diff --git "a/weights/model-base/trainer_state.json" "b/weights/model-base/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/weights/model-base/trainer_state.json"
@@ -0,0 +1,6019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.1333333333333333,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.7245,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.527570621338671e-07,
+      "loss": 2.1803,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5100842157794359e-06,
+      "loss": 2.6594,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9055141242677342e-06,
+      "loss": 2.2048,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.21223339017097e-06,
+      "loss": 2.5195,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6747272279066545e-06,
+      "loss": 2.0442,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.858271186401601e-06,
+      "loss": 2.2307,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0201684315588718e-06,
+      "loss": 2.1833,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1649904523048373e-06,
+      "loss": 2.1968,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.29599790562588e-06,
+      "loss": 1.8093,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4155983400471703e-06,
+      "loss": 1.707,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.525620074459582e-06,
+      "loss": 1.6357,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.627484290040521e-06,
+      "loss": 1.6422,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7223176059504063e-06,
+      "loss": 1.4509,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8110282485354684e-06,
+      "loss": 1.4365,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.972925493692739e-06,
+      "loss": 1.2501,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.047242937866133e-06,
+      "loss": 1.367,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.117747514438704e-06,
+      "loss": 1.2649,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.18481144368609e-06,
+      "loss": 1.2291,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2487549677597476e-06,
+      "loss": 1.3057,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.309855599633408e-06,
+      "loss": 1.2616,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.368355402181037e-06,
+      "loss": 1.3009,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.42446678034194e-06,
+      "loss": 1.0197,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.478377136593449e-06,
+      "loss": 1.3199,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.580241352174388e-06,
+      "loss": 1.1128,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.628475700819906e-06,
+      "loss": 1.1533,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.675074668084273e-06,
+      "loss": 1.1834,
+      "step": 27
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.720145521918643e-06,
+      "loss": 1.0839,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.763785310669335e-06,
+      "loss": 1.1704,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.806082121405316e-06,
+      "loss": 1.1539,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.84711615034489e-06,
+      "loss": 1.0536,
+      "step": 31
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886960618077624e-06,
+      "loss": 1.0446,
+      "step": 32
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9256825558266055e-06,
+      "loss": 1.0363,
+      "step": 33
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9633434839600245e-06,
+      "loss": 0.9855,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5e-06,
+      "loss": 1.0568,
+      "step": 35
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5e-06,
+      "loss": 1.0199,
+      "step": 36
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5e-06,
+      "loss": 1.0988,
+      "step": 37
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5e-06,
+      "loss": 1.0265,
+      "step": 38
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5e-06,
+      "loss": 0.9715,
+      "step": 39
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5e-06,
+      "loss": 0.9809,
+      "step": 40
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5e-06,
+      "loss": 1.0484,
+      "step": 41
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5e-06,
+      "loss": 1.0399,
+      "step": 42
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5e-06,
+      "loss": 1.0575,
+      "step": 43
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5e-06,
+      "loss": 1.1035,
+      "step": 44
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 0.9413,
+      "step": 45
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 1.0296,
+      "step": 46
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 1.0274,
+      "step": 47
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 0.9839,
+      "step": 48
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 0.984,
+      "step": 49
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5e-06,
+      "loss": 0.964,
+      "step": 50
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5e-06,
+      "loss": 0.8514,
+      "step": 51
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5e-06,
+      "loss": 0.9202,
+      "step": 52
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5e-06,
+      "loss": 0.9631,
+      "step": 53
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5e-06,
+      "loss": 0.8791,
+      "step": 54
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5e-06,
+      "loss": 0.9179,
+      "step": 55
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5e-06,
+      "loss": 0.9444,
+      "step": 56
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5e-06,
+      "loss": 0.9955,
+      "step": 57
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5e-06,
+      "loss": 0.7711,
+      "step": 58
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-06,
+      "loss": 0.908,
+      "step": 59
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-06,
+      "loss": 0.9389,
+      "step": 60
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-06,
+      "loss": 0.9576,
+      "step": 61
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-06,
+      "loss": 0.8092,
+      "step": 62
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-06,
+      "loss": 1.0091,
+      "step": 63
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-06,
+      "loss": 0.8951,
+      "step": 64
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-06,
+      "loss": 0.971,
+      "step": 65
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-06,
+      "loss": 0.9817,
+      "step": 66
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-06,
