793123834e-05,
+      "loss": 0.595,
+      "step": 953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.620197319235009e-05,
+      "loss": 0.6042,
+      "step": 954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6189961475957894e-05,
+      "loss": 0.5746,
+      "step": 955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.617793279464843e-05,
+      "loss": 0.6292,
+      "step": 956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.616588716102617e-05,
+      "loss": 0.5794,
+      "step": 957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.61538245877133e-05,
+      "loss": 0.5798,
+      "step": 958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.614174508734981e-05,
+      "loss": 0.5867,
+      "step": 959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.612964867259338e-05,
+      "loss": 0.5985,
+      "step": 960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.611753535611946e-05,
+      "loss": 0.5861,
+      "step": 961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6105405150621176e-05,
+      "loss": 0.6197,
+      "step": 962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6093258068809366e-05,
+      "loss": 0.6152,
+      "step": 963
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6081094123412544e-05,
+      "loss": 0.5956,
+      "step": 964
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.60689133271769e-05,
+      "loss": 0.5685,
+      "step": 965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.605671569286629e-05,
+      "loss": 0.6073,
+      "step": 966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.60445012332622e-05,
+      "loss": 0.5966,
+      "step": 967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.603226996116376e-05,
+      "loss": 0.6132,
+      "step": 968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.602002188938769e-05,
+      "loss": 0.6132,
+      "step": 969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.600775703076836e-05,
+      "loss": 0.6007,
+      "step": 970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5995475398157686e-05,
+      "loss": 0.5937,
+      "step": 971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.598317700442518e-05,
+      "loss": 0.6016,
+      "step": 972
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.597086186245792e-05,
+      "loss": 0.6132,
+      "step": 973
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.595852998516053e-05,
+      "loss": 0.5907,
+      "step": 974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.594618138545517e-05,
+      "loss": 0.5854,
+      "step": 975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.593381607628152e-05,
+      "loss": 0.582,
+      "step": 976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.592143407059677e-05,
+      "loss": 0.594,
+      "step": 977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5909035381375616e-05,
+      "loss": 0.6145,
+      "step": 978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5896620021610236e-05,
+      "loss": 0.5969,
+      "step": 979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5884188004310244e-05,
+      "loss": 0.5925,
+      "step": 980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.587173934250276e-05,
+      "loss": 0.6137,
+      "step": 981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.585927404923229e-05,
+      "loss": 0.5907,
+      "step": 982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.584679213756082e-05,
+      "loss": 0.5941,
+      "step": 983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.583429362056773e-05,
+      "loss": 0.584,
+      "step": 984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5821778511349775e-05,
+      "loss": 0.6156,
+      "step": 985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.580924682302113e-05,
+      "loss": 0.6027,
+      "step": 986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5796698568713334e-05,
+      "loss": 0.6126,
+      "step": 987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.578413376157527e-05,
+      "loss": 0.6126,
+      "step": 988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.577155241477319e-05,
+      "loss": 0.5833,
+      "step": 989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5758954541490655e-05,
+      "loss": 0.5788,
+      "step": 990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.574634015492857e-05,
+      "loss": 0.5924,
+      "step": 991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5733709268305104e-05,
+      "loss": 0.5739,
+      "step": 992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5721061894855756e-05,
+      "loss": 0.6138,
+      "step": 993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.570839804783328e-05,
+      "loss": 0.6049,
+      "step": 994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.569571774050768e-05,
+      "loss": 0.6132,
+      "step": 995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.568302098616625e-05,
+      "loss": 0.5844,
+      "step": 996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.567030779811349e-05,
+      "loss": 0.6083,
+      "step": 997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.565757818967111e-05,
+      "loss": 0.5715,
+      "step": 998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.564483217417805e-05,
+      "loss": 0.5875,
+      "step": 999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.563206976499042e-05,
+      "loss": 0.5852,
+      "step": 1000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.561929097548153e-05,
+      "loss": 0.6181,
+      "step": 1001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.560649581904184e-05,
+      "loss": 0.5821,
+      "step": 1002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.559368430907896e-05,
+      "loss": 0.6116,
+      "step": 1003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.558085645901766e-05,
+      "loss": 0.6114,
+      "step": 1004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.55680122822998e-05,
+      "loss": 0.5844,
+      "step": 1005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5555151792384354e-05,
+      "loss": 0.5984,
+      "step": 1006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.554227500274742e-05,
+      "loss": 0.593,
+      "step": 1007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5529381926882134e-05,
+      "loss": 0.6214,
+      "step": 1008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.551647257829872e-05,
+      "loss": 0.5883,
+      "step": 1009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5503546970524465e-05,
+      "loss": 0.5981,
+      "step": 1010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.549060511710367e-05,
+      "loss": 0.5651,
+      "step": 1011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.547764703159766e-05,
+      "loss": 0.6165,
+      "step": 1012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.546467272758479e-05,
+      "loss": 0.5962,
+      "step": 1013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.545168221866039e-05,
+      "loss": 0.5927,
+      "step": 1014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.543867551843679e-05,
+      "loss": 0.5735,
+      "step": 1015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5425652640543266e-05,
+      "loss": 0.6112,
+      "step": 1016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.541261359862605e-05,
+      "loss": 0.6162,
+      "step": 1017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.539955840634833e-05,
+      "loss": 0.5896,
+      "step": 1018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.53864870773902e-05,
+      "loss": 0.592,
+      "step": 1019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5373399625448654e-05,
+      "loss": 0.5994,
+      "step": 1020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5360296064237606e-05,
+      "loss": 0.5722,
+      "step": 1021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.534717640748783e-05,
+      "loss": 0.6166,
+      "step": 1022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.533404066894699e-05,
+      "loss": 0.6139,
+      "step": 1023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.5872,
+      "step": 1024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.53077210015669e-05,
+      "loss": 0.5943,
+      "step": 1025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5294537100307166e-05,
+      "loss": 0.5854,
+      "step": 1026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.528133717241531e-05,
+      "loss": 0.5971,
+      "step": 1027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.526812123172311e-05,
+      "loss": 0.6045,
+      "step": 1028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5254889292079095e-05,
+      "loss": 0.6131,
+      "step": 1029
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.524164136734858e-05,
+      "loss": 0.5762,
+      "step": 1030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.522837747141361e-05,
+      "loss": 0.5667,
+      "step": 1031
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5215097618172986e-05,
+      "loss": 0.58,
+      "step": 1032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.520180182154222e-05,
+      "loss": 0.5909,
+      "step": 1033
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.518849009545353e-05,
+      "loss": 0.593,
+      "step": 1034
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.517516245385582e-05,
+      "loss": 0.6002,
+      "step": 1035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.516181891071469e-05,
+      "loss": 0.6052,
+      "step": 1036
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.514845948001238e-05,
+      "loss": 0.5962,
+      "step": 1037
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5135084175747795e-05,
+      "loss": 0.6232,
+      "step": 1038
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.512169301193646e-05,
+      "loss": 0.5927,
+      "step": 1039
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.510828600261054e-05,
+      "loss": 0.5954,
+      "step": 1040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.509486316181878e-05,
+      "loss": 0.5889,
+      "step": 1041
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.508142450362653e-05,
+      "loss": 0.5916,
+      "step": 1042
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5067970042115706e-05,
+      "loss": 0.6295,
+      "step": 1043
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.505449979138478e-05,
+      "loss": 0.6132,
+      "step": 1044
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5041013765548785e-05,
+      "loss": 0.5954,
+      "step": 1045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.502751197873927e-05,
+      "loss": 0.5898,
+      "step": 1046
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.501399444510432e-05,
+      "loss": 0.5869,
+      "step": 1047
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.500046117880847e-05,
+      "loss": 0.6064,
+      "step": 1048
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.498691219403281e-05,
+      "loss": 0.5921,
+      "step": 1049
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.497334750497485e-05,
+      "loss": 0.6219,
+      "step": 1050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.495976712584857e-05,
+      "loss": 0.5954,
+      "step": 1051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.494617107088441e-05,
+      "loss": 0.5985,
+      "step": 1052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4932559354329196e-05,
+      "loss": 0.5897,
+      "step": 1053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.491893199044621e-05,
+      "loss": 0.6005,
+      "step": 1054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.49052889935151e-05,
+      "loss": 0.5854,
+      "step": 1055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.48916303778319e-05,
+      "loss": 0.6074,
+      "step": 1056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4877956157709024e-05,
+      "loss": 0.5843,
+      "step": 1057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.486426634747522e-05,
+      "loss": 0.6156,
+      "step": 1058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4850560961475594e-05,
+      "loss": 0.6152,
+      "step": 1059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.483684001407155e-05,
+      "loss": 0.5903,
+      "step": 1060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4823103519640805e-05,
+      "loss": 0.6159,
+      "step": 1061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.480935149257737e-05,
+      "loss": 0.5713,
+      "step": 1062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4795583947291544e-05,
+      "loss": 0.6259,
+      "step": 1063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.478180089820987e-05,
+      "loss": 0.5797,
+      "step": 1064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.476800235977514e-05,
+      "loss": 0.594,
+      "step": 1065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.475418834644638e-05,
+      "loss": 0.5721,
+      "step": 1066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.474035887269884e-05,
+      "loss": 0.598,
+      "step": 1067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4726513953023944e-05,
+      "loss": 0.5753,
+      "step": 1068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4712653601929336e-05,
+      "loss": 0.5977,
+      "step": 1069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4698777833938805e-05,
+      "loss": 0.5935,
+      "step": 1070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4684886663592295e-05,
+      "loss": 0.5959,
+      "step": 1071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.467098010544591e-05,
+      "loss": 0.5932,
+      "step": 1072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.465705817407185e-05,
+      "loss": 0.58,
+      "step": 1073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.464312088405847e-05,
+      "loss": 0.5822,
+      "step": 1074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.462916825001015e-05,
+      "loss": 0.5788,
+      "step": 1075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.461520028654742e-05,
+      "loss": 0.6095,
+      "step": 1076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4601217008306815e-05,
+      "loss": 0.6079,
+      "step": 1077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.458721842994096e-05,
+      "loss": 0.5724,
+      "step": 1078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4573204566118476e-05,
+      "loss": 0.5952,
+      "step": 1079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.455917543152404e-05,
+      "loss": 0.5824,
+      "step": 1080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4545131040858315e-05,
+      "loss": 0.5648,
+      "step": 1081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.453107140883792e-05,
+      "loss": 0.5867,
+      "step": 1082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.451699655019549e-05,
+      "loss": 0.5857,
+      "step": 1083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.450290647967961e-05,
+      "loss": 0.5892,
+      "step": 1084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.448880121205476e-05,
+      "loss": 0.5903,
+      "step": 1085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.44746807621014e-05,
+      "loss": 0.5795,
+      "step": 1086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4460545144615876e-05,
+      "loss": 0.5816,
+      "step": 1087
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4446394374410416e-05,
+      "loss": 0.614,
+      "step": 1088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4432228466313134e-05,
+      "loss": 0.5936,
+      "step": 1089
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4418047435168025e-05,
+      "loss": 0.5889,
+      "step": 1090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.44038512958349e-05,
+      "loss": 0.5671,
+      "step": 1091
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.438964006318943e-05,
+      "loss": 0.6065,
+      "step": 1092
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.437541375212308e-05,
+      "loss": 0.5927,
+      "step": 1093
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4361172377543126e-05,
+      "loss": 0.6039,
+      "step": 1094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.434691595437263e-05,
+      "loss": 0.5755,
+      "step": 1095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4332644497550424e-05,
+      "loss": 0.5872,
+      "step": 1096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4318358022031075e-05,
+      "loss": 0.5973,
+      "step": 1097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.430405654278491e-05,
+      "loss": 0.613,
+      "step": 1098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.428974007479796e-05,
+      "loss": 0.5627,
+      "step": 1099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.427540863307198e-05,
+      "loss": 0.5733,
+      "step": 1100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4261062232624405e-05,
+      "loss": 0.6207,
+      "step": 1101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.424670088848835e-05,
+      "loss": 0.5794,
+      "step": 1102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4232324615712586e-05,
+      "loss": 0.5907,
+      "step": 1103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4217933429361516e-05,
+      "loss": 0.5855,
+      "step": 1104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.420352734451519e-05,
+      "loss": 0.5937,
+      "step": 1105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.418910637626926e-05,
+      "loss": 0.5975,
+      "step": 1106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.417467053973497e-05,
+      "loss": 0.5848,
+      "step": 1107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.416021985003915e-05,
+      "loss": 0.5862,
+      "step": 1108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.41457543223242e-05,
+      "loss": 0.5984,
+      "step": 1109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4131273971748055e-05,
+      "loss": 0.6108,
+      "step": 1110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4116778813484186e-05,
+      "loss": 0.6085,
+      "step": 1111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.410226886272159e-05,
+      "loss": 0.5675,
+      "step": 1112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.408774413466475e-05,
+      "loss": 0.6044,
+      "step": 1113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.407320464453365e-05,
+      "loss": 0.5947,
+      "step": 1114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.405865040756372e-05,
+      "loss": 0.5891,
+      "step": 1115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.404408143900587e-05,
+      "loss": 0.5866,
+      "step": 1116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4029497754126425e-05,
+      "loss": 0.5921,
+      "step": 1117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.401489936820715e-05,
+      "loss": 0.5909,
+      "step": 1118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.400028629654519e-05,
+      "loss": 0.5998,
+      "step": 1119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.398565855445309e-05,
+      "loss": 0.5985,
+      "step": 1120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.397101615725879e-05,
+      "loss": 0.5964,
+      "step": 1121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3956359120305547e-05,
+      "loss": 0.6166,
+      "step": 1122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.394168745895199e-05,
+      "loss": 0.5974,
+      "step": 1123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.392700118857205e-05,
+      "loss": 0.588,
+      "step": 1124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.391230032455498e-05,
+      "loss": 0.5753,
+      "step": 1125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.389758488230533e-05,
+      "loss": 0.5797,
+      "step": 1126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.388285487724292e-05,
+      "loss": 0.5991,
+      "step": 1127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.386811032480281e-05,
+      "loss": 0.5942,
+      "step": 1128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.385335124043533e-05,
+      "loss": 0.5745,
+      "step": 1129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3838577639606046e-05,
+      "loss": 0.5856,
+      "step": 1130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.38237895377957e-05,
+      "loss": 0.5865,
+      "step": 1131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.380898695050024e-05,
+      "loss": 0.5768,
+      "step": 1132
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3794169893230827e-05,
+      "loss": 0.5912,
+      "step": 1133
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.377933838151374e-05,
+      "loss": 0.5717,
+      "step": 1134
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3764492430890414e-05,
+      "loss": 0.6019,
+      "step": 1135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.374963205691744e-05,
+      "loss": 0.5935,
+      "step": 1136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.373475727516651e-05,
+      "loss": 0.586,
+      "step": 1137
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.371986810122438e-05,
+      "loss": 0.5951,
+      "step": 1138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3704964550692946e-05,
+      "loss": 0.547,
+      "step": 1139
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.369004663918912e-05,
+      "loss": 0.5463,
+      "step": 1140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.367511438234488e-05,
+      "loss": 0.5389,
+      "step": 1141
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.366016779580726e-05,
+      "loss": 0.5139,
+      "step": 1142
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3645206895238254e-05,
+      "loss": 0.524,
+      "step": 1143
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3630231696314913e-05,
+      "loss": 0.5418,
+      "step": 1144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3615242214729226e-05,
+      "loss": 0.5224,
+      "step": 1145
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3600238466188185e-05,
+      "loss": 0.5412,
+      "step": 1146
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.358522046641371e-05,
+      "loss": 0.533,
+      "step": 1147
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3570188231142647e-05,
+      "loss": 0.5421,
+      "step": 1148
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.355514177612678e-05,
+      "loss": 0.5459,
+      "step": 1149
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.354008111713279e-05,
+      "loss": 0.5251,
+      "step": 1150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.352500626994222e-05,
+      "loss": 0.5324,
+      "step": 1151
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.350991725035152e-05,
+      "loss": 0.5474,
+      "step": 1152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3494814074171946e-05,
+      "loss": 0.5449,
+      "step": 1153
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.347969675722961e-05,
+      "loss": 0.5299,
+      "step": 1154
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3464565315365466e-05,
+      "loss": 0.5459,
+      "step": 1155
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.344941976443521e-05,
+      "loss": 0.5397,
+      "step": 1156
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.343426012030938e-05,
+      "loss": 0.5089,
+      "step": 1157
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.341908639887325e-05,
+      "loss": 0.5353,
+      "step": 1158
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3403898616026844e-05,
+      "loss": 0.5194,
+      "step": 1159
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.338869678768495e-05,
+      "loss": 0.5331,
+      "step": 1160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.337348092977704e-05,
+      "loss": 0.5238,
+      "step": 1161
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.335825105824731e-05,
+      "loss": 0.5263,
+      "step": 1162
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3343007189054625e-05,
+      "loss": 0.5281,
+      "step": 1163
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.332774933817253e-05,
+      "loss": 0.5201,
+      "step": 1164
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3312477521589194e-05,
+      "loss": 0.5355,
+      "step": 1165
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.329719175530747e-05,
+      "loss": 0.5321,
+      "step": 1166
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.328189205534479e-05,
+      "loss": 0.5664,
+      "step": 1167
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3266578437733174e-05,
+      "loss": 0.525,
+      "step": 1168
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.325125091851928e-05,
+      "loss": 0.5193,
+      "step": 1169
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3235909513764286e-05,
+      "loss": 0.5375,
+      "step": 1170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.322055423954393e-05,
+      "loss": 0.552,
+      "step": 1171
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.32051851119485e-05,
+      "loss": 0.5239,
+      "step": 1172
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.318980214708279e-05,
+      "loss": 0.5291,
+      "step": 1173
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3174405361066075e-05,
+      "loss": 0.5292,
+      "step": 1174
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3158994770032146e-05,
+      "loss": 0.545,
+      "step": 1175
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3143570390129244e-05,
+      "loss": 0.525,
+      "step": 1176
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.312813223752005e-05,
+      "loss": 0.511,
+      "step": 1177
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.311268032838169e-05,
+      "loss": 0.5458,
+      "step": 1178
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5245,
+      "step": 1179
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3081735305298024e-05,
+      "loss": 0.5431,
+      "step": 1180
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.306624222377897e-05,
+      "loss": 0.545,
+      "step": 1181
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3050735450583216e-05,
+      "loss": 0.543,
+      "step": 1182
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3035215001959797e-05,
+      "loss": 0.5291,
+      "step": 1183
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.301968089417206e-05,
+      "loss": 0.5389,
+      "step": 1184
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.300413314349768e-05,
+      "loss": 0.5107,
+      "step": 1185
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.298857176622863e-05,
+      "loss": 0.548,
+      "step": 1186
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.297299677867115e-05,
+      "loss": 0.5305,
+      "step": 1187
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.295740819714574e-05,
+      "loss": 0.5392,
+      "step": 1188
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.294180603798716e-05,
+      "loss": 0.5504,
+      "step": 1189
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.292619031754438e-05,
+      "loss": 0.528,
+      "step": 1190
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.29105610521806e-05,
+      "loss": 0.5508,
+      "step": 1191
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.28949182582732e-05,
+      "loss": 0.538,
+      "step": 1192
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.287926195221373e-05,
+      "loss": 0.5411,
+      "step": 1193
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2863592150407925e-05,
+      "loss": 0.5366,
+      "step": 1194
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.284790886927562e-05,
+      "loss": 0.5404,
+      "step": 1195
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.283221212525082e-05,
+      "loss": 0.5308,
+      "step": 1196
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.281650193478161e-05,
+      "loss": 0.5247,
+      "step": 1197
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.280077831433017e-05,
+      "loss": 0.5317,
+      "step": 1198
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.278504128037275e-05,
+      "loss": 0.5596,
+      "step": 1199
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.276929084939967e-05,
+      "loss": 0.5336,
+      "step": 1200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.275352703791527e-05,
+      "loss": 0.5126,
+      "step": 1201
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.273774986243792e-05,
+      "loss": 0.5534,
+      "step": 1202
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.27219593395e-05,
+      "loss": 0.5487,
+      "step": 1203
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.270615548564786e-05,
+      "loss": 0.5321,
+      "step": 1204
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2690338317441824e-05,
+      "loss": 0.5418,
+      "step": 1205
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.267450785145618e-05,
+      "loss": 0.5404,
+      "step": 1206
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.265866410427914e-05,
+      "loss": 0.5268,
+      "step": 1207
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.264280709251285e-05,
+      "loss": 0.5517,
+      "step": 1208
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2626936832773306e-05,
+      "loss": 0.5467,
+      "step": 1209
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.261105334169045e-05,
+      "loss": 0.5134,
+      "step": 1210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.259515663590805e-05,
+      "loss": 0.5324,
+      "step": 1211
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2579246732083726e-05,
+      "loss": 0.5268,
+      "step": 1212
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2563323646888945e-05,
+      "loss": 0.526,
+      "step": 1213
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.254738739700896e-05,
+      "loss": 0.5451,
+      "step": 1214
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.253143799914283e-05,
+      "loss": 0.5458,
+      "step": 1215
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.251547547000342e-05,
+      "loss": 0.5073,
+      "step": 1216
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.249949982631729e-05,
+      "loss": 0.5272,
+      "step": 1217
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2483511084824805e-05,
+      "loss": 0.5425,
+      "step": 1218
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.246750926228004e-05,
+      "loss": 0.5257,
+      "step": 1219
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.245149437545074e-05,
+      "loss": 0.5494,
+      "step": 1220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.243546644111838e-05,
+      "loss": 0.5253,
+      "step": 1221
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.24194254760781e-05,
+      "loss": 0.5553,
+      "step": 1222
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.240337149713868e-05,
+      "loss": 0.5332,
+      "step": 1223
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.238730452112256e-05,
+      "loss": 0.5379,
+      "step": 1224
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.237122456486577e-05,
+      "loss": 0.5373,
+      "step": 1225
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.235513164521796e-05,
+      "loss": 0.5233,
+      "step": 1226
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.233902577904236e-05,
+      "loss": 0.5268,
+      "step": 1227
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.232290698321578e-05,
+      "loss": 0.539,
+      "step": 1228
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2306775274628544e-05,
+      "loss": 0.5365,
+      "step": 1229
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2290630670184544e-05,
+      "loss": 0.5369,
+      "step": 1230
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.227447318680115e-05,
+      "loss": 0.5399,
+      "step": 1231
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.225830284140927e-05,
+      "loss": 0.5192,
+      "step": 1232
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.224211965095326e-05,
+      "loss": 0.5107,
+      "step": 1233
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.222592363239093e-05,
+      "loss": 0.5598,
+      "step": 1234
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.220971480269356e-05,
+      "loss": 0.555,
+      "step": 1235
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.219349317884582e-05,
+      "loss": 0.5396,
+      "step": 1236
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2177258777845825e-05,
+      "loss": 0.5489,
+      "step": 1237
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.216101161670505e-05,
+      "loss": 0.5558,
+      "step": 1238
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.214475171244835e-05,
+      "loss": 0.5441,
+      "step": 1239
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2128479082113934e-05,
+      "loss": 0.5393,
+      "step": 1240
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2112193742753333e-05,
+      "loss": 0.5289,
+      "step": 1241
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.209589571143143e-05,
+      "loss": 0.5184,
+      "step": 1242
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.207958500522637e-05,
+      "loss": 0.5376,
+      "step": 1243
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.20632616412296e-05,
+      "loss": 0.5327,
+      "step": 1244
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.204692563654582e-05,
+      "loss": 0.5473,
+      "step": 1245
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.203057700829299e-05,
+      "loss": 0.5411,
+      "step": 1246
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.2014215773602276e-05,
+      "loss": 0.5546,
+      "step": 1247
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.199784194961808e-05,
+      "loss": 0.5466,
+      "step": 1248
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.198145555349798e-05,
+      "loss": 0.5422,
+      "step": 1249
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.196505660241272e-05,
+      "loss": 0.5339,
+      "step": 1250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.194864511354622e-05,
+      "loss": 0.5349,
+      "step": 1251
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.193222110409553e-05,
+      "loss": 0.556,
+      "step": 1252
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1915784591270807e-05,
+      "loss": 0.5275,
+      "step": 1253
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.189933559229533e-05,
+      "loss": 0.5449,
+      "step": 1254
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.188287412440546e-05,
+      "loss": 0.5474,
+      "step": 1255
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.18664002048506e-05,
+      "loss": 0.5205,
+      "step": 1256
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1849913850893226e-05,
+      "loss": 0.5325,
+      "step": 1257
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.183341507980884e-05,
+      "loss": 0.533,
+      "step": 1258
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.181690390888594e-05,
+      "loss": 0.5355,
+      "step": 1259
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1800380355426036e-05,
+      "loss": 0.5678,
+      "step": 1260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1783844436743604e-05,
+      "loss": 0.534,
+      "step": 1261
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.176729617016607e-05,
+      "loss": 0.5461,
+      "step": 1262