+      "loss": 0.8937,
+      "step": 67
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-06,
+      "loss": 0.9479,
+      "step": 68
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-06,
+      "loss": 0.9326,
+      "step": 69
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-06,
+      "loss": 0.8633,
+      "step": 70
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-06,
+      "loss": 0.9969,
+      "step": 71
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-06,
+      "loss": 0.7864,
+      "step": 72
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 0.8292,
+      "step": 73
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 0.848,
+      "step": 74
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 0.8667,
+      "step": 75
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 0.9081,
+      "step": 76
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 0.8571,
+      "step": 77
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5e-06,
+      "loss": 0.8555,
+      "step": 78
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5e-06,
+      "loss": 1.0347,
+      "step": 79
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5e-06,
+      "loss": 0.8916,
+      "step": 80
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5e-06,
+      "loss": 0.9043,
+      "step": 81
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5e-06,
+      "loss": 0.9539,
+      "step": 82
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5e-06,
+      "loss": 0.834,
+      "step": 83
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5e-06,
+      "loss": 0.9133,
+      "step": 84
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5e-06,
+      "loss": 0.8902,
+      "step": 85
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5e-06,
+      "loss": 0.9883,
+      "step": 86
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5e-06,
+      "loss": 0.8168,
+      "step": 87
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5e-06,
+      "loss": 0.887,
+      "step": 88
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5e-06,
+      "loss": 0.8416,
+      "step": 89
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5e-06,
+      "loss": 0.8718,
+      "step": 90
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5e-06,
+      "loss": 0.9709,
+      "step": 91
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-06,
+      "loss": 0.8788,
+      "step": 92
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-06,
+      "loss": 0.8702,
+      "step": 93
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-06,
+      "loss": 0.8478,
+      "step": 94
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-06,
+      "loss": 0.8489,
+      "step": 95
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-06,
+      "loss": 0.9661,
+      "step": 96
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5e-06,
+      "loss": 0.878,
+      "step": 97
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5e-06,
+      "loss": 0.8501,
+      "step": 98
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5e-06,
+      "loss": 0.8409,
+      "step": 99
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5e-06,
+      "loss": 0.969,
+      "step": 100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5e-06,
+      "loss": 0.9025,
+      "step": 101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5e-06,
+      "loss": 0.9762,
+      "step": 102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5e-06,
+      "loss": 0.911,
+      "step": 103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5e-06,
+      "loss": 0.9844,
+      "step": 104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5e-06,
+      "loss": 0.8977,
+      "step": 105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5e-06,
+      "loss": 0.8984,
+      "step": 106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5e-06,
+      "loss": 0.8466,
+      "step": 107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5e-06,
+      "loss": 0.8538,
+      "step": 108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5e-06,
+      "loss": 0.8805,
+      "step": 109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5e-06,
+      "loss": 0.8249,
+      "step": 110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5e-06,
+      "loss": 0.7971,
+      "step": 111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5e-06,
+      "loss": 0.8538,
+      "step": 112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5e-06,
+      "loss": 0.904,
+      "step": 113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5e-06,
+      "loss": 0.9772,
+      "step": 114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5e-06,
+      "loss": 0.8674,
+      "step": 115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5e-06,
+      "loss": 0.8733,
+      "step": 116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5e-06,
+      "loss": 0.874,
+      "step": 117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5e-06,
+      "loss": 0.944,
+      "step": 118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5e-06,
+      "loss": 0.9016,
+      "step": 119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5e-06,
+      "loss": 0.8858,
+      "step": 120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5e-06,
+      "loss": 0.8547,
+      "step": 121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5e-06,
+      "loss": 0.8012,
+      "step": 122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5e-06,
+      "loss": 0.7712,
+      "step": 123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5e-06,
+      "loss": 0.8255,
+      "step": 124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5e-06,
+      "loss": 0.8307,
+      "step": 125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5e-06,
+      "loss": 0.8892,
+      "step": 126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5e-06,
+      "loss": 0.899,
+      "step": 127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5e-06,
+      "loss": 0.8566,
+      "step": 128
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5e-06,
+      "loss": 0.807,
+      "step": 129
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5e-06,
+      "loss": 0.8177,
+      "step": 130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5e-06,
+      "loss": 0.8705,
+      "step": 131
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5e-06,
+      "loss": 0.7446,
+      "step": 132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5e-06,
+      "loss": 0.8639,
+      "step": 133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5e-06,
+      "loss": 0.8176,
+      "step": 134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5e-06,
+      "loss": 0.8843,
+      "step": 135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5e-06,
+      "loss": 0.8807,
+      "step": 136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5e-06,
+      "loss": 0.8815,
+      "step": 137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5e-06,
+      "loss": 0.8774,
+      "step": 138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5e-06,
+      "loss": 0.8063,
+      "step": 139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5e-06,
+      "loss": 0.8184,
+      "step": 140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5e-06,
+      "loss": 0.7814,
+      "step": 141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5e-06,
+      "loss": 0.9093,
+      "step": 142
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5e-06,
+      "loss": 0.8119,
+      "step": 143
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5e-06,
+      "loss": 0.8116,
+      "step": 144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5e-06,
+      "loss": 0.8736,
+      "step": 145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5e-06,
+      "loss": 0.8226,
+      "step": 146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5e-06,
+      "loss": 0.8339,
+      "step": 147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5e-06,
+      "loss": 0.7903,
+      "step": 148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5e-06,
+      "loss": 0.8177,
+      "step": 149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5e-06,
+      "loss": 0.8505,
+      "step": 150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5e-06,
+      "loss": 0.8066,
+      "step": 151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5e-06,
+      "loss": 0.8478,
+      "step": 152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.8183,
+      "step": 153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.8489,
+      "step": 154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.8301,
+      "step": 155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.8546,
+      "step": 156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.7771,
+      "step": 157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5e-06,
+      "loss": 0.808,
+      "step": 158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5e-06,
+      "loss": 0.896,
+      "step": 159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5e-06,
+      "loss": 0.8946,
+      "step": 160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5e-06,
+      "loss": 0.9003,
+      "step": 161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5e-06,
+      "loss": 0.8551,
+      "step": 162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5e-06,
+      "loss": 0.8998,
+      "step": 163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5e-06,
+      "loss": 0.8363,
+      "step": 164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5e-06,
+      "loss": 0.8494,
+      "step": 165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5e-06,
+      "loss": 0.8651,
+      "step": 166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5e-06,
+      "loss": 0.816,
+      "step": 167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5e-06,
+      "loss": 0.8236,
+      "step": 168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5e-06,
+      "loss": 0.8982,
+      "step": 169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5e-06,
+      "loss": 0.8074,
+      "step": 170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5e-06,
+      "loss": 0.8012,
+      "step": 171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5e-06,
+      "loss": 0.8539,
+      "step": 172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5e-06,
+      "loss": 0.8164,
+      "step": 173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5e-06,
+      "loss": 0.8505,
+      "step": 174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5e-06,
+      "loss": 0.8541,
+      "step": 175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5e-06,
+      "loss": 0.9559,
+      "step": 176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5e-06,
+      "loss": 0.7716,
+      "step": 177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5e-06,
+      "loss": 0.781,
+      "step": 178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5e-06,
+      "loss": 0.8176,
+      "step": 179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5e-06,
+      "loss": 0.8437,
+      "step": 180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5e-06,
+      "loss": 0.7996,
+      "step": 181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5e-06,
+      "loss": 0.7764,
+      "step": 182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5e-06,
+      "loss": 0.8984,
+      "step": 183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5e-06,
+      "loss": 0.9123,
+      "step": 184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5e-06,
+      "loss": 0.8965,
+      "step": 185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5e-06,
+      "loss": 0.825,
+      "step": 186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5e-06,
+      "loss": 0.9136,
+      "step": 187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5e-06,
+      "loss": 0.8742,
+      "step": 188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5e-06,
+      "loss": 0.8452,
+      "step": 189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5e-06,
+      "loss": 0.8612,
+      "step": 190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5e-06,
+      "loss": 0.8185,
+      "step": 191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5e-06,
+      "loss": 0.8671,
+      "step": 192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5e-06,
+      "loss": 0.8548,
+      "step": 193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5e-06,
+      "loss": 0.8687,
+      "step": 194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5e-06,
+      "loss": 0.8321,
+      "step": 195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5e-06,
+      "loss": 0.7973,
+      "step": 196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5e-06,
+      "loss": 0.8885,
+      "step": 197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5e-06,
+      "loss": 0.8577,
+      "step": 198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5e-06,
+      "loss": 0.8114,
+      "step": 199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5e-06,
+      "loss": 0.7769,
+      "step": 200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5e-06,
+      "loss": 0.8875,
+      "step": 201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5e-06,
+      "loss": 0.7767,