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.175073557303382e-05,
+      "loss": 0.5441,
+      "step": 1263
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.173416266270014e-05,
+      "loss": 0.5378,
+      "step": 1264
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.171757745653122e-05,
+      "loss": 0.5402,
+      "step": 1265
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.170097997190615e-05,
+      "loss": 0.5509,
+      "step": 1266
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.168437022621689e-05,
+      "loss": 0.5515,
+      "step": 1267
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.166774823686821e-05,
+      "loss": 0.5462,
+      "step": 1268
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.165111402127774e-05,
+      "loss": 0.5328,
+      "step": 1269
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1634467596875935e-05,
+      "loss": 0.5461,
+      "step": 1270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1617808981106014e-05,
+      "loss": 0.5259,
+      "step": 1271
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1601138191423966e-05,
+      "loss": 0.543,
+      "step": 1272
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.158445524529857e-05,
+      "loss": 0.543,
+      "step": 1273
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.156776016021132e-05,
+      "loss": 0.5241,
+      "step": 1274
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.155105295365644e-05,
+      "loss": 0.5473,
+      "step": 1275
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.153433364314084e-05,
+      "loss": 0.5435,
+      "step": 1276
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.151760224618413e-05,
+      "loss": 0.5385,
+      "step": 1277
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1500858780318586e-05,
+      "loss": 0.5233,
+      "step": 1278
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.148410326308912e-05,
+      "loss": 0.5449,
+      "step": 1279
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.146733571205327e-05,
+      "loss": 0.5392,
+      "step": 1280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.145055614478119e-05,
+      "loss": 0.5496,
+      "step": 1281
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1433764578855633e-05,
+      "loss": 0.55,
+      "step": 1282
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.141696103187192e-05,
+      "loss": 0.5438,
+      "step": 1283
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1400145521437915e-05,
+      "loss": 0.5459,
+      "step": 1284
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1383318065174026e-05,
+      "loss": 0.5327,
+      "step": 1285
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.136647868071319e-05,
+      "loss": 0.5742,
+      "step": 1286
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.134962738570082e-05,
+      "loss": 0.5583,
+      "step": 1287
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1332764197794825e-05,
+      "loss": 0.5261,
+      "step": 1288
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.131588913466559e-05,
+      "loss": 0.5369,
+      "step": 1289
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1299002213995907e-05,
+      "loss": 0.5461,
+      "step": 1290
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1282103453481014e-05,
+      "loss": 0.5336,
+      "step": 1291
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.126519287082858e-05,
+      "loss": 0.54,
+      "step": 1292
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.124827048375861e-05,
+      "loss": 0.5403,
+      "step": 1293
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1231336310003514e-05,
+      "loss": 0.5539,
+      "step": 1294
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.121439036730805e-05,
+      "loss": 0.5471,
+      "step": 1295
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.119743267342929e-05,
+      "loss": 0.5243,
+      "step": 1296
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1180463246136644e-05,
+      "loss": 0.526,
+      "step": 1297
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.11634821032118e-05,
+      "loss": 0.5613,
+      "step": 1298
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.114648926244873e-05,
+      "loss": 0.529,
+      "step": 1299
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.112948474165366e-05,
+      "loss": 0.5451,
+      "step": 1300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.111246855864506e-05,
+      "loss": 0.5364,
+      "step": 1301
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1095440731253616e-05,
+      "loss": 0.5449,
+      "step": 1302
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.107840127732221e-05,
+      "loss": 0.5263,
+      "step": 1303
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.106135021470592e-05,
+      "loss": 0.5319,
+      "step": 1304
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.104428756127198e-05,
+      "loss": 0.5071,
+      "step": 1305
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1027213334899774e-05,
+      "loss": 0.5566,
+      "step": 1306
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.101012755348081e-05,
+      "loss": 0.519,
+      "step": 1307
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.099303023491871e-05,
+      "loss": 0.5232,
+      "step": 1308
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.097592139712917e-05,
+      "loss": 0.5419,
+      "step": 1309
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.095880105803997e-05,
+      "loss": 0.5645,
+      "step": 1310
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.094166923559095e-05,
+      "loss": 0.5533,
+      "step": 1311
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.092452594773395e-05,
+      "loss": 0.5556,
+      "step": 1312
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0907371212432864e-05,
+      "loss": 0.5208,
+      "step": 1313
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.089020504766356e-05,
+      "loss": 0.5437,
+      "step": 1314
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0873027471413884e-05,
+      "loss": 0.5404,
+      "step": 1315
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.085583850168364e-05,
+      "loss": 0.5491,
+      "step": 1316
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.083863815648458e-05,
+      "loss": 0.5319,
+      "step": 1317
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.082142645384035e-05,
+      "loss": 0.5357,
+      "step": 1318
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.080420341178654e-05,
+      "loss": 0.5346,
+      "step": 1319
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.078696904837057e-05,
+      "loss": 0.5477,
+      "step": 1320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.076972338165178e-05,
+      "loss": 0.582,
+      "step": 1321
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.075246642970131e-05,
+      "loss": 0.54,
+      "step": 1322
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0735198210602136e-05,
+      "loss": 0.5461,
+      "step": 1323
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.071791874244906e-05,
+      "loss": 0.5535,
+      "step": 1324
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.070062804334865e-05,
+      "loss": 0.5454,
+      "step": 1325
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0683326131419245e-05,
+      "loss": 0.5311,
+      "step": 1326
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0666013024790936e-05,
+      "loss": 0.5342,
+      "step": 1327
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0648688741605566e-05,
+      "loss": 0.536,
+      "step": 1328
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.063135330001664e-05,
+      "loss": 0.5394,
+      "step": 1329
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.06140067181894e-05,
+      "loss": 0.5222,
+      "step": 1330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.059664901430076e-05,
+      "loss": 0.5327,
+      "step": 1331
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.057928020653925e-05,
+      "loss": 0.5248,
+      "step": 1332
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.056190031310508e-05,
+      "loss": 0.5488,
+      "step": 1333
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.054450935221005e-05,
+      "loss": 0.5604,
+      "step": 1334
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.052710734207756e-05,
+      "loss": 0.5283,
+      "step": 1335
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.050969430094261e-05,
+      "loss": 0.5503,
+      "step": 1336
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0492270247051715e-05,
+      "loss": 0.5515,
+      "step": 1337
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.047483519866298e-05,
+      "loss": 0.5347,
+      "step": 1338
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0457389174045993e-05,
+      "loss": 0.5755,
+      "step": 1339
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0439932191481854e-05,
+      "loss": 0.5435,
+      "step": 1340
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0422464269263172e-05,
+      "loss": 0.5431,
+      "step": 1341
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0404985425693972e-05,
+      "loss": 0.5307,
+      "step": 1342
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0387495679089753e-05,
+      "loss": 0.5475,
+      "step": 1343
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.036999504777745e-05,
+      "loss": 0.5281,
+      "step": 1344
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.035248355009536e-05,
+      "loss": 0.5199,
+      "step": 1345
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0334961204393214e-05,
+      "loss": 0.5207,
+      "step": 1346
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0317428029032087e-05,
+      "loss": 0.5113,
+      "step": 1347
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0299884042384398e-05,
+      "loss": 0.5393,
+      "step": 1348
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0282329262833906e-05,
+      "loss": 0.5495,
+      "step": 1349
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.026476370877568e-05,
+      "loss": 0.5552,
+      "step": 1350
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.024718739861607e-05,
+      "loss": 0.5414,
+      "step": 1351
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.022960035077269e-05,
+      "loss": 0.5538,
+      "step": 1352
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0212002583674446e-05,
+      "loss": 0.5491,
+      "step": 1353
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0194394115761415e-05,
+      "loss": 0.5326,
+      "step": 1354
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.017677496548494e-05,
+      "loss": 0.5443,
+      "step": 1355
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.015914515130754e-05,
+      "loss": 0.5429,
+      "step": 1356
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.014150469170288e-05,
+      "loss": 0.5507,
+      "step": 1357
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0123853605155834e-05,
+      "loss": 0.5232,
+      "step": 1358
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.010619191016237e-05,
+      "loss": 0.5419,
+      "step": 1359
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0088519625229588e-05,
+      "loss": 0.5399,
+      "step": 1360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.007083676887568e-05,
+      "loss": 0.532,
+      "step": 1361
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.005314335962992e-05,
+      "loss": 0.5323,
+      "step": 1362
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0035439416032647e-05,
+      "loss": 0.5302,
+      "step": 1363
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.00177249566352e-05,
+      "loss": 0.5437,
+      "step": 1364
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5519,
+      "step": 1365
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9982264564700424e-05,
+      "loss": 0.5452,
+      "step": 1366
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9964518669320824e-05,
+      "loss": 0.548,
+      "step": 1367
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9946762332456548e-05,
+      "loss": 0.5109,
+      "step": 1368
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9928995572713854e-05,
+      "loss": 0.5285,
+      "step": 1369
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9911218408709933e-05,
+      "loss": 0.5508,
+      "step": 1370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9893430859072887e-05,
+      "loss": 0.5366,
+      "step": 1371
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9875632942441685e-05,
+      "loss": 0.5266,
+      "step": 1372
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9857824677466173e-05,
+      "loss": 0.551,
+      "step": 1373
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9840006082807024e-05,
+      "loss": 0.5316,
+      "step": 1374
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9822177177135752e-05,
+      "loss": 0.5346,
+      "step": 1375
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.980433797913467e-05,
+      "loss": 0.529,
+      "step": 1376
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9786488507496877e-05,
+      "loss": 0.5508,
+      "step": 1377
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9768628780926225e-05,
+      "loss": 0.5367,
+      "step": 1378
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9750758818137328e-05,
+      "loss": 0.5364,
+      "step": 1379
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9732878637855526e-05,
+      "loss": 0.5546,
+      "step": 1380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.971498825881685e-05,
+      "loss": 0.5346,
+      "step": 1381
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.969708769976803e-05,
+      "loss": 0.5407,
+      "step": 1382
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9679176979466477e-05,
+      "loss": 0.5441,
+      "step": 1383
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9661256116680204e-05,
+      "loss": 0.5603,
+      "step": 1384
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9643325130187907e-05,
+      "loss": 0.5472,
+      "step": 1385
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.962538403877885e-05,
+      "loss": 0.5492,
+      "step": 1386
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.960743286125291e-05,
+      "loss": 0.5346,
+      "step": 1387
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.958947161642052e-05,
+      "loss": 0.5367,
+      "step": 1388
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9571500323102662e-05,
+      "loss": 0.5656,
+      "step": 1389
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9553519000130856e-05,
+      "loss": 0.5365,
+      "step": 1390
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.953552766634712e-05,
+      "loss": 0.5339,
+      "step": 1391
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9517526340603987e-05,
+      "loss": 0.5344,
+      "step": 1392
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.949951504176442e-05,
+      "loss": 0.5403,
+      "step": 1393
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.948149378870186e-05,
+      "loss": 0.5445,
+      "step": 1394
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9463462600300184e-05,
+      "loss": 0.552,
+      "step": 1395
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.944542149545366e-05,
+      "loss": 0.5274,
+      "step": 1396
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9427370493066956e-05,
+      "loss": 0.5469,
+      "step": 1397
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9409309612055116e-05,
+      "loss": 0.5433,
+      "step": 1398
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.939123887134353e-05,
+      "loss": 0.5394,
+      "step": 1399
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.937315828986792e-05,
+      "loss": 0.5265,
+      "step": 1400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9355067886574315e-05,
+      "loss": 0.5397,
+      "step": 1401
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9336967680419053e-05,
+      "loss": 0.549,
+      "step": 1402
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9318857690368715e-05,
+      "loss": 0.5419,
+      "step": 1403
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.930073793540017e-05,
+      "loss": 0.5416,
+      "step": 1404
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9282608434500504e-05,
+      "loss": 0.5235,
+      "step": 1405
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.926446920666699e-05,
+      "loss": 0.5391,
+      "step": 1406
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9246320270907133e-05,
+      "loss": 0.5663,
+      "step": 1407
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.922816164623859e-05,
+      "loss": 0.5513,
+      "step": 1408
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.920999335168917e-05,
+      "loss": 0.536,
+      "step": 1409
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9191815406296827e-05,
+      "loss": 0.5304,
+      "step": 1410
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9173627829109613e-05,
+      "loss": 0.5333,
+      "step": 1411
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9155430639185686e-05,
+      "loss": 0.531,
+      "step": 1412
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.913722385559326e-05,
+      "loss": 0.5481,
+      "step": 1413
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9119007497410624e-05,
+      "loss": 0.5278,
+      "step": 1414
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.910078158372609e-05,
+      "loss": 0.5455,
+      "step": 1415
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9082546133637964e-05,
+      "loss": 0.5525,
+      "step": 1416
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.906430116625458e-05,
+      "loss": 0.5196,
+      "step": 1417
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9046046700694214e-05,
+      "loss": 0.5206,
+      "step": 1418
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.902778275608512e-05,
+      "loss": 0.5343,
+      "step": 1419
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9009509351565462e-05,
+      "loss": 0.5443,
+      "step": 1420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8991226506283335e-05,
+      "loss": 0.5355,
+      "step": 1421
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8972934239396717e-05,
+      "loss": 0.543,
+      "step": 1422
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8954632570073458e-05,
+      "loss": 0.5274,
+      "step": 1423
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8936321517491262e-05,
+      "loss": 0.54,
+      "step": 1424
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8918001100837673e-05,
+      "loss": 0.5282,
+      "step": 1425
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.889967133931003e-05,
+      "loss": 0.5425,
+      "step": 1426
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8881332252115486e-05,
+      "loss": 0.5397,
+      "step": 1427
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.886298385847095e-05,
+      "loss": 0.563,
+      "step": 1428
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.884462617760309e-05,
+      "loss": 0.5387,
+      "step": 1429
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8826259228748298e-05,
+      "loss": 0.5664,
+      "step": 1430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.880788303115269e-05,
+      "loss": 0.5281,
+      "step": 1431
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8789497604072064e-05,
+      "loss": 0.5565,
+      "step": 1432
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8771102966771888e-05,
+      "loss": 0.5465,
+      "step": 1433
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8752699138527296e-05,
+      "loss": 0.5209,
+      "step": 1434
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8734286138623032e-05,
+      "loss": 0.5552,
+      "step": 1435
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.871586398635346e-05,
+      "loss": 0.5326,
+      "step": 1436
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8697432701022544e-05,
+      "loss": 0.5442,
+      "step": 1437
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8678992301943802e-05,
+      "loss": 0.5442,
+      "step": 1438
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.866054280844031e-05,
+      "loss": 0.5624,
+      "step": 1439
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8642084239844675e-05,
+      "loss": 0.5259,
+      "step": 1440
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8623616615499014e-05,
+      "loss": 0.5304,
+      "step": 1441
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8605139954754923e-05,
+      "loss": 0.5479,
+      "step": 1442
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.858665427697348e-05,
+      "loss": 0.5128,
+      "step": 1443
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8568159601525214e-05,
+      "loss": 0.5293,
+      "step": 1444
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8549655947790053e-05,
+      "loss": 0.5408,
+      "step": 1445
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8531143335157378e-05,
+      "loss": 0.5555,
+      "step": 1446
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8512621783025916e-05,
+      "loss": 0.5415,
+      "step": 1447
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8494091310803792e-05,
+      "loss": 0.5299,
+      "step": 1448
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8475551937908464e-05,
+      "loss": 0.5126,
+      "step": 1449
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.845700368376672e-05,
+      "loss": 0.5498,
+      "step": 1450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.843844656781465e-05,
+      "loss": 0.5449,
+      "step": 1451
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.841988060949762e-05,
+      "loss": 0.5143,
+      "step": 1452
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8401305828270302e-05,
+      "loss": 0.5497,
+      "step": 1453
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8382722243596567e-05,
+      "loss": 0.5055,
+      "step": 1454
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8364129874949528e-05,
+      "loss": 0.5516,
+      "step": 1455
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.834552874181152e-05,
+      "loss": 0.5289,
+      "step": 1456
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8326918863674028e-05,
+      "loss": 0.5262,
+      "step": 1457
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.5233,
+      "step": 1458
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8289672950412435e-05,
+      "loss": 0.5389,
+      "step": 1459
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8271036954317072e-05,
+      "loss": 0.5278,
+      "step": 1460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8252392291279678e-05,
+      "loss": 0.5233,
+      "step": 1461
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.823373898083736e-05,
+      "loss": 0.5353,
+      "step": 1462
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.821507704253632e-05,
+      "loss": 0.5175,
+      "step": 1463
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8196406495931753e-05,
+      "loss": 0.5196,
+      "step": 1464
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.817772736058791e-05,
+      "loss": 0.5376,
+      "step": 1465
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8159039656078025e-05,
+      "loss": 0.5568,
+      "step": 1466
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8140343401984317e-05,
+      "loss": 0.5354,
+      "step": 1467
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8121638617897968e-05,
+      "loss": 0.5501,
+      "step": 1468
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8102925323419087e-05,
+      "loss": 0.5283,
+      "step": 1469
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.808420353815671e-05,
+      "loss": 0.5383,
+      "step": 1470
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.806547328172877e-05,
+      "loss": 0.5315,
+      "step": 1471
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8046734573762067e-05,
+      "loss": 0.5327,
+      "step": 1472
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8027987433892273e-05,
+      "loss": 0.5439,
+      "step": 1473
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8009231881763877e-05,
+      "loss": 0.5414,
+      "step": 1474
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.799046793703021e-05,
+      "loss": 0.5533,
+      "step": 1475
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7971695619353362e-05,
+      "loss": 0.5452,
+      "step": 1476
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7952914948404214e-05,
+      "loss": 0.5506,
+      "step": 1477
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.793412594386242e-05,
+      "loss": 0.5322,
+      "step": 1478
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.791532862541632e-05,
+      "loss": 0.5267,
+      "step": 1479
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7896523012763017e-05,
+      "loss": 0.5393,
+      "step": 1480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.787770912560827e-05,
+      "loss": 0.5307,
+      "step": 1481
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.785888698366652e-05,
+      "loss": 0.5369,
+      "step": 1482
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7840056606660864e-05,
+      "loss": 0.5423,
+      "step": 1483
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.782121801432302e-05,
+      "loss": 0.5175,
+      "step": 1484
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.780237122639331e-05,
+      "loss": 0.557,
+      "step": 1485
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7783516262620657e-05,
+      "loss": 0.541,
+      "step": 1486
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.776465314276254e-05,
+      "loss": 0.5204,
+      "step": 1487
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7745781886584997e-05,
+      "loss": 0.5434,
+      "step": 1488
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7726902513862572e-05,
+      "loss": 0.5496,
+      "step": 1489
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7708015044378344e-05,
+      "loss": 0.5372,
+      "step": 1490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7689119497923836e-05,
+      "loss": 0.5417,
+      "step": 1491
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7670215894299076e-05,
+      "loss": 0.5342,
+      "step": 1492
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.765130425331251e-05,
+      "loss": 0.5459,
+      "step": 1493
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7632384594781006e-05,
+      "loss": 0.5241,
+      "step": 1494
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.761345693852984e-05,
+      "loss": 0.5385,
+      "step": 1495
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7594521304392676e-05,
+      "loss": 0.5362,
+      "step": 1496
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7575577712211524e-05,
+      "loss": 0.5436,
+      "step": 1497
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7556626181836738e-05,
+      "loss": 0.5217,
+      "step": 1498
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.753766673312699e-05,
+      "loss": 0.5402,
+      "step": 1499
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7518699385949257e-05,
+      "loss": 0.5227,
+      "step": 1500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7499724160178775e-05,
+      "loss": 0.5133,
+      "step": 1501
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7480741075699055e-05,
+      "loss": 0.562,
+      "step": 1502
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7461750152401825e-05,
+      "loss": 0.5142,
+      "step": 1503
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.744275141018705e-05,
+      "loss": 0.5262,
+      "step": 1504
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7423744868962864e-05,
+      "loss": 0.5363,
+      "step": 1505
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7404730548645582e-05,
+      "loss": 0.5463,
+      "step": 1506
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7385708469159674e-05,
+      "loss": 0.5174,
+      "step": 1507
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.736667865043775e-05,
+      "loss": 0.5403,
+      "step": 1508
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7347641112420497e-05,
+      "loss": 0.5429,
+      "step": 1509
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.732859587505673e-05,
+      "loss": 0.513,
+      "step": 1510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7309542958303307e-05,
+      "loss": 0.5491,
+      "step": 1511
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7290482382125145e-05,
+      "loss": 0.5405,
+      "step": 1512
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7271414166495174e-05,
+      "loss": 0.5294,
+      "step": 1513
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7252338331394343e-05,
+      "loss": 0.5262,
+      "step": 1514
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7233254896811576e-05,
+      "loss": 0.5522,
+      "step": 1515
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7214163882743765e-05,
+      "loss": 0.5302,
+      "step": 1516
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7195065309195748e-05,
+      "loss": 0.5107,
+      "step": 1517
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7175959196180267e-05,
+      "loss": 0.5386,
+      "step": 1518
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7156845563717987e-05,
+      "loss": 0.5283,
+      "step": 1519
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.713772443183744e-05,
+      "loss": 0.5444,
+      "step": 1520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7118595820575017e-05,
+      "loss": 0.5234,
+      "step": 1521
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7099459749974947e-05,
+      "loss": 0.5158,
+      "step": 1522
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.708031624008928e-05,
+      "loss": 0.5256,
+      "step": 1523
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7061165310977852e-05,
+      "loss": 0.5413,
+      "step": 1524
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7042006982708282e-05,
+      "loss": 0.532,
+      "step": 1525
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7022841275355934e-05,
+      "loss": 0.524,
+      "step": 1526
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7003668209003926e-05,
+      "loss": 0.5306,
+      "step": 1527
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6984487803743048e-05,
+      "loss": 0.5188,
+      "step": 1528
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6965300079671818e-05,
+      "loss": 0.5425,
+      "step": 1529
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6946105056896406e-05,
+      "loss": 0.5402,
+      "step": 1530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.692690275553063e-05,
+      "loss": 0.5287,
+      "step": 1531
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6907693195695935e-05,
+      "loss": 0.5305,
+      "step": 1532
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.688847639752138e-05,
+      "loss": 0.5336,
+      "step": 1533
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.68692523811436e-05,
+      "loss": 0.5606,
+      "step": 1534
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6850021166706787e-05,
+      "loss": 0.5233,
+      "step": 1535
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6830782774362702e-05,
+      "loss": 0.5291,
+      "step": 1536
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6811537224270604e-05,
+      "loss": 0.5425,
+      "step": 1537
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.679228453659725e-05,
+      "loss": 0.5217,
+      "step": 1538
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6773024731516896e-05,
+      "loss": 0.5484,
+      "step": 1539
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6753757829211246e-05,
+      "loss": 0.54,
+      "step": 1540
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.673448384986943e-05,
+      "loss": 0.5402,
+      "step": 1541
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6715202813688015e-05,
+      "loss": 0.5303,
+      "step": 1542
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6695914740870948e-05,
+      "loss": 0.5384,
+      "step": 1543
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6676619651629548e-05,
+      "loss": 0.5497,
+      "step": 1544
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.66573175661825e-05,
+      "loss": 0.5519,
+      "step": 1545
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6638008504755807e-05,
+      "loss": 0.5644,
+      "step": 1546
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6618692487582784e-05,
+      "loss": 0.5273,
+      "step": 1547
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6599369534904044e-05,
+      "loss": 0.536,
+      "step": 1548
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6580039666967454e-05,
+      "loss": 0.5092,
+      "step": 1549
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6560702904028134e-05,
+      "loss": 0.5285,
+      "step": 1550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6541359266348437e-05,
+      "loss": 0.5486,
+      "step": 1551
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6522008774197902e-05,
+      "loss": 0.5085,
+      "step": 1552
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6502651447853264e-05,
+      "loss": 0.5165,
+      "step": 1553
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6483287307598425e-05,
+      "loss": 0.5288,
+      "step": 1554
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6463916373724394e-05,
+      "loss": 0.5338,
+      "step": 1555
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6444538666529343e-05,
+      "loss": 0.5584,
+      "step": 1556
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6425154206318517e-05,
+      "loss": 0.5199,
+      "step": 1557
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.640576301340424e-05,
+      "loss": 0.5237,
+      "step": 1558
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.638636510810589e-05,
+      "loss": 0.5287,
+      "step": 1559
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.636696051074988e-05,
+      "loss": 0.5234,
+      "step": 1560
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6347549241669646e-05,
+      "loss": 0.5429,
+      "step": 1561
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6328131321205588e-05,
+      "loss": 0.5401,
+      "step": 1562
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6308706769705118e-05,
+      "loss": 0.5191,
+      "step": 1563
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.628927560752254e-05,
+      "loss": 0.5306,
+      "step": 1564
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6269837855019145e-05,
+      "loss": 0.5321,
+      "step": 1565
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.625039353256309e-05,
+      "loss": 0.5375,
+      "step": 1566
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6230942660529423e-05,
+      "loss": 0.5389,
+      "step": 1567
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6211485259300062e-05,
+      "loss": 0.5456,
+      "step": 1568
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6192021349263773e-05,
+      "loss": 0.5549,
+      "step": 1569
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.617255095081613e-05,
+      "loss": 0.5407,
+      "step": 1570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6153074084359505e-05,
+      "loss": 0.5305,
+      "step": 1571
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6133590770303064e-05,
+      "loss": 0.5595,
+      "step": 1572
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6114101029062707e-05,
+      "loss": 0.5368,
+      "step": 1573
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6094604881061076e-05,
+      "loss": 0.5296,
+      "step": 1574
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.607510234672755e-05,
+      "loss": 0.5218,
+      "step": 1575
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.605559344649816e-05,
+      "loss": 0.5472,
+      "step": 1576
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.603607820081563e-05,
+      "loss": 0.5408,
+      "step": 1577
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6016556630129352e-05,
+      "loss": 0.5457,
+      "step": 1578
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5997028754895294e-05,
+      "loss": 0.5406,
+      "step": 1579
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.597749459557608e-05,
+      "loss": 0.5227,
+      "step": 1580
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.59579541726409e-05,
+      "loss": 0.551,
+      "step": 1581
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.59384075065655e-05,
+      "loss": 0.5316,
+      "step": 1582
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5918854617832182e-05,
+      "loss": 0.5298,
+      "step": 1583
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5899295526929758e-05,
+      "loss": 0.5348,
+      "step": 1584
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5879730254353543e-05,
+      "loss": 0.5369,
+      "step": 1585
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.586015882060533e-05,
+      "loss": 0.5241,
+      "step": 1586
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5840581246193364e-05,
+      "loss": 0.5247,
+      "step": 1587
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.582099755163234e-05,
+      "loss": 0.526,
+      "step": 1588
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5801407757443335e-05,
+      "loss": 0.5242,
+      "step": 1589
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5781811884153856e-05,
+      "loss": 0.536,
+      "step": 1590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.576220995229775e-05,
+      "loss": 0.5519,
+      "step": 1591
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.574260198241523e-05,
+      "loss": 0.5389,
+      "step": 1592
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5722987995052817e-05,
+      "loss": 0.552,
+      "step": 1593
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5703368010763367e-05,
+      "loss": 0.5177,
+      "step": 1594
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5683742050105994e-05,
+      "loss": 0.5385,
+      "step": 1595
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.566411013364608e-05,
+      "loss": 0.5301,
+      "step": 1596
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5644472281955267e-05,
+      "loss": 0.5285,
+      "step": 1597
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.562482851561138e-05,
+      "loss": 0.5254,
+      "step": 1598
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.560517885519847e-05,
+      "loss": 0.5285,
+      "step": 1599
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5585523321306767e-05,
+      "loss": 0.5485,
+      "step": 1600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5565861934532632e-05,
+      "loss": 0.5407,
+      "step": 1601
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5546194715478576e-05,
+      "loss": 0.5389,
+      "step": 1602
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5526521684753226e-05,
+      "loss": 0.5226,
+      "step": 1603
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.550684286297128e-05,
+      "loss": 0.5504,
+      "step": 1604
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.548715827075352e-05,
+      "loss": 0.5353,
+      "step": 1605
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5467467928726764e-05,
+      "loss": 0.5293,
+      "step": 1606
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5447771857523868e-05,
+      "loss": 0.5487,
+      "step": 1607
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5428070077783682e-05,
+      "loss": 0.5309,
+      "step": 1608
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5408362610151037e-05,
+      "loss": 0.5195,
+      "step": 1609
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5388649475276726e-05,
+      "loss": 0.521,
+      "step": 1610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5368930693817483e-05,
+      "loss": 0.5233,
+      "step": 1611
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5349206286435966e-05,
+      "loss": 0.5303,
+      "step": 1612
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.53294762738007e-05,
+      "loss": 0.5263,
+      "step": 1613
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5309740676586122e-05,
+      "loss": 0.5315,
+      "step": 1614
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.52899995154725e-05,
+      "loss": 0.5361,
+      "step": 1615
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.527025281114592e-05,
+      "loss": 0.5387,
+      "step": 1616
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5250500584298297e-05,
+      "loss": 0.5175,
+      "step": 1617
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.523074285562734e-05,
+      "loss": 0.5186,
+      "step": 1618
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.521097964583649e-05,
+      "loss": 0.5814,
+      "step": 1619
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5191210975634962e-05,
+      "loss": 0.534,
+      "step": 1620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5171436865737687e-05,
+      "loss": 0.5428,
+      "step": 1621
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5151657336865275e-05,
+      "loss": 0.5287,
+      "step": 1622
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.513187240974404e-05,
+      "loss": 0.5381,
+      "step": 1623
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5112082105105954e-05,
+      "loss": 0.5337,
+      "step": 1624
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5092286443688592e-05,
+      "loss": 0.5293,
+      "step": 1625
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5072485446235168e-05,
+      "loss": 0.5284,
+      "step": 1626
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5052679133494504e-05,
+      "loss": 0.5261,
+      "step": 1627
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5032867526220938e-05,
+      "loss": 0.5281,
+      "step": 1628
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5013050645174414e-05,
+      "loss": 0.5502,
+      "step": 1629
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.4993228511120365e-05,
+      "loss": 0.5197,
+      "step": 1630
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.497340114482974e-05,
+      "loss": 0.5328,
+      "step": 1631
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.4953568567078973e-05,
+      "loss": 0.5095,
+      "step": 1632
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.493373079864996e-05,
+      "loss": 0.5384,
+      "step": 1633
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4913887860330026e-05,
+      "loss": 0.5322,
+      "step": 1634
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.489403977291192e-05,
+      "loss": 0.5588,
+      "step": 1635
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.48741865571938e-05,
+      "loss": 0.5388,
+      "step": 1636
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4854328233979167e-05,
+      "loss": 0.5525,
+      "step": 1637
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.48344648240769e-05,
+      "loss": 0.5207,
+      "step": 1638
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.48145963483012e-05,
+      "loss": 0.5433,
+      "step": 1639
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.479472282747157e-05,
+      "loss": 0.53,
+      "step": 1640
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4774844282412808e-05,
+      "loss": 0.5271,
+      "step": 1641
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.475496073395498e-05,
+      "loss": 0.5225,
+      "step": 1642
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4735072202933387e-05,
+      "loss": 0.5361,
+      "step": 1643
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.471517871018855e-05,
+      "loss": 0.5363,
+      "step": 1644
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.469528027656619e-05,
+      "loss": 0.5395,
+      "step": 1645
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4675376922917215e-05,
+      "loss": 0.5194,
+      "step": 1646
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4655468670097675e-05,
+      "loss": 0.5282,
+      "step": 1647
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4635555538968764e-05,
+      "loss": 0.5327,
+      "step": 1648
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.461563755039678e-05,
+      "loss": 0.5374,
+      "step": 1649
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.459571472525312e-05,
+      "loss": 0.5378,
+      "step": 1650
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4575787084414244e-05,
+      "loss": 0.527,
+      "step": 1651
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4555854648761644e-05,
+      "loss": 0.5332,
+      "step": 1652
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.453591743918187e-05,
+      "loss": 0.5374,
+      "step": 1653
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4515975476566444e-05,
+      "loss": 0.5221,
+      "step": 1654
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4496028781811882e-05,
+      "loss": 0.5253,
+      "step": 1655
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4476077375819657e-05,
+      "loss": 0.5344,
+      "step": 1656
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.445612127949617e-05,
+      "loss": 0.5432,
+      "step": 1657
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.443616051375276e-05,
+      "loss": 0.5177,
+      "step": 1658
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4416195099505627e-05,
+      "loss": 0.5349,
+      "step": 1659
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4396225057675863e-05,
+      "loss": 0.5286,
+      "step": 1660
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4376250409189415e-05,
+      "loss": 0.5474,
+      "step": 1661
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.435627117497703e-05,
+      "loss": 0.5383,
+      "step": 1662
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4336287375974294e-05,
+      "loss": 0.5438,
+      "step": 1663
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.431629903312155e-05,
+      "loss": 0.5215,
+      "step": 1664
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.429630616736391e-05,
+      "loss": 0.5351,
+      "step": 1665
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4276308799651234e-05,
+      "loss": 0.5315,
+      "step": 1666
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4256306950938086e-05,
+      "loss": 0.537,
+      "step": 1667
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.423630064218373e-05,
+      "loss": 0.5318,
+      "step": 1668
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4216289894352118e-05,
+      "loss": 0.5396,
+      "step": 1669
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.419627472841183e-05,
+      "loss": 0.5217,
+      "step": 1670
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4176255165336092e-05,
+      "loss": 0.5371,
+      "step": 1671
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.415623122610272e-05,
+      "loss": 0.539,
+      "step": 1672
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.413620293169415e-05,
+      "loss": 0.5278,
+      "step": 1673
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4116170303097336e-05,
+      "loss": 0.5053,
+      "step": 1674
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4096133361303816e-05,
+      "loss": 0.5514,
+      "step": 1675
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4076092127309624e-05,
+      "loss": 0.515,
+      "step": 1676
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.405604662211528e-05,
+      "loss": 0.5307,
+      "step": 1677
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4035996866725824e-05,
+      "loss": 0.5399,
+      "step": 1678
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.40159428821507e-05,
+      "loss": 0.516,
+      "step": 1679
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3995884689403814e-05,
+      "loss": 0.5341,
+      "step": 1680
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3975822309503473e-05,
+      "loss": 0.5318,
+      "step": 1681
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3955755763472373e-05,
+      "loss": 0.5312,
+      "step": 1682
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.393568507233757e-05,
+      "loss": 0.5451,
+      "step": 1683
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3915610257130464e-05,
+      "loss": 0.5583,
+      "step": 1684
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3895531338886793e-05,
+      "loss": 0.5281,
+      "step": 1685
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.387544833864657e-05,
+      "loss": 0.5317,
+      "step": 1686
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.38553612774541e-05,
+      "loss": 0.5401,
+      "step": 1687
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3835270176357948e-05,
+      "loss": 0.5381,
+      "step": 1688
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.381517505641089e-05,
+      "loss": 0.5384,
+      "step": 1689
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3795075938669935e-05,
+      "loss": 0.5263,
+      "step": 1690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3774972844196277e-05,
+      "loss": 0.5283,
+      "step": 1691
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3754865794055274e-05,
+      "loss": 0.5398,
+      "step": 1692
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3734754809316423e-05,
+      "loss": 0.5309,
+      "step": 1693
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3714639911053352e-05,
+      "loss": 0.5493,
+      "step": 1694
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.369452112034379e-05,
+      "loss": 0.5193,
+      "step": 1695
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3674398458269537e-05,
+      "loss": 0.5288,
+      "step": 1696
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.365427194591647e-05,
+      "loss": 0.5084,
+      "step": 1697
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.363414160437447e-05,
+      "loss": 0.5386,
+      "step": 1698
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3614007454737448e-05,
+      "loss": 0.5082,
+      "step": 1699
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3593869518103318e-05,
+      "loss": 0.5437,
+      "step": 1700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.357372781557393e-05,
+      "loss": 0.5239,
+      "step": 1701
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3553582368255107e-05,
+      "loss": 0.5408,
+      "step": 1702
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3533433197256583e-05,
+      "loss": 0.5499,
+      "step": 1703
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3513280323692004e-05,
+      "loss": 0.5305,
+      "step": 1704
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.349312376867889e-05,
+      "loss": 0.5377,
+      "step": 1705
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.5384,
+      "step": 1706
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3452799698796382e-05,
+      "loss": 0.5286,
+      "step": 1707
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3432632226181237e-05,
+      "loss": 0.5275,
+      "step": 1708
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3412461156625987e-05,
+      "loss": 0.533,
+      "step": 1709
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3392286511267215e-05,
+      "loss": 0.5343,
+      "step": 1710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.337210831124526e-05,
+      "loss": 0.5054,
+      "step": 1711
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.335192657770419e-05,
+      "loss": 0.5479,
+      "step": 1712
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3331741331791748e-05,
+      "loss": 0.5421,
+      "step": 1713
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3311552594659387e-05,
+      "loss": 0.5468,
+      "step": 1714
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3291360387462207e-05,
+      "loss": 0.5562,
+      "step": 1715
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3271164731358936e-05,
+      "loss": 0.5156,
+      "step": 1716
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.325096564751193e-05,
+      "loss": 0.5401,
+      "step": 1717
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3230763157087136e-05,
+      "loss": 0.5337,
+      "step": 1718
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3210557281254058e-05,
+      "loss": 0.5246,
+      "step": 1719
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3190348041185756e-05,
+      "loss": 0.5357,
+      "step": 1720
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3170135458058812e-05,
+      "loss": 0.5317,
+      "step": 1721
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3149919553053333e-05,
+      "loss": 0.5369,
+      "step": 1722
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.312970034735286e-05,
+      "loss": 0.5362,
+      "step": 1723
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3109477862144432e-05,
+      "loss": 0.5229,
+      "step": 1724
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3089252118618522e-05,
+      "loss": 0.5162,
+      "step": 1725
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.306902313796899e-05,
+      "loss": 0.5352,
+      "step": 1726
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3048790941393113e-05,
+      "loss": 0.5492,
+      "step": 1727
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3028555550091536e-05,
+      "loss": 0.5218,
+      "step": 1728
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.300831698526824e-05,
+      "loss": 0.5428,
+      "step": 1729
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2988075268130542e-05,
+      "loss": 0.5316,
+      "step": 1730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2967830419889058e-05,
+      "loss": 0.5229,
+      "step": 1731
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.294758246175768e-05,
+      "loss": 0.5283,
+      "step": 1732
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.292733141495357e-05,
+      "loss": 0.5395,
+      "step": 1733
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2907077300697116e-05,
+      "loss": 0.5206,
+      "step": 1734
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2886820140211926e-05,
+      "loss": 0.5241,
+      "step": 1735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2866559954724798e-05,
+      "loss": 0.5126,
+      "step": 1736
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.5547,
+      "step": 1737
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2826030593667758e-05,
+      "loss": 0.5414,
+      "step": 1738
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2805761460567197e-05,
+      "loss": 0.5449,
+      "step": 1739
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2785489387403367e-05,
+      "loss": 0.5284,
+      "step": 1740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2765214395418695e-05,
+      "loss": 0.5497,
+      "step": 1741
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.274493650585867e-05,
+      "loss": 0.5343,
+      "step": 1742
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.27246557399718e-05,
+      "loss": 0.5366,
+      "step": 1743
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2704372119009626e-05,
+      "loss": 0.5311,
+      "step": 1744
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.268408566422667e-05,
+      "loss": 0.5361,
+      "step": 1745
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.266379639688044e-05,
+      "loss": 0.5357,
+      "step": 1746
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.264350433823136e-05,
+      "loss": 0.5307,
+      "step": 1747
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2623209509542802e-05,
+      "loss": 0.552,
+      "step": 1748
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.260291193208105e-05,
+      "loss": 0.5358,
+      "step": 1749
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.258261162711523e-05,
+      "loss": 0.5235,
+      "step": 1750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2562308615917373e-05,
+      "loss": 0.5274,
+      "step": 1751
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.254200291976231e-05,
+      "loss": 0.5375,
+      "step": 1752
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2521694559927703e-05,
+      "loss": 0.522,
+      "step": 1753
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2501383557694002e-05,
+      "loss": 0.5333,
+      "step": 1754
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2481069934344418e-05,
+      "loss": 0.518,
+      "step": 1755
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2460753711164927e-05,
+      "loss": 0.5424,
+      "step": 1756
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2440434909444204e-05,
+      "loss": 0.5341,
+      "step": 1757
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.242011355047365e-05,
+      "loss": 0.5242,
+      "step": 1758
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.239978965554732e-05,
+      "loss": 0.5465,
+      "step": 1759
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.237946324596196e-05,
+      "loss": 0.5436,
+      "step": 1760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2359134343016926e-05,
+      "loss": 0.541,
+      "step": 1761
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2338802968014184e-05,
+      "loss": 0.5257,
+      "step": 1762
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.231846914225831e-05,
+      "loss": 0.5339,
+      "step": 1763
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2298132887056437e-05,
+      "loss": 0.5384,
+      "step": 1764
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.227779422371824e-05,
+      "loss": 0.5347,
+      "step": 1765
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2257453173555922e-05,
+      "loss": 0.5107,
+      "step": 1766
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.223710975788419e-05,
+      "loss": 0.5431,
+      "step": 1767
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2216763998020222e-05,
+      "loss": 0.5354,
+      "step": 1768
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.219641591528366e-05,
+      "loss": 0.5177,
+      "step": 1769
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2176065530996583e-05,
+      "loss": 0.5396,
+      "step": 1770
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2155712866483465e-05,
+      "loss": 0.5148,
+      "step": 1771
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.213535794307118e-05,
+      "loss": 0.5321,
+      "step": 1772
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2115000782088977e-05,
+      "loss": 0.5169,
+      "step": 1773
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.209464140486844e-05,
+      "loss": 0.5345,
+      "step": 1774
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.207427983274347e-05,
+      "loss": 0.5239,
+      "step": 1775
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.205391608705029e-05,
+      "loss": 0.5348,
+      "step": 1776
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.203355018912736e-05,
+      "loss": 0.5091,
+      "step": 1777
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2013182160315448e-05,
+      "loss": 0.522,
+      "step": 1778
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.1992812021957513e-05,
+      "loss": 0.5347,
+      "step": 1779
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.197243979539875e-05,
+      "loss": 0.5389,
+      "step": 1780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1952065501986518e-05,
+      "loss": 0.5346,
+      "step": 1781
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1931689163070358e-05,
+      "loss": 0.5156,
+      "step": 1782
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1911310800001967e-05,
+      "loss": 0.5389,
+      "step": 1783
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1890930434135125e-05,
+      "loss": 0.5248,
+      "step": 1784
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.187054808682575e-05,
+      "loss": 0.5277,
+      "step": 1785
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.185016377943182e-05,
+      "loss": 0.5105,
+      "step": 1786
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1829777533313346e-05,
+      "loss": 0.5473,
+      "step": 1787
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1809389369832418e-05,
+      "loss": 0.5378,
+      "step": 1788
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.178899931035309e-05,
+      "loss": 0.5492,
+      "step": 1789
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.176860737624143e-05,
+      "loss": 0.5386,
+      "step": 1790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1748213588865455e-05,
+      "loss": 0.5413,
+      "step": 1791
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.172781796959514e-05,
+      "loss": 0.5153,
+      "step": 1792
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.170742053980236e-05,
+      "loss": 0.5409,
+      "step": 1793
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1687021320860893e-05,
+      "loss": 0.5392,
+      "step": 1794
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.166662033414642e-05,
+      "loss": 0.5262,
+      "step": 1795
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1646217601036422e-05,
+      "loss": 0.532,
+      "step": 1796
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1625813142910258e-05,
+      "loss": 0.5313,
+      "step": 1797
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1605406981149066e-05,
+      "loss": 0.521,
+      "step": 1798
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.158499913713577e-05,
+      "loss": 0.5274,
+      "step": 1799
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1564589632255082e-05,
+      "loss": 0.5251,
+      "step": 1800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1544178487893427e-05,
+      "loss": 0.5293,
+      "step": 1801
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1523765725438962e-05,
+      "loss": 0.5324,
+      "step": 1802
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.150335136628153e-05,
+      "loss": 0.5268,
+      "step": 1803
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1482935431812663e-05,
+      "loss": 0.5337,
+      "step": 1804
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1462517943425523e-05,
+      "loss": 0.5482,
+      "step": 1805
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.144209892251491e-05,
+      "loss": 0.5487,
+      "step": 1806
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.142167839047725e-05,
+      "loss": 0.5284,
+      "step": 1807
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.140125636871051e-05,
+      "loss": 0.5187,
+      "step": 1808
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1380832878614247e-05,
+      "loss": 0.5216,
+      "step": 1809
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.136040794158956e-05,
+      "loss": 0.5351,
+      "step": 1810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1339981579039045e-05,
+      "loss": 0.5209,
+      "step": 1811
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1319553812366806e-05,
+      "loss": 0.5223,
+      "step": 1812
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.129912466297842e-05,
+      "loss": 0.5222,
+      "step": 1813
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1278694152280895e-05,
+      "loss": 0.5327,
+      "step": 1814
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1258262301682685e-05,
+      "loss": 0.5311,
+      "step": 1815
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.123782913259364e-05,
+      "loss": 0.5323,
+      "step": 1816
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.121739466642499e-05,
+      "loss": 0.542,
+      "step": 1817
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.119695892458933e-05,
+      "loss": 0.5147,
+      "step": 1818
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.117652192850058e-05,
+      "loss": 0.528,
+      "step": 1819
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.115608369957399e-05,
+      "loss": 0.5406,
+      "step": 1820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1135644259226087e-05,
+      "loss": 0.5135,
+      "step": 1821
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.111520362887469e-05,
+      "loss": 0.5392,
+      "step": 1822
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1094761829938826e-05,
+      "loss": 0.5189,
+      "step": 1823
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.107431888383879e-05,
+      "loss": 0.5233,
+      "step": 1824
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.105387481199606e-05,
+      "loss": 0.559,
+      "step": 1825
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.103342963583327e-05,
+      "loss": 0.5136,
+      "step": 1826
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1012983376774255e-05,
+      "loss": 0.5231,
+      "step": 1827
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.099253605624396e-05,
+      "loss": 0.528,
+      "step": 1828
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.097208769566844e-05,
+      "loss": 0.5298,
+      "step": 1829
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.095163831647485e-05,
+      "loss": 0.527,
+      "step": 1830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0931187940091407e-05,
+      "loss": 0.5249,
+      "step": 1831
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0910736587947373e-05,
+      "loss": 0.5375,
+      "step": 1832
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.089028428147303e-05,
+      "loss": 0.5317,
+      "step": 1833
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0869831042099678e-05,
+      "loss": 0.5242,
+      "step": 1834
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0849376891259565e-05,
+      "loss": 0.5318,
+      "step": 1835
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0828921850385914e-05,
+      "loss": 0.5361,
+      "step": 1836
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0808465940912882e-05,
+      "loss": 0.5417,
+      "step": 1837
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0788009184275514e-05,
+      "loss": 0.5268,
+      "step": 1838
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0767551601909774e-05,
+      "loss": 0.5453,
+      "step": 1839
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0747093215252476e-05,
+      "loss": 0.534,
+      "step": 1840
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0726634045741262e-05,
+      "loss": 0.5149,
+      "step": 1841
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.070617411481463e-05,
+      "loss": 0.5231,
+      "step": 1842
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0685713443911837e-05,
+      "loss": 0.5337,
+      "step": 1843
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.066525205447294e-05,
+      "loss": 0.5225,
+      "step": 1844
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0644789967938738e-05,
+      "loss": 0.5374,
+      "step": 1845
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0624327205750773e-05,
+      "loss": 0.5224,
+      "step": 1846
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.060386378935129e-05,
+      "loss": 0.548,
+      "step": 1847
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.05833997401832e-05,
+      "loss": 0.5306,
+      "step": 1848
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.05629350796901e-05,
+      "loss": 0.5426,
+      "step": 1849
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0542469829316234e-05,
+      "loss": 0.5421,
+      "step": 1850
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0522004010506436e-05,
+      "loss": 0.5236,
+      "step": 1851
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0501537644706153e-05,
+      "loss": 0.5242,
+      "step": 1852
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0481070753361404e-05,
+      "loss": 0.521,
+      "step": 1853
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.046060335791876e-05,
+      "loss": 0.5373,
+      "step": 1854
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0440135479825313e-05,
+      "loss": 0.5304,
+      "step": 1855
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0419667140528675e-05,
+      "loss": 0.531,
+      "step": 1856
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0399198361476925e-05,
+      "loss": 0.5399,
+      "step": 1857
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.03787291641186e-05,
+      "loss": 0.522,
+      "step": 1858
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.035825956990271e-05,
+      "loss": 0.5037,
+      "step": 1859
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0337789600278623e-05,
+      "loss": 0.5282,
+      "step": 1860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.031731927669616e-05,
+      "loss": 0.5223,
+      "step": 1861
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0296848620605475e-05,
+      "loss": 0.5289,
+      "step": 1862
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0276377653457087e-05,
+      "loss": 0.5226,
+      "step": 1863
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.025590639670183e-05,
+      "loss": 0.5213,
+      "step": 1864
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.023543487179085e-05,
+      "loss": 0.539,
+      "step": 1865
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0214963100175565e-05,
+      "loss": 0.5332,
+      "step": 1866
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0194491103307662e-05,
+      "loss": 0.5308,
+      "step": 1867
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0174018902639067e-05,
+      "loss": 0.5298,
+      "step": 1868
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.01535465196219e-05,
+      "loss": 0.5048,
+      "step": 1869
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0133073975708482e-05,
+      "loss": 0.5189,
+      "step": 1870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0112601292351322e-05,
+      "loss": 0.5494,
+      "step": 1871
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0092128491003038e-05,
+      "loss": 0.4863,
+      "step": 1872
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0071655593116403e-05,
+      "loss": 0.528,
+      "step": 1873
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.005118262014428e-05,
+      "loss": 0.5243,
+      "step": 1874
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0030709593539607e-05,
+      "loss": 0.5195,
+      "step": 1875
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0010236534755378e-05,
+      "loss": 0.5233,
+      "step": 1876
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9989763465244625e-05,
+      "loss": 0.5373,
+      "step": 1877