+      "step": 202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5e-06,
+      "loss": 0.8299,
+      "step": 203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 0.8233,
+      "step": 204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 0.8327,
+      "step": 205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 0.8867,
+      "step": 206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 0.7686,
+      "step": 207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 0.7699,
+      "step": 208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-06,
+      "loss": 0.7803,
+      "step": 209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-06,
+      "loss": 0.8683,
+      "step": 210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-06,
+      "loss": 0.768,
+      "step": 211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-06,
+      "loss": 0.9081,
+      "step": 212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-06,
+      "loss": 0.881,
+      "step": 213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5e-06,
+      "loss": 0.8392,
+      "step": 214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5e-06,
+      "loss": 0.8253,
+      "step": 215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5e-06,
+      "loss": 0.7927,
+      "step": 216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5e-06,
+      "loss": 0.7847,
+      "step": 217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5e-06,
+      "loss": 0.7971,
+      "step": 218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5e-06,
+      "loss": 0.7791,
+      "step": 219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5e-06,
+      "loss": 0.9004,
+      "step": 220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5e-06,
+      "loss": 0.7701,
+      "step": 221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5e-06,
+      "loss": 0.851,
+      "step": 222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5e-06,
+      "loss": 0.8141,
+      "step": 223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5e-06,
+      "loss": 0.8577,
+      "step": 224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5e-06,
+      "loss": 0.8316,
+      "step": 225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5e-06,
+      "loss": 0.8024,
+      "step": 226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5e-06,
+      "loss": 0.8621,
+      "step": 227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5e-06,
+      "loss": 0.8171,
+      "step": 228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5e-06,
+      "loss": 0.9156,
+      "step": 229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5e-06,
+      "loss": 0.7921,
+      "step": 230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5e-06,
+      "loss": 0.8667,
+      "step": 231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5e-06,
+      "loss": 0.7824,
+      "step": 232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5e-06,
+      "loss": 0.7559,
+      "step": 233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5e-06,
+      "loss": 0.8283,
+      "step": 234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5e-06,
+      "loss": 0.8411,
+      "step": 235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5e-06,
+      "loss": 0.7826,
+      "step": 236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 0.7758,
+      "step": 237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 0.7793,
+      "step": 238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 0.821,
+      "step": 239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 0.8918,
+      "step": 240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 0.7602,
+      "step": 241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.7801,
+      "step": 242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.8424,
+      "step": 243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.8459,
+      "step": 244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.8942,
+      "step": 245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.8498,
+      "step": 246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 0.8518,
+      "step": 247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 0.8608,
+      "step": 248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 0.7739,
+      "step": 249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 0.801,
+      "step": 250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5e-06,
+      "loss": 0.7802,
+      "step": 251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5e-06,
+      "loss": 0.8428,
+      "step": 252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5e-06,
+      "loss": 0.8491,
+      "step": 253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5e-06,
+      "loss": 0.8621,
+      "step": 254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5e-06,
+      "loss": 0.8286,
+      "step": 255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.7941,
+      "step": 256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.8012,
+      "step": 257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.8377,
+      "step": 258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.8161,
+      "step": 259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.8678,
+      "step": 260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5e-06,
+      "loss": 0.8201,
+      "step": 261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5e-06,
+      "loss": 0.7687,
+      "step": 262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5e-06,
+      "loss": 0.8941,
+      "step": 263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5e-06,
+      "loss": 0.8022,
+      "step": 264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7848,
+      "step": 265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7773,
+      "step": 266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7948,
+      "step": 267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7199,
+      "step": 268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 0.749,
+      "step": 269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5e-06,
+      "loss": 0.7476,
+      "step": 270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5e-06,
+      "loss": 0.8641,
+      "step": 271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5e-06,
+      "loss": 0.7909,
+      "step": 272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5e-06,
+      "loss": 0.7109,
+      "step": 273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5e-06,
+      "loss": 0.7805,
+      "step": 274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5e-06,
+      "loss": 0.7702,
+      "step": 275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5e-06,