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9969290406460403e-05,
+      "loss": 0.5084,
+      "step": 1878
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9948817379855728e-05,
+      "loss": 0.5212,
+      "step": 1879
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9928344406883597e-05,
+      "loss": 0.5348,
+      "step": 1880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.990787150899697e-05,
+      "loss": 0.5085,
+      "step": 1881
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.988739870764869e-05,
+      "loss": 0.5389,
+      "step": 1882
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9866926024291518e-05,
+      "loss": 0.5423,
+      "step": 1883
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9846453480378106e-05,
+      "loss": 0.5347,
+      "step": 1884
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9825981097360946e-05,
+      "loss": 0.5361,
+      "step": 1885
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.980550889669234e-05,
+      "loss": 0.5352,
+      "step": 1886
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9785036899824438e-05,
+      "loss": 0.5144,
+      "step": 1887
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.976456512820916e-05,
+      "loss": 0.524,
+      "step": 1888
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9744093603298177e-05,
+      "loss": 0.5168,
+      "step": 1889
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.972362234654292e-05,
+      "loss": 0.5238,
+      "step": 1890
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9703151379394528e-05,
+      "loss": 0.5435,
+      "step": 1891
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9682680723303842e-05,
+      "loss": 0.5324,
+      "step": 1892
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.966221039972138e-05,
+      "loss": 0.5105,
+      "step": 1893
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9641740430097304e-05,
+      "loss": 0.5358,
+      "step": 1894
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9621270835881402e-05,
+      "loss": 0.5426,
+      "step": 1895
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9600801638523085e-05,
+      "loss": 0.5319,
+      "step": 1896
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9580332859471335e-05,
+      "loss": 0.5443,
+      "step": 1897
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.955986452017469e-05,
+      "loss": 0.5274,
+      "step": 1898
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9539396642081244e-05,
+      "loss": 0.5294,
+      "step": 1899
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9518929246638606e-05,
+      "loss": 0.5375,
+      "step": 1900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9498462355293854e-05,
+      "loss": 0.522,
+      "step": 1901
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.947799598949357e-05,
+      "loss": 0.5252,
+      "step": 1902
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9457530170683773e-05,
+      "loss": 0.5338,
+      "step": 1903
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9437064920309895e-05,
+      "loss": 0.5126,
+      "step": 1904
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9416600259816806e-05,
+      "loss": 0.5111,
+      "step": 1905
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.939613621064872e-05,
+      "loss": 0.5199,
+      "step": 1906
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9375672794249227e-05,
+      "loss": 0.5202,
+      "step": 1907
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9355210032061266e-05,
+      "loss": 0.5483,
+      "step": 1908
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.933474794552707e-05,
+      "loss": 0.5197,
+      "step": 1909
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9314286556088173e-05,
+      "loss": 0.5417,
+      "step": 1910
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9293825885185378e-05,
+      "loss": 0.517,
+      "step": 1911
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9273365954258745e-05,
+      "loss": 0.5472,
+      "step": 1912
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.925290678474753e-05,
+      "loss": 0.5227,
+      "step": 1913
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9232448398090232e-05,
+      "loss": 0.518,
+      "step": 1914
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9211990815724496e-05,
+      "loss": 0.5155,
+      "step": 1915
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9191534059087125e-05,
+      "loss": 0.5294,
+      "step": 1916
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9171078149614092e-05,
+      "loss": 0.5064,
+      "step": 1917
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9150623108740442e-05,
+      "loss": 0.5121,
+      "step": 1918
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9130168957900325e-05,
+      "loss": 0.5195,
+      "step": 1919
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9109715718526972e-05,
+      "loss": 0.5206,
+      "step": 1920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9089263412052637e-05,
+      "loss": 0.5425,
+      "step": 1921
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.90688120599086e-05,
+      "loss": 0.535,
+      "step": 1922
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9048361683525155e-05,
+      "loss": 0.532,
+      "step": 1923
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.902791230433157e-05,
+      "loss": 0.5278,
+      "step": 1924
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9007463943756045e-05,
+      "loss": 0.5368,
+      "step": 1925
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8987016623225748e-05,
+      "loss": 0.5343,
+      "step": 1926
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.896657036416674e-05,
+      "loss": 0.5302,
+      "step": 1927
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8946125188003946e-05,
+      "loss": 0.5284,
+      "step": 1928
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8925681116161213e-05,
+      "loss": 0.541,
+      "step": 1929
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8905238170061177e-05,
+      "loss": 0.5184,
+      "step": 1930
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8884796371125316e-05,
+      "loss": 0.5269,
+      "step": 1931
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8864355740773916e-05,
+      "loss": 0.5392,
+      "step": 1932
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.884391630042602e-05,
+      "loss": 0.5457,
+      "step": 1933
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8823478071499427e-05,
+      "loss": 0.5228,
+      "step": 1934
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8803041075410677e-05,
+      "loss": 0.5392,
+      "step": 1935
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8782605333575018e-05,
+      "loss": 0.5183,
+      "step": 1936
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8762170867406366e-05,
+      "loss": 0.5302,
+      "step": 1937
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.874173769831732e-05,
+      "loss": 0.5436,
+      "step": 1938
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8721305847719115e-05,
+      "loss": 0.5541,
+      "step": 1939
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.870087533702159e-05,
+      "loss": 0.5373,
+      "step": 1940
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8680446187633198e-05,
+      "loss": 0.5295,
+      "step": 1941
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.866001842096096e-05,
+      "loss": 0.5227,
+      "step": 1942
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8639592058410445e-05,
+      "loss": 0.5131,
+      "step": 1943
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.861916712138576e-05,
+      "loss": 0.5204,
+      "step": 1944
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8598743631289498e-05,
+      "loss": 0.5329,
+      "step": 1945
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.857832160952276e-05,
+      "loss": 0.5027,
+      "step": 1946
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8557901077485093e-05,
+      "loss": 0.5129,
+      "step": 1947
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.853748205657448e-05,
+      "loss": 0.5293,
+      "step": 1948
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8517064568187344e-05,
+      "loss": 0.5197,
+      "step": 1949
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8496648633718473e-05,
+      "loss": 0.5064,
+      "step": 1950
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.847623427456104e-05,
+      "loss": 0.5469,
+      "step": 1951
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.845582151210658e-05,
+      "loss": 0.5573,
+      "step": 1952
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8435410367744925e-05,
+      "loss": 0.524,
+      "step": 1953
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.841500086286423e-05,
+      "loss": 0.5264,
+      "step": 1954
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.839459301885094e-05,
+      "loss": 0.5217,
+      "step": 1955
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8374186857089752e-05,
+      "loss": 0.5433,
+      "step": 1956
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.835378239896358e-05,
+      "loss": 0.5152,
+      "step": 1957
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8333379665853585e-05,
+      "loss": 0.5326,
+      "step": 1958
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.831297867913911e-05,
+      "loss": 0.5091,
+      "step": 1959
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8292579460197643e-05,
+      "loss": 0.5292,
+      "step": 1960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8272182030404866e-05,
+      "loss": 0.5373,
+      "step": 1961
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8251786411134548e-05,
+      "loss": 0.527,
+      "step": 1962
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.823139262375857e-05,
+      "loss": 0.5347,
+      "step": 1963
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8211000689646916e-05,
+      "loss": 0.5481,
+      "step": 1964
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.819061063016759e-05,
+      "loss": 0.5297,
+      "step": 1965
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8170222466686654e-05,
+      "loss": 0.5202,
+      "step": 1966
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8149836220568188e-05,
+      "loss": 0.509,
+      "step": 1967
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8129451913174256e-05,
+      "loss": 0.5349,
+      "step": 1968
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8109069565864878e-05,
+      "loss": 0.5222,
+      "step": 1969
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.808868919999804e-05,
+      "loss": 0.5313,
+      "step": 1970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8068310836929645e-05,
+      "loss": 0.5418,
+      "step": 1971
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.804793449801349e-05,
+      "loss": 0.5141,
+      "step": 1972
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8027560204601256e-05,
+      "loss": 0.5272,
+      "step": 1973
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.800718797804249e-05,
+      "loss": 0.5219,
+      "step": 1974
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7986817839684552e-05,
+      "loss": 0.5211,
+      "step": 1975
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7966449810872645e-05,
+      "loss": 0.5161,
+      "step": 1976
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7946083912949723e-05,
+      "loss": 0.5271,
+      "step": 1977
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7925720167256533e-05,
+      "loss": 0.5298,
+      "step": 1978
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.790535859513157e-05,
+      "loss": 0.5314,
+      "step": 1979
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7884999217911033e-05,
+      "loss": 0.5281,
+      "step": 1980
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7864642056928823e-05,
+      "loss": 0.519,
+      "step": 1981
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7844287133516542e-05,
+      "loss": 0.5214,
+      "step": 1982
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7823934469003427e-05,
+      "loss": 0.5022,
+      "step": 1983
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7803584084716342e-05,
+      "loss": 0.5136,
+      "step": 1984
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.778323600197978e-05,
+      "loss": 0.5311,
+      "step": 1985
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.776289024211582e-05,
+      "loss": 0.5288,
+      "step": 1986
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7742546826444078e-05,
+      "loss": 0.516,
+      "step": 1987
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7722205776281767e-05,
+      "loss": 0.4978,
+      "step": 1988
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.770186711294357e-05,
+      "loss": 0.5152,
+      "step": 1989
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7681530857741692e-05,
+      "loss": 0.5083,
+      "step": 1990
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7661197031985822e-05,
+      "loss": 0.5185,
+      "step": 1991
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7640865656983084e-05,
+      "loss": 0.5097,
+      "step": 1992
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7620536754038044e-05,
+      "loss": 0.5229,
+      "step": 1993
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7600210344452682e-05,
+      "loss": 0.5157,
+      "step": 1994
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7579886449526364e-05,
+      "loss": 0.5204,
+      "step": 1995
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7559565090555802e-05,
+      "loss": 0.5309,
+      "step": 1996
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.753924628883508e-05,
+      "loss": 0.5202,
+      "step": 1997
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.751893006565559e-05,
+      "loss": 0.5408,
+      "step": 1998
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7498616442305998e-05,
+      "loss": 0.5366,
+      "step": 1999
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.74783054400723e-05,
+      "loss": 0.5003,
+      "step": 2000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7457997080237693e-05,
+      "loss": 0.5221,
+      "step": 2001
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.743769138408263e-05,
+      "loss": 0.5136,
+      "step": 2002
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7417388372884775e-05,
+      "loss": 0.5153,
+      "step": 2003
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.739708806791896e-05,
+      "loss": 0.5409,
+      "step": 2004
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.73767904904572e-05,
+      "loss": 0.5161,
+      "step": 2005
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7356495661768646e-05,
+      "loss": 0.5364,
+      "step": 2006
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.733620360311957e-05,
+      "loss": 0.5154,
+      "step": 2007
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7315914335773333e-05,
+      "loss": 0.5367,
+      "step": 2008
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.729562788099038e-05,
+      "loss": 0.5355,
+      "step": 2009
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.727534426002821e-05,
+      "loss": 0.5313,
+      "step": 2010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7255063494141338e-05,
+      "loss": 0.5229,
+      "step": 2011
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7234785604581308e-05,
+      "loss": 0.5304,
+      "step": 2012
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7214510612596637e-05,
+      "loss": 0.515,
+      "step": 2013
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7194238539432807e-05,
+      "loss": 0.5395,
+      "step": 2014
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7173969406332252e-05,
+      "loss": 0.5275,
+      "step": 2015
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.5229,
+      "step": 2016
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.71334400452752e-05,
+      "loss": 0.5251,
+      "step": 2017
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7113179859788078e-05,
+      "loss": 0.5388,
+      "step": 2018
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.709292269930289e-05,
+      "loss": 0.5296,
+      "step": 2019
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7072668585046432e-05,
+      "loss": 0.5471,
+      "step": 2020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7052417538242322e-05,
+      "loss": 0.5158,
+      "step": 2021
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7032169580110952e-05,
+      "loss": 0.526,
+      "step": 2022
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.701192473186946e-05,
+      "loss": 0.5229,
+      "step": 2023
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6991683014731765e-05,
+      "loss": 0.539,
+      "step": 2024
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6971444449908474e-05,
+      "loss": 0.5201,
+      "step": 2025
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6951209058606886e-05,
+      "loss": 0.5236,
+      "step": 2026
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.693097686203102e-05,
+      "loss": 0.5298,
+      "step": 2027
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.691074788138149e-05,
+      "loss": 0.5236,
+      "step": 2028
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6890522137855568e-05,
+      "loss": 0.5072,
+      "step": 2029
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.687029965264715e-05,
+      "loss": 0.5388,
+      "step": 2030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.685008044694668e-05,
+      "loss": 0.5215,
+      "step": 2031
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6829864541941184e-05,
+      "loss": 0.5276,
+      "step": 2032
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6809651958814247e-05,
+      "loss": 0.5456,
+      "step": 2033
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6789442718745956e-05,
+      "loss": 0.5176,
+      "step": 2034
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.676923684291287e-05,
+      "loss": 0.5275,
+      "step": 2035
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6749034352488077e-05,
+      "loss": 0.5137,
+      "step": 2036
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6728835268641077e-05,
+      "loss": 0.5375,
+      "step": 2037
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.67086396125378e-05,
+      "loss": 0.5455,
+      "step": 2038
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6688447405340623e-05,
+      "loss": 0.5359,
+      "step": 2039
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6668258668208262e-05,
+      "loss": 0.5248,
+      "step": 2040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6648073422295815e-05,
+      "loss": 0.5427,
+      "step": 2041
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6627891688754744e-05,
+      "loss": 0.5395,
+      "step": 2042
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6607713488732792e-05,
+      "loss": 0.5293,
+      "step": 2043
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.658753884337402e-05,
+      "loss": 0.5536,
+      "step": 2044
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.656736777381877e-05,
+      "loss": 0.5186,
+      "step": 2045
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6547200301203625e-05,
+      "loss": 0.5257,
+      "step": 2046
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.5187,
+      "step": 2047
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6506876231321117e-05,
+      "loss": 0.5139,
+      "step": 2048
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6486719676308003e-05,
+      "loss": 0.5296,
+      "step": 2049
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.646656680274342e-05,
+      "loss": 0.5356,
+      "step": 2050
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.64464176317449e-05,
+      "loss": 0.5191,
+      "step": 2051
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.642627218442608e-05,
+      "loss": 0.5236,
+      "step": 2052
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6406130481896685e-05,
+      "loss": 0.53,
+      "step": 2053
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6385992545262556e-05,
+      "loss": 0.5188,
+      "step": 2054
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.636585839562554e-05,
+      "loss": 0.5309,
+      "step": 2055
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6345728054083533e-05,
+      "loss": 0.526,
+      "step": 2056
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6325601541730466e-05,
+      "loss": 0.5313,
+      "step": 2057
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.630547887965622e-05,
+      "loss": 0.5177,
+      "step": 2058
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6285360088946654e-05,
+      "loss": 0.5192,
+      "step": 2059
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6265245190683584e-05,
+      "loss": 0.5239,
+      "step": 2060
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.624513420594474e-05,
+      "loss": 0.5274,
+      "step": 2061
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.622502715580373e-05,
+      "loss": 0.5178,
+      "step": 2062
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.620492406133007e-05,
+      "loss": 0.5512,
+      "step": 2063
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.618482494358912e-05,
+      "loss": 0.5279,
+      "step": 2064
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6164729823642062e-05,
+      "loss": 0.523,
+      "step": 2065
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6144638722545906e-05,
+      "loss": 0.5381,
+      "step": 2066
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.612455166135344e-05,
+      "loss": 0.5324,
+      "step": 2067
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.610446866111321e-05,
+      "loss": 0.5268,
+      "step": 2068
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6084389742869543e-05,
+      "loss": 0.5285,
+      "step": 2069
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6064314927662433e-05,
+      "loss": 0.5274,
+      "step": 2070
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6044244236527633e-05,
+      "loss": 0.5254,
+      "step": 2071
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.602417769049653e-05,
+      "loss": 0.516,
+      "step": 2072
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6004115310596183e-05,
+      "loss": 0.5484,
+      "step": 2073
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.5984057117849304e-05,
+      "loss": 0.5333,
+      "step": 2074
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.5964003133274183e-05,
+      "loss": 0.4926,
+      "step": 2075
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.5943953377884718e-05,
+      "loss": 0.521,
+      "step": 2076
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5923907872690383e-05,
+      "loss": 0.5155,
+      "step": 2077
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.590386663869619e-05,
+      "loss": 0.5243,
+      "step": 2078
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5883829696902664e-05,
+      "loss": 0.5223,
+      "step": 2079
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.586379706830586e-05,
+      "loss": 0.5272,
+      "step": 2080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5843768773897285e-05,
+      "loss": 0.5429,
+      "step": 2081
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5823744834663918e-05,
+      "loss": 0.514,
+      "step": 2082
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5803725271588176e-05,
+      "loss": 0.5279,
+      "step": 2083
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.578371010564789e-05,
+      "loss": 0.5284,
+      "step": 2084
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.576369935781627e-05,
+      "loss": 0.5153,
+      "step": 2085
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5743693049061917e-05,
+      "loss": 0.5173,
+      "step": 2086
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5723691200348773e-05,
+      "loss": 0.5447,
+      "step": 2087
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5703693832636092e-05,
+      "loss": 0.5266,
+      "step": 2088
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5683700966878457e-05,
+      "loss": 0.5251,
+      "step": 2089
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.566371262402571e-05,
+      "loss": 0.5173,
+      "step": 2090
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.564372882502297e-05,
+      "loss": 0.5132,
+      "step": 2091
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.562374959081059e-05,
+      "loss": 0.519,
+      "step": 2092
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5603774942324144e-05,
+      "loss": 0.4994,
+      "step": 2093
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.558380490049438e-05,
+      "loss": 0.5141,
+      "step": 2094
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.556383948624725e-05,
+      "loss": 0.5076,
+      "step": 2095
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5543878720503835e-05,
+      "loss": 0.5232,
+      "step": 2096
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5523922624180346e-05,
+      "loss": 0.5195,
+      "step": 2097
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.550397121818812e-05,
+      "loss": 0.5202,
+      "step": 2098
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.548402452343356e-05,
+      "loss": 0.5262,
+      "step": 2099
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5464082560818133e-05,
+      "loss": 0.5353,
+      "step": 2100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.544414535123836e-05,
+      "loss": 0.5078,
+      "step": 2101
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5424212915585766e-05,
+      "loss": 0.5226,
+      "step": 2102
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5404285274746883e-05,
+      "loss": 0.5188,
+      "step": 2103
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5384362449603222e-05,
+      "loss": 0.5088,
+      "step": 2104
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5364444461031246e-05,
+      "loss": 0.5334,
+      "step": 2105
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.534453132990233e-05,
+      "loss": 0.5194,
+      "step": 2106
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5324623077082792e-05,
+      "loss": 0.5159,
+      "step": 2107
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.530471972343382e-05,
+      "loss": 0.5024,
+      "step": 2108
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5284821289811453e-05,
+      "loss": 0.4986,
+      "step": 2109
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.526492779706662e-05,
+      "loss": 0.5098,
+      "step": 2110
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5245039266045025e-05,
+      "loss": 0.529,
+      "step": 2111
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5225155717587192e-05,
+      "loss": 0.5025,
+      "step": 2112
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5205277172528438e-05,
+      "loss": 0.5354,
+      "step": 2113
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5185403651698812e-05,
+      "loss": 0.528,
+      "step": 2114
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5165535175923105e-05,
+      "loss": 0.5154,
+      "step": 2115
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5145671766020838e-05,
+      "loss": 0.5204,
+      "step": 2116
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.512581344280621e-05,
+      "loss": 0.5052,
+      "step": 2117
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.510596022708808e-05,
+      "loss": 0.5131,
+      "step": 2118
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5086112139669977e-05,
+      "loss": 0.529,
+      "step": 2119
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.506626920135005e-05,
+      "loss": 0.5254,
+      "step": 2120
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.504643143292103e-05,
+      "loss": 0.5278,
+      "step": 2121
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5026598855170265e-05,
+      "loss": 0.5288,
+      "step": 2122
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.5006771488879642e-05,
+      "loss": 0.5163,
+      "step": 2123
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.498694935482559e-05,
+      "loss": 0.5296,
+      "step": 2124
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4967132473779067e-05,
+      "loss": 0.5289,
+      "step": 2125
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4947320866505508e-05,
+      "loss": 0.5215,
+      "step": 2126
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.492751455376483e-05,
+      "loss": 0.538,
+      "step": 2127
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4907713556311414e-05,
+      "loss": 0.5328,
+      "step": 2128
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.488791789489406e-05,
+      "loss": 0.5219,
+      "step": 2129
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4868127590255961e-05,
+      "loss": 0.5179,
+      "step": 2130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.484834266313473e-05,
+      "loss": 0.521,
+      "step": 2131
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4828563134262326e-05,
+      "loss": 0.5125,
+      "step": 2132
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.480878902436504e-05,
+      "loss": 0.5296,
+      "step": 2133
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4789020354163515e-05,
+      "loss": 0.5131,
+      "step": 2134
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4769257144372668e-05,
+      "loss": 0.4924,
+      "step": 2135
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.47494994157017e-05,
+      "loss": 0.529,
+      "step": 2136
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4729747188854089e-05,
+      "loss": 0.5252,
+      "step": 2137
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4710000484527511e-05,
+      "loss": 0.5161,
+      "step": 2138
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.469025932341388e-05,
+      "loss": 0.5229,
+      "step": 2139
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4670523726199304e-05,
+      "loss": 0.5187,
+      "step": 2140
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4650793713564046e-05,
+      "loss": 0.4965,
+      "step": 2141
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4631069306182518e-05,
+      "loss": 0.4991,
+      "step": 2142
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.461135052472328e-05,
+      "loss": 0.5479,
+      "step": 2143
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4591637389848976e-05,
+      "loss": 0.5287,
+      "step": 2144
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4571929922216326e-05,
+      "loss": 0.5177,
+      "step": 2145
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4552228142476138e-05,
+      "loss": 0.5045,
+      "step": 2146
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4532532071273245e-05,
+      "loss": 0.5294,
+      "step": 2147
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4512841729246483e-05,
+      "loss": 0.5292,
+      "step": 2148
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4493157137028729e-05,
+      "loss": 0.5055,
+      "step": 2149
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.447347831524678e-05,
+      "loss": 0.5368,
+      "step": 2150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4453805284521422e-05,
+      "loss": 0.522,
+      "step": 2151
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4434138065467373e-05,
+      "loss": 0.5022,
+      "step": 2152
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4414476678693242e-05,
+      "loss": 0.5166,
+      "step": 2153
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.439482114480153e-05,
+      "loss": 0.5278,
+      "step": 2154
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4375171484388623e-05,
+      "loss": 0.5301,
+      "step": 2155
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4355527718044746e-05,
+      "loss": 0.5102,
+      "step": 2156
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.433588986635392e-05,
+      "loss": 0.5031,
+      "step": 2157
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.431625794989401e-05,
+      "loss": 0.5195,
+      "step": 2158
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4296631989236643e-05,
+      "loss": 0.5181,
+      "step": 2159
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.427701200494718e-05,
+      "loss": 0.506,
+      "step": 2160
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425739801758478e-05,
+      "loss": 0.51,
+      "step": 2161
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4237790047702258e-05,
+      "loss": 0.5254,
+      "step": 2162
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4218188115846148e-05,
+      "loss": 0.5183,
+      "step": 2163
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4198592242556673e-05,
+      "loss": 0.5328,
+      "step": 2164
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4179002448367673e-05,
+      "loss": 0.5224,
+      "step": 2165
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4159418753806642e-05,
+      "loss": 0.5109,
+      "step": 2166
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4139841179394678e-05,
+      "loss": 0.5039,
+      "step": 2167
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4120269745646469e-05,
+      "loss": 0.5351,
+      "step": 2168
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.410070447307025e-05,
+      "loss": 0.5048,
+      "step": 2169
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4081145382167825e-05,
+      "loss": 0.5222,
+      "step": 2170
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4061592493434511e-05,
+      "loss": 0.5393,
+      "step": 2171
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4042045827359108e-05,
+      "loss": 0.52,
+      "step": 2172
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4022505404423924e-05,
+      "loss": 0.5008,
+      "step": 2173
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4002971245104713e-05,
+      "loss": 0.5024,
+      "step": 2174
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3983443369870657e-05,
+      "loss": 0.533,
+      "step": 2175
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3963921799184372e-05,
+      "loss": 0.497,
+      "step": 2176
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.394440655350185e-05,
+      "loss": 0.5189,
+      "step": 2177
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3924897653272456e-05,
+      "loss": 0.5284,
+      "step": 2178
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3905395118938929e-05,
+      "loss": 0.5331,
+      "step": 2179
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3885898970937304e-05,
+      "loss": 0.5373,
+      "step": 2180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3866409229696942e-05,
+      "loss": 0.5472,
+      "step": 2181
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3846925915640496e-05,
+      "loss": 0.4985,
+      "step": 2182
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.382744904918388e-05,
+      "loss": 0.533,
+      "step": 2183
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3807978650736228e-05,
+      "loss": 0.5073,
+      "step": 2184
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3788514740699941e-05,
+      "loss": 0.5195,
+      "step": 2185
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.376905733947059e-05,
+      "loss": 0.5204,
+      "step": 2186
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3749606467436919e-05,
+      "loss": 0.5319,
+      "step": 2187
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3730162144980863e-05,
+      "loss": 0.521,
+      "step": 2188
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3710724392477466e-05,
+      "loss": 0.5128,
+      "step": 2189
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.369129323029489e-05,
+      "loss": 0.528,
+      "step": 2190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3671868678794416e-05,
+      "loss": 0.5192,
+      "step": 2191
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3652450758330364e-05,
+      "loss": 0.529,
+      "step": 2192
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3633039489250121e-05,
+      "loss": 0.5064,
+      "step": 2193
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3613634891894117e-05,
+      "loss": 0.5299,
+      "step": 2194
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.359423698659577e-05,
+      "loss": 0.512,
+      "step": 2195
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3574845793681488e-05,
+      "loss": 0.5314,
+      "step": 2196
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.355546133347066e-05,
+      "loss": 0.5251,
+      "step": 2197
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3536083626275606e-05,
+      "loss": 0.5244,
+      "step": 2198
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3516712692401584e-05,
+      "loss": 0.5368,
+      "step": 2199
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3497348552146741e-05,
+      "loss": 0.5044,
+      "step": 2200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3477991225802103e-05,
+      "loss": 0.5245,
+      "step": 2201
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.345864073365157e-05,
+      "loss": 0.5023,
+      "step": 2202
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3439297095971872e-05,
+      "loss": 0.5187,
+      "step": 2203
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3419960333032551e-05,
+      "loss": 0.53,
+      "step": 2204
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3400630465095961e-05,
+      "loss": 0.533,
+      "step": 2205
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.338130751241722e-05,
+      "loss": 0.5185,
+      "step": 2206
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3361991495244195e-05,
+      "loss": 0.5097,
+      "step": 2207
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3342682433817507e-05,
+      "loss": 0.5091,
+      "step": 2208
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3323380348370459e-05,
+      "loss": 0.5218,
+      "step": 2209
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3304085259129057e-05,
+      "loss": 0.5092,
+      "step": 2210
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3284797186311994e-05,
+      "loss": 0.5198,
+      "step": 2211
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3265516150130577e-05,
+      "loss": 0.5272,
+      "step": 2212
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.324624217078876e-05,
+      "loss": 0.5317,
+      "step": 2213
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3226975268483107e-05,
+      "loss": 0.5287,
+      "step": 2214
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3207715463402759e-05,
+      "loss": 0.5356,
+      "step": 2215
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3188462775729404e-05,
+      "loss": 0.5238,
+      "step": 2216
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3169217225637301e-05,
+      "loss": 0.5189,
+      "step": 2217
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.314997883329322e-05,
+      "loss": 0.526,
+      "step": 2218
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3130747618856403e-05,
+      "loss": 0.5338,
+      "step": 2219
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3111523602478626e-05,
+      "loss": 0.515,
+      "step": 2220
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3092306804304067e-05,
+      "loss": 0.5143,
+      "step": 2221
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.307309724446937e-05,
+      "loss": 0.5109,
+      "step": 2222
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3053894943103598e-05,
+      "loss": 0.5412,
+      "step": 2223
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3034699920328185e-05,
+      "loss": 0.5334,
+      "step": 2224
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3015512196256952e-05,
+      "loss": 0.5313,
+      "step": 2225
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2996331790996078e-05,
+      "loss": 0.5341,
+      "step": 2226
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2977158724644068e-05,
+      "loss": 0.4958,
+      "step": 2227
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2957993017291721e-05,
+      "loss": 0.5101,
+      "step": 2228
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2938834689022153e-05,
+      "loss": 0.5211,
+      "step": 2229
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2919683759910731e-05,
+      "loss": 0.5341,
+      "step": 2230
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2900540250025053e-05,
+      "loss": 0.5098,
+      "step": 2231
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2881404179424988e-05,
+      "loss": 0.5006,
+      "step": 2232
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2862275568162566e-05,
+      "loss": 0.5163,
+      "step": 2233
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2843154436282014e-05,
+      "loss": 0.5218,
+      "step": 2234
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.282404080381974e-05,
+      "loss": 0.525,
+      "step": 2235
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2804934690804262e-05,
+      "loss": 0.5263,
+      "step": 2236
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2785836117256239e-05,
+      "loss": 0.518,
+      "step": 2237
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2766745103188427e-05,
+      "loss": 0.5433,
+      "step": 2238
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2747661668605667e-05,
+      "loss": 0.5152,
+      "step": 2239
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.272858583350483e-05,
+      "loss": 0.5069,
+      "step": 2240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.270951761787486e-05,
+      "loss": 0.5175,
+      "step": 2241
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.26904570416967e-05,
+      "loss": 0.5001,
+      "step": 2242
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2671404124943272e-05,
+      "loss": 0.5162,
+      "step": 2243
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2652358887579506e-05,
+      "loss": 0.5148,
+      "step": 2244
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.263332134956226e-05,
+      "loss": 0.4943,
+      "step": 2245
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2614291530840323e-05,
+      "loss": 0.4969,
+      "step": 2246
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2595269451354426e-05,
+      "loss": 0.5188,
+      "step": 2247
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2576255131037144e-05,
+      "loss": 0.5317,
+      "step": 2248
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2557248589812953e-05,
+      "loss": 0.4893,
+      "step": 2249
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2538249847598178e-05,
+      "loss": 0.5245,
+      "step": 2250
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2519258924300952e-05,
+      "loss": 0.5265,
+      "step": 2251
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2500275839821227e-05,
+      "loss": 0.522,
+      "step": 2252
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2481300614050747e-05,
+      "loss": 0.5209,
+      "step": 2253
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2462333266873015e-05,
+      "loss": 0.5328,
+      "step": 2254
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2443373818163267e-05,
+      "loss": 0.4976,
+      "step": 2255
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.242442228778848e-05,
+      "loss": 0.5243,
+      "step": 2256
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2405478695607334e-05,
+      "loss": 0.5209,
+      "step": 2257
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2386543061470162e-05,
+      "loss": 0.5301,
+      "step": 2258
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2367615405219003e-05,
+      "loss": 0.5275,
+      "step": 2259
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2348695746687497e-05,
+      "loss": 0.5323,
+      "step": 2260
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2329784105700926e-05,
+      "loss": 0.5007,
+      "step": 2261
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2310880502076169e-05,
+      "loss": 0.504,
+      "step": 2262
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2291984955621668e-05,
+      "loss": 0.5226,
+      "step": 2263
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2273097486137426e-05,
+      "loss": 0.5168,
+      "step": 2264
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.225421811341501e-05,
+      "loss": 0.507,
+      "step": 2265
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2235346857237467e-05,
+      "loss": 0.5255,
+      "step": 2266
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.221648373737935e-05,
+      "loss": 0.5153,
+      "step": 2267
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2197628773606695e-05,
+      "loss": 0.5175,
+      "step": 2268
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2178781985676991e-05,
+      "loss": 0.5058,
+      "step": 2269
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2159943393339135e-05,
+      "loss": 0.5192,
+      "step": 2270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.214111301633348e-05,
+      "loss": 0.5019,
+      "step": 2271
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2122290874391735e-05,
+      "loss": 0.5035,
+      "step": 2272
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2103476987236986e-05,
+      "loss": 0.4969,
+      "step": 2273
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2084671374583686e-05,
+      "loss": 0.5256,
+      "step": 2274
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2065874056137594e-05,
+      "loss": 0.5244,
+      "step": 2275
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2047085051595788e-05,
+      "loss": 0.4636,
+      "step": 2276
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2028304380646648e-05,
+      "loss": 0.4278,
+      "step": 2277
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2009532062969801e-05,
+      "loss": 0.4426,
+      "step": 2278
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1990768118236123e-05,
+      "loss": 0.4428,
+      "step": 2279
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.197201256610773e-05,
+      "loss": 0.4375,
+      "step": 2280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1953265426237942e-05,
+      "loss": 0.4448,
+      "step": 2281
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1934526718271236e-05,
+      "loss": 0.4346,
+      "step": 2282
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1915796461843296e-05,
+      "loss": 0.4454,
+      "step": 2283
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.189707467658092e-05,
+      "loss": 0.4372,
+      "step": 2284
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1878361382102037e-05,
+      "loss": 0.4482,
+      "step": 2285
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.185965659801569e-05,
+      "loss": 0.4349,
+      "step": 2286
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1840960343921983e-05,
+      "loss": 0.4368,
+      "step": 2287
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1822272639412096e-05,
+      "loss": 0.4411,
+      "step": 2288
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1803593504068256e-05,
+      "loss": 0.4303,
+      "step": 2289
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1784922957463688e-05,
+      "loss": 0.4485,
+      "step": 2290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1766261019162638e-05,
+      "loss": 0.4462,
+      "step": 2291
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.174760770872033e-05,
+      "loss": 0.4271,
+      "step": 2292
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1728963045682938e-05,
+      "loss": 0.4404,
+      "step": 2293
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1710327049587571e-05,
+      "loss": 0.4597,
+      "step": 2294
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.4431,
+      "step": 2295
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1673081136325982e-05,
+      "loss": 0.4314,
+      "step": 2296
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1654471258188484e-05,
+      "loss": 0.4448,
+      "step": 2297
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1635870125050472e-05,
+      "loss": 0.4208,
+      "step": 2298
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1617277756403446e-05,
+      "loss": 0.4275,
+      "step": 2299
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1598694171729703e-05,
+      "loss": 0.4519,
+      "step": 2300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.158011939050238e-05,
+      "loss": 0.4389,
+      "step": 2301
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1561553432185369e-05,
+      "loss": 0.4476,
+      "step": 2302
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1542996316233286e-05,
+      "loss": 0.4419,
+      "step": 2303
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1524448062091537e-05,
+      "loss": 0.4464,
+      "step": 2304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.150590868919621e-05,
+      "loss": 0.4338,
+      "step": 2305
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1487378216974087e-05,
+      "loss": 0.431,
+      "step": 2306
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.146885666484263e-05,
+      "loss": 0.4267,
+      "step": 2307
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1450344052209952e-05,
+      "loss": 0.4311,
+      "step": 2308
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1431840398474799e-05,
+      "loss": 0.4342,
+      "step": 2309
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1413345723026525e-05,
+      "loss": 0.4286,
+      "step": 2310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1394860045245084e-05,
+      "loss": 0.4321,
+      "step": 2311
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1376383384500994e-05,
+      "loss": 0.4299,
+      "step": 2312
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1357915760155332e-05,
+      "loss": 0.401,
+      "step": 2313
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1339457191559697e-05,
+      "loss": 0.4383,
+      "step": 2314
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.13210076980562e-05,
+      "loss": 0.4391,
+      "step": 2315
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1302567298977466e-05,
+      "loss": 0.4435,
+      "step": 2316
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1284136013646547e-05,
+      "loss": 0.4303,
+      "step": 2317
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1265713861376973e-05,
+      "loss": 0.4274,
+      "step": 2318
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1247300861472716e-05,
+      "loss": 0.4083,
+      "step": 2319
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1228897033228119e-05,
+      "loss": 0.4305,
+      "step": 2320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.121050239592794e-05,
+      "loss": 0.4255,
+      "step": 2321
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1192116968847313e-05,
+      "loss": 0.4347,
+      "step": 2322
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1173740771251714e-05,
+      "loss": 0.4321,
+      "step": 2323
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1155373822396915e-05,
+      "loss": 0.4531,
+      "step": 2324
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1137016141529054e-05,
+      "loss": 0.417,
+      "step": 2325
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1118667747884517e-05,
+      "loss": 0.4269,
+      "step": 2326
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1100328660689974e-05,
+      "loss": 0.4354,
+      "step": 2327
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1081998899162336e-05,
+      "loss": 0.4354,
+      "step": 2328
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1063678482508744e-05,
+      "loss": 0.4234,
+      "step": 2329
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.104536742992655e-05,
+      "loss": 0.4398,
+      "step": 2330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.102706576060329e-05,
+      "loss": 0.4321,
+      "step": 2331
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1008773493716668e-05,
+      "loss": 0.4596,
+      "step": 2332
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0990490648434541e-05,
+      "loss": 0.4287,
+      "step": 2333
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0972217243914888e-05,
+      "loss": 0.4336,
+      "step": 2334
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0953953299305784e-05,
+      "loss": 0.4365,
+      "step": 2335
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.093569883374543e-05,
+      "loss": 0.429,
+      "step": 2336
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0917453866362044e-05,
+      "loss": 0.4399,
+      "step": 2337
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0899218416273913e-05,
+      "loss": 0.4263,
+      "step": 2338
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0880992502589382e-05,
+      "loss": 0.4357,
+      "step": 2339
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0862776144406745e-05,
+      "loss": 0.4429,
+      "step": 2340
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0844569360814317e-05,
+      "loss": 0.4372,
+      "step": 2341
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0826372170890386e-05,
+      "loss": 0.4313,
+      "step": 2342
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0808184593703181e-05,
+      "loss": 0.4363,
+      "step": 2343
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0790006648310828e-05,
+      "loss": 0.431,
+      "step": 2344
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0771838353761413e-05,
+      "loss": 0.4421,
+      "step": 2345
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0753679729092877e-05,
+      "loss": 0.4117,
+      "step": 2346
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0735530793333011e-05,
+      "loss": 0.4486,
+      "step": 2347
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0717391565499503e-05,
+      "loss": 0.4191,
+      "step": 2348
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0699262064599827e-05,
+      "loss": 0.4278,
+      "step": 2349
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0681142309631283e-05,
+      "loss": 0.4469,
+      "step": 2350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0663032319580956e-05,
+      "loss": 0.4447,
+      "step": 2351
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.064493211342569e-05,
+      "loss": 0.4336,
+      "step": 2352
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.062684171013209e-05,
+      "loss": 0.4312,
+      "step": 2353
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0608761128656477e-05,
+      "loss": 0.4335,
+      "step": 2354
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.059069038794489e-05,
+      "loss": 0.4399,
+      "step": 2355
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0572629506933053e-05,
+      "loss": 0.4255,
+      "step": 2356
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.055457850454635e-05,
+      "loss": 0.4382,
+      "step": 2357
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0536537399699824e-05,
+      "loss": 0.4326,
+      "step": 2358
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0518506211298148e-05,
+      "loss": 0.4307,
+      "step": 2359
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.050048495823559e-05,
+      "loss": 0.44,
+      "step": 2360
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0482473659396026e-05,
+      "loss": 0.4246,
+      "step": 2361
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0464472333652878e-05,
+      "loss": 0.4221,
+      "step": 2362
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0446480999869152e-05,
+      "loss": 0.4312,
+      "step": 2363
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0428499676897346e-05,
+      "loss": 0.4309,
+      "step": 2364
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0410528383579481e-05,
+      "loss": 0.4344,
+      "step": 2365
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0392567138747101e-05,
+      "loss": 0.429,
+      "step": 2366
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0374615961221157e-05,
+      "loss": 0.4297,
+      "step": 2367
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0356674869812095e-05,
+      "loss": 0.4396,
+      "step": 2368
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0338743883319795e-05,
+      "loss": 0.4447,
+      "step": 2369
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0320823020533536e-05,
+      "loss": 0.4359,
+      "step": 2370
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0302912300231967e-05,
+      "loss": 0.433,
+      "step": 2371
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0285011741183152e-05,
+      "loss": 0.4172,
+      "step": 2372
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0267121362144486e-05,
+      "loss": 0.4212,
+      "step": 2373
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0249241181862675e-05,
+      "loss": 0.4297,
+      "step": 2374
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0231371219073778e-05,
+      "loss": 0.4248,
+      "step": 2375
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.021351149250313e-05,
+      "loss": 0.4337,
+      "step": 2376
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0195662020865333e-05,
+      "loss": 0.4221,
+      "step": 2377
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0177822822864252e-05,
+      "loss": 0.4267,
+      "step": 2378
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0159993917192983e-05,
+      "loss": 0.4282,
+      "step": 2379
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0142175322533837e-05,
+      "loss": 0.4215,
+      "step": 2380
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0124367057558322e-05,
+      "loss": 0.4262,
+      "step": 2381
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0106569140927118e-05,
+      "loss": 0.4392,
+      "step": 2382
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.008878159129007e-05,
+      "loss": 0.4512,
+      "step": 2383
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0071004427286155e-05,
+      "loss": 0.4353,
+      "step": 2384
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.005323766754346e-05,
+      "loss": 0.4291,
+      "step": 2385
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0035481330679176e-05,
+      "loss": 0.4376,
+      "step": 2386
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0017735435299588e-05,
+      "loss": 0.4331,
+      "step": 2387
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.4435,
+      "step": 2388
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.982275043364795e-06,
+      "loss": 0.4342,
+      "step": 2389
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.964560583967367e-06,
+      "loss": 0.4212,
+      "step": 2390
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.946856640370085e-06,
+      "loss": 0.4285,
+      "step": 2391
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.929163231124319e-06,
+      "loss": 0.4268,
+      "step": 2392
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.911480374770412e-06,
+      "loss": 0.4328,
+      "step": 2393
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.893808089837638e-06,
+      "loss": 0.4375,
+      "step": 2394
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.876146394844169e-06,
+      "loss": 0.4336,
+      "step": 2395
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.85849530829712e-06,
+      "loss": 0.4393,
+      "step": 2396
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.840854848692478e-06,
+      "loss": 0.4295,
+      "step": 2397
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.823225034515062e-06,
+      "loss": 0.4309,
+      "step": 2398
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.805605884238587e-06,
+      "loss": 0.4297,
+      "step": 2399
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.78799741632556e-06,
+      "loss": 0.4315,
+      "step": 2400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.77039964922731e-06,
+      "loss": 0.4271,
+      "step": 2401
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.752812601383937e-06,
+      "loss": 0.4429,
+      "step": 2402
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.735236291224323e-06,
+      "loss": 0.4356,
+      "step": 2403
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.717670737166094e-06,
+      "loss": 0.4504,
+      "step": 2404
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.700115957615606e-06,
+      "loss": 0.4496,
+      "step": 2405
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.682571970967918e-06,
+      "loss": 0.4351,
+      "step": 2406
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.665038795606789e-06,
+      "loss": 0.4554,
+      "step": 2407
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.647516449904644e-06,
+      "loss": 0.4237,
+      "step": 2408
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.63000495222256e-06,
+      "loss": 0.4424,
+      "step": 2409
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.612504320910249e-06,
+      "loss": 0.4308,
+      "step": 2410
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.595014574306037e-06,
+      "loss": 0.4393,
+      "step": 2411
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.577535730736838e-06,
+      "loss": 0.447,
+      "step": 2412
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.56006780851814e-06,
+      "loss": 0.4234,
+      "step": 2413
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.542610825954015e-06,
+      "loss": 0.4209,
+      "step": 2414
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.525164801337027e-06,
+      "loss": 0.4353,
+      "step": 2415
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.507729752948283e-06,
+      "loss": 0.4341,
+      "step": 2416
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.490305699057401e-06,
+      "loss": 0.4209,
+      "step": 2417
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.472892657922444e-06,
+      "loss": 0.4328,
+      "step": 2418
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.455490647789951e-06,
+      "loss": 0.4268,
+      "step": 2419
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.438099686894922e-06,
+      "loss": 0.4366,
+      "step": 2420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.420719793460758e-06,
+      "loss": 0.4229,
+      "step": 2421
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.403350985699246e-06,
+      "loss": 0.4324,
+      "step": 2422
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.3859932818106e-06,
+      "loss": 0.4284,
+      "step": 2423
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.368646699983372e-06,
+      "loss": 0.4318,
+      "step": 2424
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.351311258394444e-06,
+      "loss": 0.4254,
+      "step": 2425
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.333986975209064e-06,
+      "loss": 0.4265,
+      "step": 2426
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.316673868580762e-06,
+      "loss": 0.4316,
+      "step": 2427
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.299371956651355e-06,
+      "loss": 0.4283,
+      "step": 2428
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.28208125755094e-06,
+      "loss": 0.4424,
+      "step": 2429
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.264801789397863e-06,
+      "loss": 0.4307,
+      "step": 2430
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.247533570298694e-06,
+      "loss": 0.4334,
+      "step": 2431
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.230276618348224e-06,
+      "loss": 0.444,
+      "step": 2432
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.213030951629431e-06,
+      "loss": 0.4325,
+      "step": 2433
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.195796588213472e-06,
+      "loss": 0.4401,
+      "step": 2434
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.178573546159656e-06,
+      "loss": 0.4282,
+      "step": 2435
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.161361843515433e-06,
+      "loss": 0.4378,
+      "step": 2436
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.144161498316365e-06,
+      "loss": 0.4229,
+      "step": 2437
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.126972528586124e-06,
+      "loss": 0.4227,
+      "step": 2438
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.109794952336446e-06,
+      "loss": 0.4211,
+      "step": 2439
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.092628787567135e-06,
+      "loss": 0.4346,
+      "step": 2440
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.075474052266055e-06,
+      "loss": 0.4423,
+      "step": 2441
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.058330764409062e-06,
+      "loss": 0.4166,
+      "step": 2442
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.04119894196003e-06,
+      "loss": 0.4342,
+      "step": 2443
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.02407860287084e-06,
+      "loss": 0.4319,
+      "step": 2444
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.006969765081302e-06,
+      "loss": 0.4343,
+      "step": 2445
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.98987244651919e-06,
+      "loss": 0.4298,
+      "step": 2446
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.972786665100228e-06,
+      "loss": 0.446,
+      "step": 2447
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.95571243872803e-06,
+      "loss": 0.4187,
+      "step": 2448
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.938649785294085e-06,
+      "loss": 0.4243,
+      "step": 2449
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.921598722677796e-06,
+      "loss": 0.4398,
+      "step": 2450
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.90455926874639e-06,
+      "loss": 0.4136,
+      "step": 2451
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.887531441354944e-06,
+      "loss": 0.4266,
+      "step": 2452
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.870515258346342e-06,
+      "loss": 0.4444,
+      "step": 2453
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.853510737551274e-06,
+      "loss": 0.423,
+      "step": 2454
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.836517896788204e-06,
+      "loss": 0.4399,
+      "step": 2455
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.819536753863363e-06,
+      "loss": 0.4257,
+      "step": 2456
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.80256732657071e-06,
+      "loss": 0.4334,
+      "step": 2457
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.78560963269196e-06,
+      "loss": 0.4338,
+      "step": 2458
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.768663689996493e-06,
+      "loss": 0.4509,
+      "step": 2459
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.751729516241391e-06,
+      "loss": 0.4305,
+      "step": 2460
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.734807129171429e-06,
+      "loss": 0.4266,
+      "step": 2461
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.717896546518987e-06,
+      "loss": 0.4265,
+      "step": 2462
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.700997786004097e-06,
+      "loss": 0.439,
+      "step": 2463
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.684110865334417e-06,
+      "loss": 0.4372,
+      "step": 2464
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.667235802205183e-06,
+      "loss": 0.4333,
+      "step": 2465
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.650372614299186e-06,
+      "loss": 0.428,
+      "step": 2466
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.63352131928682e-06,
+      "loss": 0.4298,
+      "step": 2467
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.616681934825986e-06,
+      "loss": 0.4334,
+      "step": 2468
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.59985447856209e-06,
+      "loss": 0.433,
+      "step": 2469
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.583038968128085e-06,
+      "loss": 0.4251,
+      "step": 2470
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.566235421144368e-06,
+      "loss": 0.4467,
+      "step": 2471
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.549443855218813e-06,
+      "loss": 0.4257,
+      "step": 2472
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.532664287946737e-06,
+      "loss": 0.4333,
+      "step": 2473
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.515896736910887e-06,
+      "loss": 0.4369,
+      "step": 2474
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.499141219681417e-06,
+      "loss": 0.4303,
+      "step": 2475
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.482397753815872e-06,
+      "loss": 0.416,
+      "step": 2476
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.465666356859166e-06,
+      "loss": 0.4418,
+      "step": 2477
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.448947046343568e-06,
+      "loss": 0.4285,
+      "step": 2478
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.432239839788685e-06,
+      "loss": 0.4347,
+      "step": 2479
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.415544754701434e-06,
+      "loss": 0.4277,
+      "step": 2480
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.39886180857604e-06,
+      "loss": 0.4365,
+      "step": 2481
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.382191018893998e-06,
+      "loss": 0.4233,
+      "step": 2482
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.365532403124068e-06,
+      "loss": 0.4138,
+      "step": 2483
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.348885978722252e-06,
+      "loss": 0.4252,
+      "step": 2484
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.332251763131799e-06,
+      "loss": 0.4277,
+      "step": 2485
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.31562977378312e-06,
+      "loss": 0.4265,
+      "step": 2486
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.299020028093844e-06,
+      "loss": 0.414,
+      "step": 2487
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.28242254346879e-06,
+      "loss": 0.425,
+      "step": 2488
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.265837337299874e-06,
+      "loss": 0.4094,
+      "step": 2489
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.249264426966185e-06,
+      "loss": 0.4166,
+      "step": 2490
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.232703829833935e-06,
+      "loss": 0.4242,
+      "step": 2491
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.216155563256411e-06,
+      "loss": 0.4284,
+      "step": 2492
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.199619644573969e-06,