+      "loss": 0.7731,
+      "step": 276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5e-06,
+      "loss": 0.7864,
+      "step": 277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5e-06,
+      "loss": 0.8503,
+      "step": 278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5e-06,
+      "loss": 0.7806,
+      "step": 279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5e-06,
+      "loss": 0.7456,
+      "step": 280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5e-06,
+      "loss": 0.796,
+      "step": 281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5e-06,
+      "loss": 0.8074,
+      "step": 282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5e-06,
+      "loss": 0.7386,
+      "step": 283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-06,
+      "loss": 0.8755,
+      "step": 284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-06,
+      "loss": 0.778,
+      "step": 285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-06,
+      "loss": 0.8379,
+      "step": 286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-06,
+      "loss": 0.8333,
+      "step": 287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-06,
+      "loss": 0.8542,
+      "step": 288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5e-06,
+      "loss": 0.8313,
+      "step": 289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5e-06,
+      "loss": 0.8217,
+      "step": 290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5e-06,
+      "loss": 0.7527,
+      "step": 291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5e-06,
+      "loss": 0.8289,
+      "step": 292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5e-06,
+      "loss": 0.8188,
+      "step": 293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5e-06,
+      "loss": 0.788,
+      "step": 294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5e-06,
+      "loss": 0.8625,
+      "step": 295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5e-06,
+      "loss": 0.8349,
+      "step": 296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5e-06,
+      "loss": 0.7633,
+      "step": 297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 0.8316,
+      "step": 298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 0.8268,
+      "step": 299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 0.7911,
+      "step": 300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 0.8755,
+      "step": 301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 0.8246,
+      "step": 302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5e-06,
+      "loss": 0.8712,
+      "step": 303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5e-06,
+      "loss": 0.8298,
+      "step": 304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5e-06,
+      "loss": 0.6828,
+      "step": 305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5e-06,
+      "loss": 0.8095,
+      "step": 306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5e-06,
+      "loss": 0.7864,
+      "step": 307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5e-06,
+      "loss": 0.9315,
+      "step": 308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5e-06,
+      "loss": 0.8837,
+      "step": 309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5e-06,
+      "loss": 0.8142,
+      "step": 310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5e-06,
+      "loss": 0.7912,
+      "step": 311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5e-06,
+      "loss": 0.7876,
+      "step": 312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5e-06,
+      "loss": 0.8222,
+      "step": 313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5e-06,
+      "loss": 0.8174,
+      "step": 314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5e-06,
+      "loss": 0.7928,
+      "step": 315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5e-06,
+      "loss": 0.8431,
+      "step": 316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7816,
+      "step": 317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7582,
+      "step": 318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7574,
+      "step": 319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7906,
+      "step": 320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7769,
+      "step": 321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5e-06,
+      "loss": 0.7402,
+      "step": 322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5e-06,
+      "loss": 0.766,
+      "step": 323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5e-06,
+      "loss": 0.7896,
+      "step": 324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5e-06,
+      "loss": 0.8332,
+      "step": 325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5e-06,
+      "loss": 0.8729,
+      "step": 326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5e-06,
+      "loss": 0.8006,
+      "step": 327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5e-06,
+      "loss": 0.7927,
+      "step": 328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5e-06,
+      "loss": 0.8015,
+      "step": 329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5e-06,
+      "loss": 0.7774,
+      "step": 330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5e-06,
+      "loss": 0.8244,
+      "step": 331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5e-06,
+      "loss": 0.7924,
+      "step": 332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5e-06,
+      "loss": 0.8237,
+      "step": 333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5e-06,
+      "loss": 0.8378,
+      "step": 334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5e-06,
+      "loss": 0.8893,
+      "step": 335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5e-06,
+      "loss": 0.8416,
+      "step": 336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5e-06,
+      "loss": 0.7577,
+      "step": 337
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5e-06,
+      "loss": 0.8068,
+      "step": 338
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5e-06,
+      "loss": 0.7592,
+      "step": 339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5e-06,
+      "loss": 0.8397,
+      "step": 340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5e-06,
+      "loss": 0.8228,
+      "step": 341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5e-06,
+      "loss": 0.7523,
+      "step": 342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5e-06,
+      "loss": 0.7865,
+      "step": 343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5e-06,
+      "loss": 0.7838,