+      "loss": 0.4268,
+      "step": 2493
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.183096091114064e-06,
+      "loss": 0.4316,
+      "step": 2494
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.166584920191175e-06,
+      "loss": 0.4252,
+      "step": 2495
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.150086149106778e-06,
+      "loss": 0.4387,
+      "step": 2496
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.133599795149408e-06,
+      "loss": 0.4368,
+      "step": 2497
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.11712587559455e-06,
+      "loss": 0.425,
+      "step": 2498
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.100664407704673e-06,
+      "loss": 0.4314,
+      "step": 2499
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.084215408729198e-06,
+      "loss": 0.4446,
+      "step": 2500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.06777889590448e-06,
+      "loss": 0.4445,
+      "step": 2501
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.051354886453787e-06,
+      "loss": 0.4361,
+      "step": 2502
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.034943397587287e-06,
+      "loss": 0.4227,
+      "step": 2503
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.01854444650203e-06,
+      "loss": 0.434,
+      "step": 2504
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.002158050381926e-06,
+      "loss": 0.4168,
+      "step": 2505
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.985784226397728e-06,
+      "loss": 0.4118,
+      "step": 2506
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.969422991707019e-06,
+      "loss": 0.4326,
+      "step": 2507
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.953074363454185e-06,
+      "loss": 0.4328,
+      "step": 2508
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.936738358770409e-06,
+      "loss": 0.447,
+      "step": 2509
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.920414994773638e-06,
+      "loss": 0.4202,
+      "step": 2510
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.90410428856857e-06,
+      "loss": 0.4254,
+      "step": 2511
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.887806257246671e-06,
+      "loss": 0.4226,
+      "step": 2512
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.87152091788608e-06,
+      "loss": 0.4331,
+      "step": 2513
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.855248287551653e-06,
+      "loss": 0.4338,
+      "step": 2514
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.83898838329495e-06,
+      "loss": 0.4316,
+      "step": 2515
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.82274122215418e-06,
+      "loss": 0.4194,
+      "step": 2516
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.806506821154177e-06,
+      "loss": 0.4226,
+      "step": 2517
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.790285197306445e-06,
+      "loss": 0.4455,
+      "step": 2518
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.774076367609075e-06,
+      "loss": 0.4392,
+      "step": 2519
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.757880349046742e-06,
+      "loss": 0.4385,
+      "step": 2520
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.741697158590728e-06,
+      "loss": 0.4296,
+      "step": 2521
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.72552681319885e-06,
+      "loss": 0.4272,
+      "step": 2522
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.709369329815464e-06,
+      "loss": 0.4322,
+      "step": 2523
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.693224725371462e-06,
+      "loss": 0.444,
+      "step": 2524
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.677093016784228e-06,
+      "loss": 0.419,
+      "step": 2525
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.660974220957642e-06,
+      "loss": 0.4225,
+      "step": 2526
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.644868354782045e-06,
+      "loss": 0.4238,
+      "step": 2527
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.6287754351342366e-06,
+      "loss": 0.4441,
+      "step": 2528
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.612695478877447e-06,
+      "loss": 0.4216,
+      "step": 2529
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.596628502861323e-06,
+      "loss": 0.4305,
+      "step": 2530
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.580574523921906e-06,
+      "loss": 0.4275,
+      "step": 2531
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.5645335588816284e-06,
+      "loss": 0.427,
+      "step": 2532
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.548505624549272e-06,
+      "loss": 0.4273,
+      "step": 2533
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.5324907377199755e-06,
+      "loss": 0.4084,
+      "step": 2534
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.516488915175193e-06,
+      "loss": 0.4479,
+      "step": 2535
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.500500173682716e-06,
+      "loss": 0.4299,
+      "step": 2536
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.484524529996595e-06,
+      "loss": 0.4338,
+      "step": 2537
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.468562000857167e-06,
+      "loss": 0.4471,
+      "step": 2538
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.452612602991049e-06,
+      "loss": 0.4338,
+      "step": 2539
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.436676353111065e-06,
+      "loss": 0.422,
+      "step": 2540
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.420753267916272e-06,
+      "loss": 0.4439,
+      "step": 2541
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.404843364091951e-06,
+      "loss": 0.4178,
+      "step": 2542
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.388946658309557e-06,
+      "loss": 0.4373,
+      "step": 2543
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.373063167226693e-06,
+      "loss": 0.4451,
+      "step": 2544
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.3571929074871586e-06,
+      "loss": 0.4148,
+      "step": 2545
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.341335895720865e-06,
+      "loss": 0.4429,
+      "step": 2546
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.32549214854382e-06,
+      "loss": 0.4209,
+      "step": 2547
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.30966168255818e-06,
+      "loss": 0.4302,
+      "step": 2548
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.293844514352149e-06,
+      "loss": 0.4354,
+      "step": 2549
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.278040660500008e-06,
+      "loss": 0.421,
+      "step": 2550
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.2622501375620815e-06,
+      "loss": 0.4152,
+      "step": 2551
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.246472962084732e-06,
+      "loss": 0.4437,
+      "step": 2552
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.2307091506003325e-06,
+      "loss": 0.4255,
+      "step": 2553
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.214958719627249e-06,
+      "loss": 0.4301,
+      "step": 2554
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.199221685669835e-06,
+      "loss": 0.4301,
+      "step": 2555
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.183498065218395e-06,
+      "loss": 0.4361,
+      "step": 2556
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.167787874749184e-06,
+      "loss": 0.4393,
+      "step": 2557
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.152091130724384e-06,
+      "loss": 0.446,
+      "step": 2558
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.136407849592086e-06,
+      "loss": 0.4308,
+      "step": 2559
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.120738047786275e-06,
+      "loss": 0.4166,
+      "step": 2560
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.105081741726809e-06,
+      "loss": 0.4336,
+      "step": 2561
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.0894389478194e-06,
+      "loss": 0.4272,
+      "step": 2562
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.073809682455628e-06,
+      "loss": 0.4178,
+      "step": 2563
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.0581939620128515e-06,
+      "loss": 0.4331,
+      "step": 2564
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.042591802854264e-06,
+      "loss": 0.4439,
+      "step": 2565
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.0270032213288635e-06,
+      "loss": 0.4455,
+      "step": 2566
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.01142823377138e-06,
+      "loss": 0.4109,
+      "step": 2567
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.9958668565023205e-06,
+      "loss": 0.431,
+      "step": 2568
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.980319105827944e-06,
+      "loss": 0.4205,
+      "step": 2569
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.964784998040215e-06,
+      "loss": 0.4286,
+      "step": 2570
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.949264549416787e-06,
+      "loss": 0.4192,
+      "step": 2571
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.933757776221035e-06,
+      "loss": 0.4365,
+      "step": 2572
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.918264694701986e-06,
+      "loss": 0.4265,
+      "step": 2573
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.4396,
+      "step": 2574
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.887319671618315e-06,
+      "loss": 0.4218,
+      "step": 2575
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.871867762479958e-06,
+      "loss": 0.4413,
+      "step": 2576
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.856429609870765e-06,
+      "loss": 0.4225,
+      "step": 2577
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.841005229967858e-06,
+      "loss": 0.4254,
+      "step": 2578
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.82559463893393e-06,
+      "loss": 0.4199,
+      "step": 2579
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.81019785291722e-06,
+      "loss": 0.4224,
+      "step": 2580
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.794814888051504e-06,
+      "loss": 0.4411,
+      "step": 2581
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.779445760456067e-06,
+      "loss": 0.4239,
+      "step": 2582
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.764090486235721e-06,
+      "loss": 0.4385,
+      "step": 2583
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.748749081480726e-06,
+      "loss": 0.4169,
+      "step": 2584
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.733421562266822e-06,
+      "loss": 0.4181,
+      "step": 2585
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.7181079446552165e-06,
+      "loss": 0.4181,
+      "step": 2586
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.702808244692536e-06,
+      "loss": 0.4229,
+      "step": 2587
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.687522478410802e-06,
+      "loss": 0.4265,
+      "step": 2588
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.672250661827475e-06,
+      "loss": 0.4262,
+      "step": 2589
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.656992810945379e-06,
+      "loss": 0.4334,
+      "step": 2590
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.641748941752688e-06,
+      "loss": 0.4232,
+      "step": 2591
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.626519070222957e-06,
+      "loss": 0.4367,
+      "step": 2592
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.611303212315048e-06,
+      "loss": 0.4358,
+      "step": 2593
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.596101383973154e-06,
+      "loss": 0.4386,
+      "step": 2594
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.580913601126757e-06,
+      "loss": 0.4227,
+      "step": 2595
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.565739879690627e-06,
+      "loss": 0.4386,
+      "step": 2596
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.550580235564794e-06,
+      "loss": 0.4301,
+      "step": 2597
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.535434684634543e-06,
+      "loss": 0.4182,
+      "step": 2598
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.52030324277039e-06,
+      "loss": 0.4264,
+      "step": 2599
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.505185925828059e-06,
+      "loss": 0.42,
+      "step": 2600
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.4900827496484875e-06,
+      "loss": 0.4253,
+      "step": 2601
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.474993730057779e-06,
+      "loss": 0.4311,
+      "step": 2602
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.4599188828672155e-06,
+      "loss": 0.4219,
+      "step": 2603
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.444858223873223e-06,
+      "loss": 0.4275,
+      "step": 2604
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.429811768857359e-06,
+      "loss": 0.4358,
+      "step": 2605
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.414779533586293e-06,
+      "loss": 0.4254,
+      "step": 2606
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.3997615338118195e-06,
+      "loss": 0.4246,
+      "step": 2607
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.384757785270777e-06,
+      "loss": 0.4381,
+      "step": 2608
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.369768303685089e-06,
+      "loss": 0.4295,
+      "step": 2609
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.354793104761752e-06,
+      "loss": 0.4313,
+      "step": 2610
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.339832204192753e-06,
+      "loss": 0.4207,
+      "step": 2611
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.3248856176551165e-06,
+      "loss": 0.4451,
+      "step": 2612
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.309953360810883e-06,
+      "loss": 0.4088,
+      "step": 2613
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.2950354493070656e-06,
+      "loss": 0.4293,
+      "step": 2614
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.280131898775621e-06,
+      "loss": 0.4185,
+      "step": 2615
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.265242724833498e-06,
+      "loss": 0.4249,
+      "step": 2616
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.250367943082565e-06,
+      "loss": 0.4522,
+      "step": 2617
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.235507569109587e-06,
+      "loss": 0.4159,
+      "step": 2618
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.220661618486268e-06,
+      "loss": 0.4242,
+      "step": 2619
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.205830106769178e-06,
+      "loss": 0.429,
+      "step": 2620
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.191013049499759e-06,
+      "loss": 0.4193,
+      "step": 2621
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.176210462204306e-06,
+      "loss": 0.422,
+      "step": 2622
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.161422360393958e-06,
+      "loss": 0.4278,
+      "step": 2623
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.146648759564666e-06,
+      "loss": 0.419,
+      "step": 2624
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.131889675197193e-06,
+      "loss": 0.4399,
+      "step": 2625
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.117145122757087e-06,
+      "loss": 0.4253,
+      "step": 2626
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.102415117694671e-06,
+      "loss": 0.4244,
+      "step": 2627
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.087699675445023e-06,
+      "loss": 0.4221,
+      "step": 2628
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.072998811427957e-06,
+      "loss": 0.4304,
+      "step": 2629
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.058312541048021e-06,
+      "loss": 0.4417,
+      "step": 2630
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.043640879694461e-06,
+      "loss": 0.4307,
+      "step": 2631
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.028983842741221e-06,
+      "loss": 0.4266,
+      "step": 2632
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.014341445546908e-06,
+      "loss": 0.4156,
+      "step": 2633
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.999713703454822e-06,
+      "loss": 0.4204,
+      "step": 2634
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.985100631792859e-06,
+      "loss": 0.4274,
+      "step": 2635
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.9705022458735724e-06,
+      "loss": 0.4445,
+      "step": 2636
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.955918560994136e-06,
+      "loss": 0.4228,
+      "step": 2637
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.941349592436285e-06,
+      "loss": 0.4114,
+      "step": 2638
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.9267953554663526e-06,
+      "loss": 0.4277,
+      "step": 2639
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.91225586533525e-06,
+      "loss": 0.4235,
+      "step": 2640
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.897731137278417e-06,
+      "loss": 0.4257,
+      "step": 2641
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.883221186515813e-06,
+      "loss": 0.4266,
+      "step": 2642
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.868726028251948e-06,
+      "loss": 0.4331,
+      "step": 2643
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.854245677675807e-06,
+      "loss": 0.41,
+      "step": 2644
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.83978014996085e-06,
+      "loss": 0.4213,
+      "step": 2645
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.8253294602650346e-06,
+      "loss": 0.4208,
+      "step": 2646
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.810893623730746e-06,
+      "loss": 0.4293,
+      "step": 2647
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.796472655484815e-06,
+      "loss": 0.4309,
+      "step": 2648
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.782066570638489e-06,
+      "loss": 0.4136,
+      "step": 2649
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7676753842874215e-06,
+      "loss": 0.4064,
+      "step": 2650
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.753299111511652e-06,
+      "loss": 0.4295,
+      "step": 2651
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.738937767375596e-06,
+      "loss": 0.4298,
+      "step": 2652
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7245913669280224e-06,
+      "loss": 0.4347,
+      "step": 2653
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7102599252020466e-06,
+      "loss": 0.4409,
+      "step": 2654
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.695943457215101e-06,
+      "loss": 0.4075,
+      "step": 2655
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.681641977968932e-06,
+      "loss": 0.4221,
+      "step": 2656
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.667355502449579e-06,
+      "loss": 0.4368,
+      "step": 2657
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.653084045627371e-06,
+      "loss": 0.4282,
+      "step": 2658
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.638827622456875e-06,
+      "loss": 0.4234,
+      "step": 2659
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.62458624787692e-06,
+      "loss": 0.443,
+      "step": 2660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.610359936810577e-06,
+      "loss": 0.4248,
+      "step": 2661
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.5961487041651055e-06,
+      "loss": 0.4252,
+      "step": 2662
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.581952564831978e-06,
+      "loss": 0.4304,
+      "step": 2663
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.567771533686865e-06,
+      "loss": 0.3976,
+      "step": 2664
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.553605625589595e-06,
+      "loss": 0.4123,
+      "step": 2665
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.539454855384127e-06,
+      "loss": 0.4349,
+      "step": 2666
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.5253192378985966e-06,
+      "loss": 0.4251,
+      "step": 2667
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.5111987879452445e-06,
+      "loss": 0.4258,
+      "step": 2668
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.497093520320396e-06,
+      "loss": 0.4333,
+      "step": 2669
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.483003449804505e-06,
+      "loss": 0.4403,
+      "step": 2670
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.468928591162082e-06,
+      "loss": 0.4296,
+      "step": 2671
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.454868959141695e-06,
+      "loss": 0.4475,
+      "step": 2672
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.4408245684759595e-06,
+      "loss": 0.4235,
+      "step": 2673
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.426795433881527e-06,
+      "loss": 0.4245,
+      "step": 2674
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.412781570059052e-06,
+      "loss": 0.4383,
+      "step": 2675
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.398782991693192e-06,
+      "loss": 0.4218,
+      "step": 2676
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.384799713452591e-06,
+      "loss": 0.4217,
+      "step": 2677
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.370831749989853e-06,
+      "loss": 0.4178,
+      "step": 2678
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.35687911594154e-06,
+      "loss": 0.423,
+      "step": 2679
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.342941825928147e-06,
+      "loss": 0.4272,
+      "step": 2680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.329019894554097e-06,
+      "loss": 0.4241,
+      "step": 2681
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.315113336407711e-06,
+      "loss": 0.4288,
+      "step": 2682
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.301222166061204e-06,
+      "loss": 0.4275,
+      "step": 2683
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.287346398070663e-06,
+      "loss": 0.4175,
+      "step": 2684
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.273486046976057e-06,
+      "loss": 0.4289,
+      "step": 2685
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.259641127301167e-06,
+      "loss": 0.4246,
+      "step": 2686
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.2458116535536144e-06,
+      "loss": 0.4251,
+      "step": 2687
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.231997640224864e-06,
+      "loss": 0.416,
+      "step": 2688
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.2181991017901355e-06,
+      "loss": 0.4393,
+      "step": 2689
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.204416052708454e-06,
+      "loss": 0.4235,
+      "step": 2690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.1906485074226286e-06,
+      "loss": 0.4207,
+      "step": 2691
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.176896480359206e-06,
+      "loss": 0.4173,
+      "step": 2692
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.1631599859284565e-06,
+      "loss": 0.4205,
+      "step": 2693
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.1494390385244105e-06,
+      "loss": 0.4186,
+      "step": 2694
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.1357336525247855e-06,
+      "loss": 0.4115,
+      "step": 2695
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.12204384229098e-06,
+      "loss": 0.422,
+      "step": 2696
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.108369622168103e-06,
+      "loss": 0.4341,
+      "step": 2697
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.094711006484907e-06,
+      "loss": 0.4145,
+      "step": 2698
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.081068009553796e-06,
+      "loss": 0.4384,
+      "step": 2699
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.067440645670807e-06,
+      "loss": 0.4262,
+      "step": 2700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.053828929115598e-06,
+      "loss": 0.4021,
+      "step": 2701
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.040232874151432e-06,
+      "loss": 0.4173,
+      "step": 2702
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.0266524950251575e-06,
+      "loss": 0.4257,
+      "step": 2703
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.0130878059671965e-06,
+      "loss": 0.4294,
+      "step": 2704
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.999538821191534e-06,
+      "loss": 0.4204,
+      "step": 2705
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.986005554895692e-06,
+      "loss": 0.4167,
+      "step": 2706
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.972488021260733e-06,
+      "loss": 0.424,
+      "step": 2707
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.95898623445122e-06,
+      "loss": 0.418,
+      "step": 2708
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.945500208615226e-06,
+      "loss": 0.4115,
+      "step": 2709
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.932029957884299e-06,
+      "loss": 0.4244,
+      "step": 2710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.918575496373472e-06,
+      "loss": 0.4095,
+      "step": 2711
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.905136838181226e-06,
+      "loss": 0.4383,
+      "step": 2712
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.89171399738946e-06,
+      "loss": 0.4242,
+      "step": 2713
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.878306988063539e-06,
+      "loss": 0.431,
+      "step": 2714
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.86491582425221e-06,
+      "loss": 0.4104,
+      "step": 2715
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.851540519987625e-06,
+      "loss": 0.4159,
+      "step": 2716
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.838181089285316e-06,
+      "loss": 0.4243,
+      "step": 2717
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.824837546144183e-06,
+      "loss": 0.4161,
+      "step": 2718
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.811509904546474e-06,
+      "loss": 0.4219,
+      "step": 2719
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.798198178457782e-06,
+      "loss": 0.4322,
+      "step": 2720
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.784902381827015e-06,
+      "loss": 0.4442,
+      "step": 2721
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.771622528586393e-06,
+      "loss": 0.4173,
+      "step": 2722
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.7583586326514275e-06,
+      "loss": 0.4124,
+      "step": 2723
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.74511070792091e-06,
+      "loss": 0.4121,
+      "step": 2724
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.731878768276899e-06,
+      "loss": 0.4141,
+      "step": 2725
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.718662827584694e-06,
+      "loss": 0.4089,
+      "step": 2726
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.705462899692843e-06,
+      "loss": 0.4238,
+      "step": 2727
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6922789984330975e-06,
+      "loss": 0.4378,
+      "step": 2728
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.4076,
+      "step": 2729
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.665959331053021e-06,
+      "loss": 0.4111,
+      "step": 2730
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6528235925121676e-06,
+      "loss": 0.4241,
+      "step": 2731
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.639703935762403e-06,
+      "loss": 0.4127,
+      "step": 2732
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6266003745513555e-06,
+      "loss": 0.4262,
+      "step": 2733
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.613512922609806e-06,
+      "loss": 0.4341,
+      "step": 2734
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.60044159365167e-06,
+      "loss": 0.4314,
+      "step": 2735
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.587386401373952e-06,
+      "loss": 0.4264,
+      "step": 2736
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.574347359456737e-06,
+      "loss": 0.4199,
+      "step": 2737
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.561324481563212e-06,
+      "loss": 0.4177,
+      "step": 2738
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5483177813396155e-06,
+      "loss": 0.4293,
+      "step": 2739
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.535327272415215e-06,
+      "loss": 0.4403,
+      "step": 2740
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.522352968402346e-06,
+      "loss": 0.4286,
+      "step": 2741
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.509394882896338e-06,
+      "loss": 0.4314,
+      "step": 2742
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.4964530294755385e-06,
+      "loss": 0.4357,
+      "step": 2743
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.4835274217012795e-06,
+      "loss": 0.4207,
+      "step": 2744
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.4706180731178695e-06,
+      "loss": 0.4157,
+      "step": 2745
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.457724997252584e-06,
+      "loss": 0.4239,
+      "step": 2746
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.444848207615644e-06,
+      "loss": 0.4092,
+      "step": 2747
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.431987717700206e-06,
+      "loss": 0.4309,
+      "step": 2748
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.419143540982343e-06,
+      "loss": 0.4255,
+      "step": 2749
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.4063156909210395e-06,
+      "loss": 0.4214,
+      "step": 2750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.393504180958166e-06,
+      "loss": 0.4222,
+      "step": 2751
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.380709024518477e-06,
+      "loss": 0.417,
+      "step": 2752
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.367930235009586e-06,
+      "loss": 0.4218,
+      "step": 2753
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.355167825821957e-06,
+      "loss": 0.415,
+      "step": 2754
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.342421810328887e-06,
+      "loss": 0.4277,
+      "step": 2755
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.329692201886515e-06,
+      "loss": 0.4303,
+      "step": 2756
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.316979013833748e-06,
+      "loss": 0.4291,
+      "step": 2757
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.3042822594923164e-06,
+      "loss": 0.4344,
+      "step": 2758
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.2916019521667325e-06,
+      "loss": 0.4276,
+      "step": 2759
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.278938105144255e-06,
+      "loss": 0.4285,
+      "step": 2760
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.2662907316948995e-06,
+      "loss": 0.4304,
+      "step": 2761
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.253659845071436e-06,
+      "loss": 0.4432,
+      "step": 2762
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.24104545850935e-06,
+      "loss": 0.4202,
+      "step": 2763
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.2284475852268136e-06,
+      "loss": 0.4249,
+      "step": 2764
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.2158662384247306e-06,
+      "loss": 0.421,
+      "step": 2765
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.203301431286677e-06,
+      "loss": 0.4081,
+      "step": 2766
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.190753176978872e-06,
+      "loss": 0.429,
+      "step": 2767
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.178221488650229e-06,
+      "loss": 0.4226,
+      "step": 2768
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1657063794322775e-06,
+      "loss": 0.4328,
+      "step": 2769
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.153207862439179e-06,
+      "loss": 0.4226,
+      "step": 2770
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.140725950767712e-06,
+      "loss": 0.4238,
+      "step": 2771
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.128260657497251e-06,
+      "loss": 0.4288,
+      "step": 2772
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.11581199568976e-06,
+      "loss": 0.4322,
+      "step": 2773
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.1033799783897745e-06,
+      "loss": 0.4217,
+      "step": 2774
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.090964618624385e-06,
+      "loss": 0.4268,
+      "step": 2775
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.078565929403233e-06,
+      "loss": 0.4155,
+      "step": 2776
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.066183923718488e-06,
+      "loss": 0.4175,
+      "step": 2777
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.0538186145448396e-06,
+      "loss": 0.4124,
+      "step": 2778
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.041470014839472e-06,
+      "loss": 0.4143,
+      "step": 2779
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.029138137542086e-06,
+      "loss": 0.42,
+      "step": 2780
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.016822995574827e-06,
+      "loss": 0.4216,
+      "step": 2781
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.004524601842317e-06,
+      "loss": 0.4114,
+      "step": 2782
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.992242969231647e-06,
+      "loss": 0.4235,
+      "step": 2783
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.979978110612313e-06,
+      "loss": 0.4323,
+      "step": 2784
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.9677300388362435e-06,
+      "loss": 0.4252,
+      "step": 2785
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.9554987667378e-06,
+      "loss": 0.4354,
+      "step": 2786
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.943284307133717e-06,
+      "loss": 0.4095,
+      "step": 2787
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.931086672823101e-06,
+      "loss": 0.4332,
+      "step": 2788
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.918905876587462e-06,
+      "loss": 0.4126,
+      "step": 2789
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.906741931190648e-06,
+      "loss": 0.3985,
+      "step": 2790
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.894594849378828e-06,
+      "loss": 0.4121,
+      "step": 2791
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.882464643880541e-06,
+      "loss": 0.4184,
+      "step": 2792
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8703513274066186e-06,
+      "loss": 0.4269,
+      "step": 2793
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.858254912650197e-06,
+      "loss": 0.4162,
+      "step": 2794
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.846175412286701e-06,
+      "loss": 0.4213,
+      "step": 2795
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.834112838973836e-06,
+      "loss": 0.4215,
+      "step": 2796
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8220672053515675e-06,
+      "loss": 0.4258,
+      "step": 2797
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.81003852404211e-06,
+      "loss": 0.4391,
+      "step": 2798
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.798026807649917e-06,
+      "loss": 0.4129,
+      "step": 2799
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.786032068761658e-06,
+      "loss": 0.4268,
+      "step": 2800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.774054319946223e-06,
+      "loss": 0.4345,
+      "step": 2801
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7620935737546905e-06,
+      "loss": 0.4148,
+      "step": 2802
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7501498427203254e-06,
+      "loss": 0.4274,
+      "step": 2803
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.738223139358561e-06,
+      "loss": 0.412,
+      "step": 2804
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7263134761669917e-06,
+      "loss": 0.4253,
+      "step": 2805
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7144208656253476e-06,
+      "loss": 0.431,
+      "step": 2806
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.702545320195512e-06,
+      "loss": 0.4245,
+      "step": 2807
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6906868523214543e-06,
+      "loss": 0.4174,
+      "step": 2808
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6788454744292647e-06,
+      "loss": 0.4148,
+      "step": 2809
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.667021198927143e-06,
+      "loss": 0.4228,
+      "step": 2810
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.655214038205335e-06,
+      "loss": 0.4238,
+      "step": 2811
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6434240046361645e-06,
+      "loss": 0.4227,
+      "step": 2812
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.631651110574028e-06,
+      "loss": 0.4177,
+      "step": 2813
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6198953683553396e-06,
+      "loss": 0.4126,
+      "step": 2814
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6081567902985382e-06,
+      "loss": 0.4277,
+      "step": 2815
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5964353887040957e-06,
+      "loss": 0.4109,
+      "step": 2816
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.584731175854479e-06,
+      "loss": 0.4013,
+      "step": 2817
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5730441640141235e-06,
+      "loss": 0.4168,
+      "step": 2818
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5613743654294707e-06,
+      "loss": 0.4235,
+      "step": 2819
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5497217923289083e-06,
+      "loss": 0.4203,
+      "step": 2820
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5380864569227734e-06,
+      "loss": 0.4327,
+      "step": 2821
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5264683714033467e-06,
+      "loss": 0.4284,
+      "step": 2822
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.514867547944827e-06,
+      "loss": 0.4238,
+      "step": 2823
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5032839987033306e-06,
+      "loss": 0.4156,
+      "step": 2824
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4917177358168684e-06,
+      "loss": 0.4118,
+      "step": 2825
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4801687714053413e-06,
+      "loss": 0.4238,
+      "step": 2826
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.468637117570517e-06,
+      "loss": 0.4219,
+      "step": 2827
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.457122786396032e-06,
+      "loss": 0.4171,
+      "step": 2828
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.445625789947369e-06,
+      "loss": 0.4249,
+      "step": 2829
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.434146140271841e-06,
+      "loss": 0.4188,
+      "step": 2830
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.422683849398589e-06,
+      "loss": 0.4356,
+      "step": 2831
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.411238929338563e-06,
+      "loss": 0.4348,
+      "step": 2832
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3998113920845045e-06,
+      "loss": 0.4308,
+      "step": 2833
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3884012496109642e-06,
+      "loss": 0.4165,
+      "step": 2834
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3770085138742294e-06,
+      "loss": 0.4116,
+      "step": 2835
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3656331968123657e-06,
+      "loss": 0.4228,
+      "step": 2836
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.354275310345205e-06,
+      "loss": 0.4265,
+      "step": 2837
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.34293486637427e-06,
+      "loss": 0.4284,
+      "step": 2838
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3316118767828498e-06,
+      "loss": 0.4014,
+      "step": 2839
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3203063534359203e-06,
+      "loss": 0.4084,
+      "step": 2840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.309018308180156e-06,
+      "loss": 0.4136,
+      "step": 2841
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.297747752843925e-06,
+      "loss": 0.4217,
+      "step": 2842
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.286494699237264e-06,
+      "loss": 0.4204,
+      "step": 2843
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.2752591591518666e-06,
+      "loss": 0.4306,
+      "step": 2844
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.264041144361081e-06,
+      "loss": 0.4323,
+      "step": 2845
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.2528406666198874e-06,
+      "loss": 0.4301,
+      "step": 2846
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.2416577376648918e-06,
+      "loss": 0.4167,
+      "step": 2847
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.2304923692143065e-06,
+      "loss": 0.419,
+      "step": 2848
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.219344572967951e-06,
+      "loss": 0.4181,
+      "step": 2849
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.20821436060722e-06,
+      "loss": 0.4093,
+      "step": 2850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1971017437951034e-06,
+      "loss": 0.4275,
+      "step": 2851
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.186006734176126e-06,
+      "loss": 0.4106,
+      "step": 2852
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.4268,
+      "step": 2853
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1638695830034917e-06,
+      "loss": 0.4199,
+      "step": 2854
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1528274646466108e-06,
+      "loss": 0.4173,
+      "step": 2855
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1418029998764022e-06,
+      "loss": 0.4384,
+      "step": 2856
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1307962002450385e-06,
+      "loss": 0.423,
+      "step": 2857
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.119807077286179e-06,
+      "loss": 0.4274,
+      "step": 2858
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.108835642514938e-06,
+      "loss": 0.4291,
+      "step": 2859
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.097881907427931e-06,
+      "loss": 0.4169,
+      "step": 2860
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0869458835032097e-06,
+      "loss": 0.4142,
+      "step": 2861
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0760275822002493e-06,
+      "loss": 0.4203,
+      "step": 2862
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.065127014959988e-06,
+      "loss": 0.4181,
+      "step": 2863
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0542441932047605e-06,
+      "loss": 0.4182,
+      "step": 2864
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0433791283383084e-06,
+      "loss": 0.416,
+      "step": 2865
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0325318317457686e-06,
+      "loss": 0.416,
+      "step": 2866
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0217023147936618e-06,
+      "loss": 0.4278,
+      "step": 2867
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.010890588829871e-06,
+      "loss": 0.4189,
+      "step": 2868
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.000096665183645e-06,
+      "loss": 0.4273,
+      "step": 2869
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.9893205551655713e-06,
+      "loss": 0.4236,
+      "step": 2870
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.9785622700675753e-06,
+      "loss": 0.4211,
+      "step": 2871
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.967821821162904e-06,
+      "loss": 0.4234,
+      "step": 2872
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.95709921970611e-06,
+      "loss": 0.4239,
+      "step": 2873
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9463944769330497e-06,
+      "loss": 0.4235,
+      "step": 2874
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9357076040608646e-06,
+      "loss": 0.422,
+      "step": 2875
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9250386122879694e-06,
+      "loss": 0.4265,
+      "step": 2876
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.914387512794037e-06,
+      "loss": 0.4229,
+      "step": 2877
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.903754316740013e-06,
+      "loss": 0.4169,
+      "step": 2878
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.893139035268049e-06,
+      "loss": 0.4148,
+      "step": 2879
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.8825416795015448e-06,
+      "loss": 0.4438,
+      "step": 2880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.8719622605451293e-06,
+      "loss": 0.4229,
+      "step": 2881
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.861400789484603e-06,
+      "loss": 0.4174,
+      "step": 2882
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.850857277386978e-06,
+      "loss": 0.4093,
+      "step": 2883
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.840331735300459e-06,
+      "loss": 0.4088,
+      "step": 2884
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8298241742544053e-06,
+      "loss": 0.422,
+      "step": 2885
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.819334605259323e-06,
+      "loss": 0.4095,
+      "step": 2886
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.808863039306897e-06,
+      "loss": 0.4062,
+      "step": 2887
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7984094873699264e-06,
+      "loss": 0.3972,
+      "step": 2888
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7879739604023235e-06,
+      "loss": 0.4373,
+      "step": 2889
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.777556469339142e-06,
+      "loss": 0.4259,
+      "step": 2890
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7671570250965184e-06,
+      "loss": 0.4171,
+      "step": 2891
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.756775638571676e-06,
+      "loss": 0.4183,
+      "step": 2892
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7464123206429263e-06,
+      "loss": 0.4136,
+      "step": 2893
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7360670821696422e-06,
+      "loss": 0.4263,
+      "step": 2894
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.7257399339922486e-06,
+      "loss": 0.4157,
+      "step": 2895
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.7154308869322177e-06,
+      "loss": 0.4086,
+      "step": 2896
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.7051399517920575e-06,
+      "loss": 0.423,
+      "step": 2897
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6948671393552907e-06,
+      "loss": 0.4342,
+      "step": 2898
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6846124603864533e-06,
+      "loss": 0.4322,
+      "step": 2899
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6743759256310785e-06,
+      "loss": 0.4303,
+      "step": 2900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.664157545815689e-06,
+      "loss": 0.411,
+      "step": 2901
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6539573316477806e-06,
+      "loss": 0.4224,
+      "step": 2902
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6437752938158177e-06,
+      "loss": 0.4148,
+      "step": 2903
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6336114429892077e-06,
+      "loss": 0.4182,
+      "step": 2904
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.623465789818327e-06,
+      "loss": 0.4272,
+      "step": 2905
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.6133383449344487e-06,
+      "loss": 0.409,
+      "step": 2906
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.6032291189497817e-06,
+      "loss": 0.4253,
+      "step": 2907
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5931381224574594e-06,
+      "loss": 0.4224,
+      "step": 2908
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.583065366031483e-06,
+      "loss": 0.4283,
+      "step": 2909
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.573010860226761e-06,
+      "loss": 0.4232,
+      "step": 2910
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5629746155790747e-06,
+      "loss": 0.4179,
+      "step": 2911
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5529566426050733e-06,
+      "loss": 0.4283,
+      "step": 2912
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.54295695180224e-06,
+      "loss": 0.4121,
+      "step": 2913
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5329755536489266e-06,
+      "loss": 0.4135,
+      "step": 2914
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5230124586043016e-06,
+      "loss": 0.4235,
+      "step": 2915
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5130676771083585e-06,
+      "loss": 0.4198,
+      "step": 2916
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5031412195818993e-06,
+      "loss": 0.4273,
+      "step": 2917
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4932330964265263e-06,
+      "loss": 0.4257,
+      "step": 2918
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.483343318024629e-06,
+      "loss": 0.41,
+      "step": 2919
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4734718947393764e-06,
+      "loss": 0.4133,
+      "step": 2920
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4636188369146983e-06,
+      "loss": 0.4271,
+      "step": 2921
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.453784154875283e-06,
+      "loss": 0.429,
+      "step": 2922
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.443967858926568e-06,
+      "loss": 0.407,
+      "step": 2923
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4341699593547153e-06,
+      "loss": 0.4146,
+      "step": 2924
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4243904664266183e-06,
+      "loss": 0.4134,
+      "step": 2925
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4146293903898798e-06,
+      "loss": 0.4023,
+      "step": 2926
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4048867414728004e-06,
+      "loss": 0.4267,
+      "step": 2927
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.395162529884376e-06,
+      "loss": 0.4144,
+      "step": 2928
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3854567658142892e-06,
+      "loss": 0.4241,
+      "step": 2929
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.375769459432873e-06,
+      "loss": 0.4269,
+      "step": 2930
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.366100620891132e-06,
+      "loss": 0.4256,
+      "step": 2931
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3564502603207284e-06,
+      "loss": 0.4071,
+      "step": 2932
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.34681838783394e-06,
+      "loss": 0.4161,
+      "step": 2933
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3372050135236846e-06,
+      "loss": 0.4101,
+      "step": 2934
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3276101474634994e-06,
+      "loss": 0.4243,
+      "step": 2935
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.318033799707522e-06,
+      "loss": 0.4228,
+      "step": 2936
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3084759802904765e-06,
+      "loss": 0.4103,
+      "step": 2937
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2989366992276917e-06,
+      "loss": 0.4282,
+      "step": 2938
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2894159665150606e-06,
+      "loss": 0.4282,
+      "step": 2939
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2799137921290273e-06,
+      "loss": 0.4118,
+      "step": 2940
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2704301860266133e-06,
+      "loss": 0.4355,
+      "step": 2941
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2609651581453674e-06,
+      "loss": 0.411,
+      "step": 2942
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.251518718403374e-06,
+      "loss": 0.4198,
+      "step": 2943
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.242090876699241e-06,
+      "loss": 0.4145,
+      "step": 2944
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2326816429120867e-06,
+      "loss": 0.4191,
+      "step": 2945
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.223291026901533e-06,
+      "loss": 0.4187,
+      "step": 2946
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.213919038507688e-06,
+      "loss": 0.4155,
+      "step": 2947
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2045656875511455e-06,
+      "loss": 0.4319,
+      "step": 2948
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.19523098383297e-06,
+      "loss": 0.4048,
+      "step": 2949
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.185914937134681e-06,
+      "loss": 0.4249,
+      "step": 2950
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.1766175572182547e-06,
+      "loss": 0.4147,
+      "step": 2951
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1673388538260996e-06,
+      "loss": 0.4195,
+      "step": 2952
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.158078836681061e-06,
+      "loss": 0.4215,
+      "step": 2953
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.148837515486397e-06,
+      "loss": 0.4222,
+      "step": 2954
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.139614899925777e-06,
+      "loss": 0.423,
+      "step": 2955
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1304109996632795e-06,
+      "loss": 0.4121,
+      "step": 2956
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1212258243433535e-06,
+      "loss": 0.4352,
+      "step": 2957
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.112059383590832e-06,
+      "loss": 0.416,
+      "step": 2958
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1029116870109356e-06,
+      "loss": 0.4133,
+      "step": 2959
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.093782744189217e-06,
+      "loss": 0.434,
+      "step": 2960
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.0846725646915877e-06,
+      "loss": 0.4235,
+      "step": 2961
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.075581158064308e-06,
+      "loss": 0.4216,
+      "step": 2962
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.066508533833955e-06,
+      "loss": 0.4216,
+      "step": 2963
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.05745470150742e-06,
+      "loss": 0.4111,
+      "step": 2964
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0484196705719172e-06,
+      "loss": 0.4317,
+      "step": 2965
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.03940345049495e-06,
+      "loss": 0.4271,
+      "step": 2966
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0304060507243138e-06,
+      "loss": 0.4052,
+      "step": 2967
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.021427480688081e-06,
+      "loss": 0.4205,
+      "step": 2968
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0124677497945953e-06,
+      "loss": 0.417,
+      "step": 2969
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.003526867432455e-06,
+      "loss": 0.4027,
+      "step": 2970
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9946048429705133e-06,
+      "loss": 0.4265,
+      "step": 2971
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9857016857578594e-06,
+      "loss": 0.4129,
+      "step": 2972
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.976817405123812e-06,
+      "loss": 0.4125,
+      "step": 2973
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9679520103779116e-06,
+      "loss": 0.4159,
+      "step": 2974
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9591055108099023e-06,
+      "loss": 0.4195,
+      "step": 2975
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.950277915689749e-06,
+      "loss": 0.4052,
+      "step": 2976
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.941469234267579e-06,
+      "loss": 0.4009,
+      "step": 2977
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.932679475773711e-06,
+      "loss": 0.4289,
+      "step": 2978
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.923908649418651e-06,
+      "loss": 0.44,
+      "step": 2979
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9151567643930514e-06,
+      "loss": 0.4125,
+      "step": 2980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9064238298677052e-06,
+      "loss": 0.4194,
+      "step": 2981
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8977098549935745e-06,
+      "loss": 0.4118,
+      "step": 2982
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8890148489017423e-06,
+      "loss": 0.4043,
+      "step": 2983
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8803388207034001e-06,
+      "loss": 0.4284,
+      "step": 2984
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8716817794898801e-06,
+      "loss": 0.427,
+      "step": 2985
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8630437343326013e-06,
+      "loss": 0.4064,
+      "step": 2986
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.854424694283077e-06,
+      "loss": 0.4149,
+      "step": 2987
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8458246683729152e-06,
+      "loss": 0.4259,
+      "step": 2988
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8372436656137949e-06,
+      "loss": 0.4166,
+      "step": 2989
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.828681694997456e-06,
+      "loss": 0.4259,
+      "step": 2990
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8201387654957026e-06,
+      "loss": 0.4071,
+      "step": 2991
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8116148860603843e-06,
+      "loss": 0.4171,
+      "step": 2992
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.803110065623388e-06,
+      "loss": 0.4068,
+      "step": 2993
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7946243130966311e-06,
+      "loss": 0.4179,
+      "step": 2994
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7861576373720458e-06,
+      "loss": 0.4131,
+      "step": 2995
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.777710047321579e-06,
+      "loss": 0.4313,
+      "step": 2996
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7692815517971773e-06,
+      "loss": 0.4471,
+      "step": 2997
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7608721596307777e-06,
+      "loss": 0.4111,
+      "step": 2998
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7524818796342957e-06,
+      "loss": 0.4299,
+      "step": 2999
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7441107205996344e-06,
+      "loss": 0.4241,
+      "step": 3000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7357586912986434e-06,
+      "loss": 0.4442,
+      "step": 3001
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7274258004831312e-06,
+      "loss": 0.4184,
+      "step": 3002
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.719112056884864e-06,
+      "loss": 0.4145,
+      "step": 3003
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7108174692155266e-06,
+      "loss": 0.4259,
+      "step": 3004
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7025420461667374e-06,
+      "loss": 0.435,
+      "step": 3005
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6942857964100401e-06,
+      "loss": 0.4245,
+      "step": 3006
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6860487285968852e-06,
+      "loss": 0.4296,
+      "step": 3007
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6778308513586084e-06,
+      "loss": 0.4272,
+      "step": 3008
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6696321733064546e-06,
+      "loss": 0.4104,
+      "step": 3009
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6614527030315475e-06,
+      "loss": 0.3986,
+      "step": 3010
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6532924491048662e-06,
+      "loss": 0.4317,
+      "step": 3011
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6451514200772779e-06,
+      "loss": 0.4282,
+      "step": 3012
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6370296244794914e-06,
+      "loss": 0.4132,
+      "step": 3013
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6289270708220638e-06,
+      "loss": 0.4187,
+      "step": 3014
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.620843767595388e-06,
+      "loss": 0.4052,
+      "step": 3015
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6127797232696863e-06,
+      "loss": 0.4106,
+      "step": 3016
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6047349462950012e-06,
+      "loss": 0.4335,
+      "step": 3017
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.596709445101181e-06,
+      "loss": 0.4183,
+      "step": 3018
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5887032280978831e-06,
+      "loss": 0.4214,
+      "step": 3019
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5807163036745543e-06,
+      "loss": 0.4102,
+      "step": 3020
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5727486802004198e-06,
+      "loss": 0.4185,
+      "step": 3021
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5648003660244882e-06,
+      "loss": 0.4225,
+      "step": 3022
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5568713694755299e-06,
+      "loss": 0.4164,
+      "step": 3023
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.548961698862077e-06,
+      "loss": 0.4201,
+      "step": 3024
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5410713624724039e-06,
+      "loss": 0.4265,
+      "step": 3025
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5332003685745279e-06,
+      "loss": 0.4148,
+      "step": 3026
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5253487254162113e-06,
+      "loss": 0.4157,
+      "step": 3027
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5175164412249132e-06,
+      "loss": 0.4049,
+      "step": 3028
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.509703524207824e-06,
+      "loss": 0.4138,
+      "step": 3029
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5019099825518457e-06,
+      "loss": 0.4201,
+      "step": 3030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4941358244235615e-06,
+      "loss": 0.4276,
+      "step": 3031
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4863810579692484e-06,
+      "loss": 0.4276,
+      "step": 3032
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.478645691314875e-06,
+      "loss": 0.4099,
+      "step": 3033
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4709297325660709e-06,
+      "loss": 0.4111,
+      "step": 3034
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4632331898081199e-06,
+      "loss": 0.4386,
+      "step": 3035
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4555560711059792e-06,
+      "loss": 0.421,
+      "step": 3036
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4478983845042493e-06,
+      "loss": 0.4124,
+      "step": 3037
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4402601380271474e-06,
+      "loss": 0.4156,
+      "step": 3038
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4326413396785488e-06,
+      "loss": 0.4287,
+      "step": 3039
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4250419974419317e-06,
+      "loss": 0.4044,
+      "step": 3040
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4174621192803928e-06,
+      "loss": 0.415,
+      "step": 3041
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.40990171313663e-06,
+      "loss": 0.4179,
+      "step": 3042
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4023607869329436e-06,
+      "loss": 0.4295,
+      "step": 3043
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3948393485712132e-06,
+      "loss": 0.4246,
+      "step": 3044
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3873374059329004e-06,
+      "loss": 0.4333,
+      "step": 3045
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.379854966879044e-06,
+      "loss": 0.4275,
+      "step": 3046
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3723920392502366e-06,
+      "loss": 0.4239,
+      "step": 3047
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3649486308666314e-06,
+      "loss": 0.4045,
+      "step": 3048
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3575247495279275e-06,
+      "loss": 0.4159,
+      "step": 3049
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3501204030133552e-06,
+      "loss": 0.4184,
+      "step": 3050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3427355990816903e-06,
+      "loss": 0.4264,
+      "step": 3051
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3353703454712142e-06,
+      "loss": 0.4348,
+      "step": 3052
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3280246498997262e-06,
+      "loss": 0.4124,
+      "step": 3053
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.320698520064545e-06,
+      "loss": 0.4222,
+      "step": 3054
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3133919636424653e-06,
+      "loss": 0.4107,
+      "step": 3055
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3061049882897848e-06,
+      "loss": 0.405,
+      "step": 3056
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2988376016422844e-06,
+      "loss": 0.4123,
+      "step": 3057
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.291589811315217e-06,
+      "loss": 0.4136,
+      "step": 3058
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2843616249032874e-06,
+      "loss": 0.406,
+      "step": 3059
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2771530499806794e-06,
+      "loss": 0.4293,
+      "step": 3060
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2699640941010173e-06,
+      "loss": 0.4173,
+      "step": 3061
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.262794764797355e-06,
+      "loss": 0.4055,
+      "step": 3062
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2556450695822053e-06,
+      "loss": 0.4299,
+      "step": 3063
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.248515015947489e-06,
+      "loss": 0.4081,
+      "step": 3064
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2414046113645473e-06,
+      "loss": 0.4134,
+      "step": 3065
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2343138632841401e-06,
+      "loss": 0.4385,
+      "step": 3066
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2272427791364195e-06,
+      "loss": 0.3987,
+      "step": 3067
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2201913663309383e-06,
+      "loss": 0.4248,
+      "step": 3068
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2131596322566374e-06,
+      "loss": 0.4139,
+      "step": 3069
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.4212,
+      "step": 3070
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1991552297542164e-06,
+      "loss": 0.4224,
+      "step": 3071
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.19218257600084e-06,
+      "loss": 0.4174,
+      "step": 3072
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.185229630328113e-06,
+      "loss": 0.4272,
+      "step": 3073
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1782964000217944e-06,
+      "loss": 0.4182,
+      "step": 3074
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.171382892346984e-06,
+      "loss": 0.4124,
+      "step": 3075
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1644891145481152e-06,
+      "loss": 0.4086,
+      "step": 3076
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1576150738489412e-06,
+      "loss": 0.4044,
+      "step": 3077
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1507607774525487e-06,
+      "loss": 0.4146,
+      "step": 3078
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.143926232541317e-06,
+      "loss": 0.4129,
+      "step": 3079
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.137111446276935e-06,
+      "loss": 0.404,
+      "step": 3080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1303164258003974e-06,
+      "loss": 0.4156,
+      "step": 3081
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1235411782319706e-06,
+      "loss": 0.4119,
+      "step": 3082
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1167857106712066e-06,
+      "loss": 0.4094,
+      "step": 3083
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.110050030196943e-06,
+      "loss": 0.4385,
+      "step": 3084
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1033341438672673e-06,
+      "loss": 0.4277,
+      "step": 3085
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.096638058719528e-06,
+      "loss": 0.4003,
+      "step": 3086
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0899617817703344e-06,
+      "loss": 0.4331,
+      "step": 3087
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0833053200155353e-06,
+      "loss": 0.4097,
+      "step": 3088
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0766686804302018e-06,
+      "loss": 0.4186,
+      "step": 3089
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0700518699686558e-06,
+      "loss": 0.431,
+      "step": 3090
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0634548955644264e-06,
+      "loss": 0.4221,
+      "step": 3091
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0568777641302663e-06,
+      "loss": 0.3968,
+      "step": 3092
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0503204825581247e-06,
+      "loss": 0.425,
+      "step": 3093
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.043783057719161e-06,
+      "loss": 0.4151,
+      "step": 3094
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.037265496463722e-06,
+      "loss": 0.4232,
+      "step": 3095
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0307678056213422e-06,
+      "loss": 0.42,
+      "step": 3096
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.024289992000731e-06,
+      "loss": 0.4144,
+      "step": 3097
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0178320623897786e-06,
+      "loss": 0.4127,
+      "step": 3098
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0113940235555253e-06,
+      "loss": 0.4079,
+      "step": 3099
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0049758822441768e-06,
+      "loss": 0.4198,
+      "step": 3100
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.985776451810935e-07,
+      "loss": 0.4189,
+      "step": 3101
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.921993190707679e-07,
+      "loss": 0.4181,
+      "step": 3102
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.858409105968337e-07,
+      "loss": 0.4361,
+      "step": 3103
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.795024264220587e-07,
+      "loss": 0.4281,
+      "step": 3104
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.731838731883258e-07,
+      "loss": 0.4182,
+      "step": 3105
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.6688525751663e-07,
+      "loss": 0.4123,
+      "step": 3106
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.606065860070845e-07,
+      "loss": 0.4114,
+      "step": 3107
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.543478652389004e-07,
+      "loss": 0.4132,
+      "step": 3108
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.481091017703737e-07,
+      "loss": 0.4086,
+      "step": 3109
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.418903021388992e-07,
+      "loss": 0.4091,
+      "step": 3110
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.356914728609489e-07,
+      "loss": 0.4197,
+      "step": 3111
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.295126204320693e-07,
+      "loss": 0.426,
+      "step": 3112
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.233537513268698e-07,
+      "loss": 0.4113,
+      "step": 3113
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.172148719990237e-07,
+      "loss": 0.4186,
+      "step": 3114
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.110959888812565e-07,
+      "loss": 0.4342,
+      "step": 3115
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.049971083853392e-07,
+      "loss": 0.4001,
+      "step": 3116
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.989182369020866e-07,
+      "loss": 0.4059,
+      "step": 3117
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.928593808013408e-07,
+      "loss": 0.4179,
+      "step": 3118
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.868205464319768e-07,
+      "loss": 0.4072,
+      "step": 3119
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.80801740121886e-07,
+      "loss": 0.4204,
+      "step": 3120
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.748029681779701e-07,
+      "loss": 0.4174,
+      "step": 3121
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.688242368861477e-07,
+      "loss": 0.4099,
+      "step": 3122
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.62865552511325e-07,
+      "loss": 0.4117,
+      "step": 3123
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.569269212974052e-07,
+      "loss": 0.4182,
+      "step": 3124
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.510083494672905e-07,
+      "loss": 0.4297,
+      "step": 3125
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.451098432228444e-07,
+      "loss": 0.3967,
+      "step": 3126
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.392314087449139e-07,
+      "loss": 0.417,
+      "step": 3127
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.333730521933204e-07,
+      "loss": 0.4155,
+      "step": 3128
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.275347797068356e-07,
+      "loss": 0.4197,
+      "step": 3129
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.217165974031837e-07,
+      "loss": 0.4343,
+      "step": 3130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.159185113790502e-07,
+      "loss": 0.4163,
+      "step": 3131
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.4096,
+      "step": 3132
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.043826524507437e-07,
+      "loss": 0.4259,
+      "step": 3133
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.986448916346123e-07,
+      "loss": 0.4342,
+      "step": 3134
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.929272512740605e-07,
+      "loss": 0.4056,
+      "step": 3135
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.872297373604154e-07,
+      "loss": 0.4192,
+      "step": 3136
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.815523558639082e-07,
+      "loss": 0.4098,
+      "step": 3137
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.758951127336778e-07,
+      "loss": 0.4338,
+      "step": 3138
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.702580138977577e-07,
+      "loss": 0.4123,
+      "step": 3139
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.646410652630786e-07,
+      "loss": 0.4165,
+      "step": 3140
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.590442727154479e-07,
+      "loss": 0.409,
+      "step": 3141
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.534676421195586e-07,
+      "loss": 0.4213,
+      "step": 3142
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.479111793189742e-07,
+      "loss": 0.4178,
+      "step": 3143
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.423748901361261e-07,
+      "loss": 0.4177,
+      "step": 3144
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.36858780372307e-07,
+      "loss": 0.4296,
+      "step": 3145
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.313628558076602e-07,
+      "loss": 0.4159,
+      "step": 3146
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.258871222011832e-07,
+      "loss": 0.3975,
+      "step": 3147
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.204315852907106e-07,
+      "loss": 0.4272,
+      "step": 3148
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.149962507929231e-07,
+      "loss": 0.4164,
+      "step": 3149
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.095811244033201e-07,
+      "loss": 0.4359,
+      "step": 3150
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.041862117962273e-07,
+      "loss": 0.4108,
+      "step": 3151
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.988115186248001e-07,
+      "loss": 0.3937,
+      "step": 3152
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.934570505209958e-07,
+      "loss": 0.4096,
+      "step": 3153
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.881228130955775e-07,
+      "loss": 0.4277,
+      "step": 3154
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.828088119381205e-07,
+      "loss": 0.4244,
+      "step": 3155
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.775150526169883e-07,
+      "loss": 0.4155,
+      "step": 3156
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.722415406793259e-07,
+      "loss": 0.4065,
+      "step": 3157
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.669882816510776e-07,
+      "loss": 0.4102,
+      "step": 3158
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.617552810369554e-07,
+      "loss": 0.4215,
+      "step": 3159
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.565425443204398e-07,
+      "loss": 0.413,
+      "step": 3160
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.513500769637882e-07,
+      "loss": 0.4081,
+      "step": 3161
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.461778844080102e-07,
+      "loss": 0.4159,
+      "step": 3162
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.410259720728751e-07,
+      "loss": 0.4227,
+      "step": 3163
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.358943453568978e-07,
+      "loss": 0.4091,
+      "step": 3164
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.307830096373369e-07,
+      "loss": 0.4094,
+      "step": 3165
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.256919702701903e-07,
+      "loss": 0.4255,
+      "step": 3166
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.206212325901862e-07,
+      "loss": 0.4285,
+      "step": 3167
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.155708019107831e-07,
+      "loss": 0.4241,
+      "step": 3168
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.105406835241545e-07,
+      "loss": 0.4019,
+      "step": 3169
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.055308827011952e-07,
+      "loss": 0.4034,
+      "step": 3170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.005414046915059e-07,
+      "loss": 0.4236,
+      "step": 3171
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.955722547233934e-07,
+      "loss": 0.4129,
+      "step": 3172
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.906234380038633e-07,
+      "loss": 0.4023,
+      "step": 3173
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.856949597186146e-07,
+      "loss": 0.4254,
+      "step": 3174
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.807868250320337e-07,
+      "loss": 0.4262,
+      "step": 3175
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.758990390871932e-07,
+      "loss": 0.4045,
+      "step": 3176
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.710316070058364e-07,
+      "loss": 0.3986,
+      "step": 3177
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.66184533888381e-07,
+      "loss": 0.4146,
+      "step": 3178
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.613578248139173e-07,
+      "loss": 0.4205,
+      "step": 3179
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.565514848401887e-07,
+      "loss": 0.4271,
+      "step": 3180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.517655190035954e-07,
+      "loss": 0.4218,
+      "step": 3181
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.469999323191944e-07,
+      "loss": 0.4029,
+      "step": 3182
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.422547297806846e-07,
+      "loss": 0.3997,
+      "step": 3183
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.375299163603975e-07,
+      "loss": 0.4102,
+      "step": 3184
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.32825497009315e-07,
+      "loss": 0.4394,
+      "step": 3185
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.281414766570337e-07,
+      "loss": 0.413,
+      "step": 3186
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.234778602117851e-07,
+      "loss": 0.418,
+      "step": 3187
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.18834652560416e-07,
+      "loss": 0.4101,
+      "step": 3188
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.142118585683875e-07,
+      "loss": 0.4321,
+      "step": 3189
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.096094830797693e-07,
+      "loss": 0.438,
+      "step": 3190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.05027530917237e-07,
+      "loss": 0.4273,
+      "step": 3191
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.004660068820654e-07,
+      "loss": 0.4176,
+      "step": 3192
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.959249157541202e-07,
+      "loss": 0.4103,
+      "step": 3193
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.914042622918591e-07,
+      "loss": 0.4221,
+      "step": 3194
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.869040512323264e-07,
+      "loss": 0.4243,
+      "step": 3195
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.824242872911367e-07,
+      "loss": 0.3852,
+      "step": 3196
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.779649751624882e-07,
+      "loss": 0.4022,
+      "step": 3197
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.735261195191432e-07,
+      "loss": 0.4175,
+      "step": 3198
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.691077250124276e-07,
+      "loss": 0.4172,
+      "step": 3199
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.6470979627223357e-07,
+      "loss": 0.4088,
+      "step": 3200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.6033233790699907e-07,
+      "loss": 0.415,
+      "step": 3201
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.559753545037171e-07,
+      "loss": 0.4082,
+      "step": 3202
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.516388506279246e-07,
+      "loss": 0.4093,
+      "step": 3203
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.4732283082369767e-07,
+      "loss": 0.4196,
+      "step": 3204
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.4302729961364754e-07,
+      "loss": 0.4173,
+      "step": 3205
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.387522614989204e-07,
+      "loss": 0.4181,
+      "step": 3206
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.3449772095918606e-07,
+      "loss": 0.4135,
+      "step": 3207
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.302636824526274e-07,
+      "loss": 0.4257,
+      "step": 3208
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.2605015041595755e-07,
+      "loss": 0.4198,
+      "step": 3209
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.2185712926439584e-07,
+      "loss": 0.4285,
+      "step": 3210
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.1768462339166317e-07,
+      "loss": 0.4056,
+      "step": 3211
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.135326371699933e-07,
+      "loss": 0.4045,
+      "step": 3212
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.094011749501103e-07,
+      "loss": 0.4117,
+      "step": 3213
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.05290241061238e-07,
+      "loss": 0.4119,
+      "step": 3214
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.0119983981108166e-07,
+      "loss": 0.419,
+      "step": 3215
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.9712997548583933e-07,
+      "loss": 0.4143,
+      "step": 3216
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.93080652350184e-07,
+      "loss": 0.4128,
+      "step": 3217
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.89051874647266e-07,
+      "loss": 0.423,
+      "step": 3218
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.8504364659870396e-07,
+      "loss": 0.4205,
+      "step": 3219
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.8105597240458703e-07,
+      "loss": 0.4243,
+      "step": 3220
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7708885624346604e-07,
+      "loss": 0.424,
+      "step": 3221
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.731423022723468e-07,
+      "loss": 0.4176,
+      "step": 3222
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6921631462669473e-07,
+      "loss": 0.4235,
+      "step": 3223
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.653108974204145e-07,
+      "loss": 0.4171,
+      "step": 3224
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.614260547458659e-07,
+      "loss": 0.4161,
+      "step": 3225
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.575617906738438e-07,
+      "loss": 0.4149,
+      "step": 3226
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.537181092535846e-07,
+      "loss": 0.4118,
+      "step": 3227
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.498950145127467e-07,
+      "loss": 0.4338,
+      "step": 3228
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.4609251045742754e-07,
+      "loss": 0.4188,
+      "step": 3229
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.423106010721422e-07,
+      "loss": 0.4284,
+      "step": 3230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3854929031982285e-07,
+      "loss": 0.4195,
+      "step": 3231
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.348085821418279e-07,
+      "loss": 0.4135,
+      "step": 3232
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.310884804579151e-07,
+      "loss": 0.4249,
+      "step": 3233
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.2738898916625516e-07,
+      "loss": 0.4097,
+      "step": 3234
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.2371011214342053e-07,
+      "loss": 0.4117,
+      "step": 3235
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.2005185324438304e-07,
+      "loss": 0.4067,
+      "step": 3236
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.1641421630250967e-07,
+      "loss": 0.4081,
+      "step": 3237
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.12797205129558e-07,
+      "loss": 0.4178,
+      "step": 3238
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.0920082351567404e-07,
+      "loss": 0.4239,
+      "step": 3239
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.0562507522938324e-07,
+      "loss": 0.4004,
+      "step": 3240
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.02069964017595e-07,
+      "loss": 0.4157,
+      "step": 3241
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.985354936055873e-07,
+      "loss": 0.4088,
+      "step": 3242
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.95021667697013e-07,
+      "loss": 0.4121,
+      "step": 3243
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.9152848997389793e-07,
+      "loss": 0.41,
+      "step": 3244
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8805596409662075e-07,
+      "loss": 0.4002,
+      "step": 3245
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8460409370392405e-07,
+      "loss": 0.4137,
+      "step": 3246
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.811728824129123e-07,
+      "loss": 0.4227,
+      "step": 3247
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.7776233381903164e-07,
+      "loss": 0.3988,
+      "step": 3248
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.7437245149608325e-07,
+      "loss": 0.4147,
+      "step": 3249
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.710032389962103e-07,
+      "loss": 0.422,
+      "step": 3250
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.676546998499019e-07,
+      "loss": 0.3967,
+      "step": 3251
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.643268375659713e-07,
+      "loss": 0.4252,
+      "step": 3252
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.610196556315803e-07,
+      "loss": 0.4196,
+      "step": 3253
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.577331575122144e-07,
+      "loss": 0.4169,
+      "step": 3254
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.544673466516789e-07,
+      "loss": 0.4255,
+      "step": 3255
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5122222647211427e-07,
+      "loss": 0.3924,
+      "step": 3256
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.479978003739669e-07,
+      "loss": 0.4133,
+      "step": 3257
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4479407173600753e-07,
+      "loss": 0.416,
+      "step": 3258
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.4161104391531744e-07,
+      "loss": 0.4046,
+      "step": 3259
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3844872024728183e-07,
+      "loss": 0.4099,
+      "step": 3260
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3530710404559675e-07,
+      "loss": 0.4113,
+      "step": 3261
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.321861986022511e-07,
+      "loss": 0.4357,
+      "step": 3262
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2908600718754227e-07,
+      "loss": 0.4068,
+      "step": 3263
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2600653305005382e-07,
+      "loss": 0.4304,
+      "step": 3264
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2294777941666457e-07,
+      "loss": 0.4039,
+      "step": 3265
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1990974949253953e-07,
+      "loss": 0.405,
+      "step": 3266
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1689244646112774e-07,
+      "loss": 0.409,
+      "step": 3267
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.138958734841623e-07,
+      "loss": 0.4262,
+      "step": 3268
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1092003370164705e-07,
+      "loss": 0.4029,
+      "step": 3269
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0796493023186538e-07,
+      "loss": 0.4306,
+      "step": 3270
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0503056617137808e-07,
+      "loss": 0.4276,
+      "step": 3271
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0211694459499664e-07,
+      "loss": 0.4097,
+      "step": 3272
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9922406855581221e-07,
+      "loss": 0.4117,
+      "step": 3273
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9635194108517552e-07,
+      "loss": 0.4062,
+      "step": 3274
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.93500565192688e-07,
+      "loss": 0.4165,
+      "step": 3275
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9066994386621073e-07,
+      "loss": 0.4025,
+      "step": 3276
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8786008007185774e-07,
+      "loss": 0.4058,
+      "step": 3277
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8507097675399376e-07,
+      "loss": 0.4105,
+      "step": 3278
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.823026368352232e-07,
+      "loss": 0.4135,
+      "step": 3279
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.795550632163967e-07,
+      "loss": 0.4127,
+      "step": 3280
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.768282587766068e-07,
+      "loss": 0.4247,
+      "step": 3281
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7412222637317678e-07,
+      "loss": 0.408,
+      "step": 3282
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7143696884167172e-07,
+      "loss": 0.4202,
+      "step": 3283
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6877248899587863e-07,
+      "loss": 0.3902,
+      "step": 3284
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6612878962781743e-07,
+      "loss": 0.4173,
+      "step": 3285
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.635058735077344e-07,
+      "loss": 0.4375,
+      "step": 3286
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.609037433840932e-07,
+      "loss": 0.42,
+      "step": 3287
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.583224019835794e-07,
+      "loss": 0.4043,
+      "step": 3288
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5576185201109595e-07,
+      "loss": 0.4076,
+      "step": 3289
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5322209614975214e-07,
+      "loss": 0.4292,
+      "step": 3290
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.507031370608769e-07,
+      "loss": 0.4142,
+      "step": 3291
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4820497738400108e-07,
+      "loss": 0.4011,
+      "step": 3292
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.45727619736864e-07,
+      "loss": 0.4005,
+      "step": 3293
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4327106671540248e-07,
+      "loss": 0.4153,
+      "step": 3294
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4083532089376185e-07,
+      "loss": 0.4275,
+      "step": 3295
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3842038482427157e-07,
+      "loss": 0.4029,
+      "step": 3296
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.360262610374652e-07,
+      "loss": 0.4333,
+      "step": 3297
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3365295204206707e-07,
+      "loss": 0.4226,
+      "step": 3298
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3130046032498568e-07,
+      "loss": 0.4161,
+      "step": 3299
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2896878835131578e-07,
+      "loss": 0.4232,
+      "step": 3300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2665793856434516e-07,
+      "loss": 0.4179,
+      "step": 3301
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.24367913385528e-07,
+      "loss": 0.425,
+      "step": 3302
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.22098715214507e-07,
+      "loss": 0.4246,
+      "step": 3303
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.198503464291001e-07,
+      "loss": 0.4103,
+      "step": 3304
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1762280938530047e-07,
+      "loss": 0.4205,
+      "step": 3305
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1541610641726319e-07,
+      "loss": 0.4286,
+      "step": 3306
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1323023983732085e-07,
+      "loss": 0.4277,
+      "step": 3307
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.110652119359723e-07,
+      "loss": 0.418,
+      "step": 3308
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0892102498187396e-07,
+      "loss": 0.4098,
+      "step": 3309
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0679768122184853e-07,
+      "loss": 0.4068,
+      "step": 3310
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0469518288087621e-07,
+      "loss": 0.4218,
+      "step": 3311
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0261353216209691e-07,
+      "loss": 0.416,
+      "step": 3312
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0055273124680132e-07,
+      "loss": 0.4048,
+      "step": 3313
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.851278229443095e-08,
+      "loss": 0.4071,
+      "step": 3314
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.649368744258258e-08,
+      "loss": 0.4109,
+      "step": 3315
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.449544880699713e-08,
+      "loss": 0.3968,
+      "step": 3316
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.251806848156186e-08,
+      "loss": 0.4087,
+      "step": 3317
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.056154853830823e-08,
+      "loss": 0.4097,
+      "step": 3318
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.862589102740737e-08,
+      "loss": 0.4172,
+      "step": 3319
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.671109797716792e-08,
+      "loss": 0.4098,
+      "step": 3320
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.481717139403823e-08,
+      "loss": 0.4206,
+      "step": 3321
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.294411326260188e-08,
+      "loss": 0.4104,
+      "step": 3322
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.109192554557333e-08,
+      "loss": 0.4257,
+      "step": 3323
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.92606101837956e-08,
+      "loss": 0.4047,
+      "step": 3324
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.74501690962448e-08,
+      "loss": 0.4117,
+      "step": 3325
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.56606041800212e-08,
+      "loss": 0.4057,
+      "step": 3326
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.389191731034917e-08,
+      "loss": 0.4263,
+      "step": 3327
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.214411034057512e-08,
+      "loss": 0.4304,
+      "step": 3328
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.041718510217177e-08,
+      "loss": 0.4133,
+      "step": 3329
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.871114340472051e-08,
+      "loss": 0.4182,
+      "step": 3330
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.702598703592911e-08,
+      "loss": 0.4119,
+      "step": 3331
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.536171776161393e-08,
+      "loss": 0.4174,
+      "step": 3332
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.371833732570887e-08,
+      "loss": 0.4235,
+      "step": 3333
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.209584745025643e-08,
+      "loss": 0.4066,
+      "step": 3334
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.049424983540775e-08,
+      "loss": 0.4173,
+      "step": 3335
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.8913546159427014e-08,
+      "loss": 0.4237,
+      "step": 3336
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.735373807867595e-08,
+      "loss": 0.404,
+      "step": 3337
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.58148272276271e-08,
+      "loss": 0.4134,
+      "step": 3338
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.4296815218854994e-08,
+      "loss": 0.4152,
+      "step": 3339
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.2799703643031664e-08,
+      "loss": 0.4211,
+      "step": 3340
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.132349406893111e-08,
+      "loss": 0.4167,
+      "step": 3341
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.9868188043422635e-08,
+      "loss": 0.415,
+      "step": 3342
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.8433787091473064e-08,
+      "loss": 0.4138,
+      "step": 3343
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.7020292716142277e-08,
+      "loss": 0.4174,
+      "step": 3344
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.562770639858549e-08,
+      "loss": 0.4375,
+      "step": 3345
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4256029598046536e-08,
+      "loss": 0.4,
+      "step": 3346
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2905263751857884e-08,
+      "loss": 0.3957,
+      "step": 3347
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.157541027544287e-08,
+      "loss": 0.404,
+      "step": 3348
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.026647056231348e-08,
+      "loss": 0.4072,
+      "step": 3349
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.897844598405698e-08,
+      "loss": 0.411,
+      "step": 3350
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.771133789035597e-08,
+      "loss": 0.4044,
+      "step": 3351
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.646514760897057e-08,
+      "loss": 0.4104,
+      "step": 3352
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.523987644573845e-08,
+      "loss": 0.4155,
+      "step": 3353
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.403552568458146e-08,
+      "loss": 0.4173,
+      "step": 3354
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.285209658750121e-08,
+      "loss": 0.4135,
+      "step": 3355
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1689590394570204e-08,
+      "loss": 0.4209,
+      "step": 3356
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.054800832394067e-08,
+      "loss": 0.4087,
+      "step": 3357
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.942735157183796e-08,
+      "loss": 0.3905,
+      "step": 3358
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8327621312560503e-08,
+      "loss": 0.4068,
+      "step": 3359
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.724881869847984e-08,
+      "loss": 0.3957,
+      "step": 3360
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.619094486003615e-08,
+      "loss": 0.4048,
+      "step": 3361
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.5154000905738274e-08,
+      "loss": 0.4236,
+      "step": 3362
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4137987922168148e-08,
+      "loss": 0.4156,
+      "step": 3363
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.3142906973971923e-08,
+      "loss": 0.429,
+      "step": 3364
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.216875910385774e-08,
+      "loss": 0.4089,
+      "step": 3365
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1215545332604616e-08,
+      "loss": 0.4252,
+      "step": 3366
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0283266659051338e-08,
+      "loss": 0.4108,
+      "step": 3367
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.937192406010535e-08,
+      "loss": 0.4173,
+      "step": 3368
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8481518490729433e-08,
+      "loss": 0.4119,
+      "step": 3369
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7612050883950572e-08,
+      "loss": 0.4144,
+      "step": 3370
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6763522150855526e-08,
+      "loss": 0.4151,
+      "step": 3371
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5935933180586395e-08,
+      "loss": 0.4181,
+      "step": 3372
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5129284840347257e-08,
+      "loss": 0.418,
+      "step": 3373
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4343577975397538e-08,
+      "loss": 0.41,
+      "step": 3374
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3578813409054204e-08,
+      "loss": 0.4155,
+      "step": 3375
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2834991942687337e-08,
+      "loss": 0.4093,
+      "step": 3376
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2112114355722348e-08,
+      "loss": 0.43,
+      "step": 3377
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1410181405639986e-08,
+      "loss": 0.4102,
+      "step": 3378
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0729193827969663e-08,
+      "loss": 0.4334,
+      "step": 3379
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.006915233629835e-08,
+      "loss": 0.4012,
+      "step": 3380
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.430057622257237e-09,
+      "loss": 0.4011,
+      "step": 3381
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.811910355537301e-09,
+      "loss": 0.4233,
+      "step": 3382
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.214711183871515e-09,
+      "loss": 0.4066,
+      "step": 3383
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.638460733043751e-09,
+      "loss": 0.4211,
+      "step": 3384
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.083159606890988e-09,
+      "loss": 0.4186,
+      "step": 3385
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.548808387292216e-09,
+      "loss": 0.4159,
+      "step": 3386
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.035407634175094e-09,
+      "loss": 0.4042,
+      "step": 3387
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.5429578855181745e-09,
+      "loss": 0.4189,
+      "step": 3388
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.071459657339794e-09,
+      "loss": 0.4134,
+      "step": 3389
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.620913443709185e-09,
+      "loss": 0.4207,
+      "step": 3390
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.1913197167375856e-09,
+      "loss": 0.4077,
+      "step": 3391
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.782678926582683e-09,
+      "loss": 0.4278,
+      "step": 3392
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3949915014463987e-09,
+      "loss": 0.4013,
+      "step": 3393
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.028257847570437e-09,
+      "loss": 0.4122,
+      "step": 3394
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6824783492451765e-09,
+      "loss": 0.4292,
+      "step": 3395
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.357653368800783e-09,
+      "loss": 0.4079,
+      "step": 3396
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.053783246609431e-09,
+      "loss": 0.4141,
+      "step": 3397
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7708683010853046e-09,
+      "loss": 0.4162,
+      "step": 3398
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5089088286890375e-09,
+      "loss": 0.4122,
+      "step": 3399
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2679051039188317e-09,
+      "loss": 0.4337,
+      "step": 3400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.047857379312678e-09,
+      "loss": 0.4214,
+      "step": 3401
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.487658854505753e-10,
+      "loss": 0.4268,
+      "step": 3402
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.706308309545329e-10,
+      "loss": 0.4143,
+      "step": 3403
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.134524024863474e-10,
+      "loss": 0.4135,
+      "step": 3404
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.772307647498252e-10,
+      "loss": 0.4154,
+      "step": 3405
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.619660604841201e-10,
+      "loss": 0.4067,
+      "step": 3406
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6765841047261532e-10,
+      "loss": 0.4096,
+      "step": 3407
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.430791353626234e-11,
+      "loss": 0.4135,
+      "step": 3408
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1914646538021574e-11,
+      "loss": 0.4222,
+      "step": 3409
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0478664380642045e-11,
+      "loss": 0.4143,
+      "step": 3410
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.4172,
+      "step": 3411
+    },
+    {
+      "epoch": 3.0,
+      "step": 3411,
+      "total_flos": 7.80466187204048e+23,
+      "train_loss": 0.1407828797818069,
+      "train_runtime": 19972.8893,
+      "train_samples_per_second": 87.508,
+      "train_steps_per_second": 0.171
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3411,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 7.80466187204048e+23,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}