+      "step": 344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5e-06,
+      "loss": 0.7967,
+      "step": 345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5e-06,
+      "loss": 0.7741,
+      "step": 346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5e-06,
+      "loss": 0.7083,
+      "step": 347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5e-06,
+      "loss": 0.7266,
+      "step": 348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5e-06,
+      "loss": 0.7531,
+      "step": 349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5e-06,
+      "loss": 0.7957,
+      "step": 350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5e-06,
+      "loss": 0.7978,
+      "step": 351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5e-06,
+      "loss": 0.8337,
+      "step": 352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5e-06,
+      "loss": 0.8267,
+      "step": 353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5e-06,
+      "loss": 0.8636,
+      "step": 354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5e-06,
+      "loss": 0.8135,
+      "step": 355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5e-06,
+      "loss": 0.7759,
+      "step": 356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5e-06,
+      "loss": 0.8248,
+      "step": 357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5e-06,
+      "loss": 0.7732,
+      "step": 358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5e-06,
+      "loss": 0.8346,
+      "step": 359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5e-06,
+      "loss": 0.7553,
+      "step": 360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5e-06,
+      "loss": 0.846,
+      "step": 361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5e-06,
+      "loss": 0.8738,
+      "step": 362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5e-06,
+      "loss": 0.8929,
+      "step": 363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5e-06,
+      "loss": 0.7838,
+      "step": 364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5e-06,
+      "loss": 0.7575,
+      "step": 365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5e-06,
+      "loss": 0.8056,
+      "step": 366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5e-06,
+      "loss": 0.7482,
+      "step": 367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5e-06,
+      "loss": 0.8213,
+      "step": 368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5e-06,
+      "loss": 0.7903,
+      "step": 369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5e-06,
+      "loss": 0.753,
+      "step": 370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5e-06,
+      "loss": 0.7563,
+      "step": 371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5e-06,
+      "loss": 0.7936,
+      "step": 372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5e-06,
+      "loss": 0.8055,
+      "step": 373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5e-06,
+      "loss": 0.8705,
+      "step": 374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5e-06,
+      "loss": 0.8474,
+      "step": 375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5e-06,
+      "loss": 0.8214,
+      "step": 376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5e-06,
+      "loss": 0.7324,
+      "step": 377
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5e-06,
+      "loss": 0.8092,
+      "step": 378
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5e-06,
+      "loss": 0.8466,
+      "step": 379
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5e-06,
+      "loss": 0.7637,
+      "step": 380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5e-06,
+      "loss": 0.7973,
+      "step": 381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5e-06,
+      "loss": 0.744,
+      "step": 382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5e-06,
+      "loss": 0.7448,
+      "step": 383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5e-06,
+      "loss": 0.8724,
+      "step": 384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5e-06,
+      "loss": 0.7428,
+      "step": 385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5e-06,
+      "loss": 0.785,
+      "step": 386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5e-06,
+      "loss": 0.8281,
+      "step": 387
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5e-06,
+      "loss": 0.7321,
+      "step": 388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5e-06,
+      "loss": 0.8579,
+      "step": 389
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5e-06,
+      "loss": 0.7277,
+      "step": 390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5e-06,
+      "loss": 0.8282,
+      "step": 391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5e-06,
+      "loss": 0.768,
+      "step": 392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5e-06,
+      "loss": 0.7111,
+      "step": 393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5e-06,
+      "loss": 0.6998,
+      "step": 394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5e-06,
+      "loss": 0.7843,
+      "step": 395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5e-06,
+      "loss": 0.8226,
+      "step": 396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5e-06,
+      "loss": 0.7771,
+      "step": 397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5e-06,
+      "loss": 0.8156,
+      "step": 398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5e-06,
+      "loss": 0.8372,
+      "step": 399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5e-06,
+      "loss": 0.7327,
+      "step": 400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5e-06,
+      "loss": 0.7285,
+      "step": 401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5e-06,
+      "loss": 0.8282,
+      "step": 402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5e-06,
+      "loss": 0.7456,
+      "step": 403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5e-06,
+      "loss": 0.8208,
+      "step": 404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5e-06,
+      "loss": 0.7944,
+      "step": 405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5e-06,
+      "loss": 0.8016,
+      "step": 406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5e-06,
+      "loss": 0.8129,
+      "step": 407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5e-06,
+      "loss": 0.836,
+      "step": 408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5e-06,
+      "loss": 0.7911,
+      "step": 409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5e-06,
+      "loss": 0.7055,
+      "step": 410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5e-06,
+      "loss": 0.7654,
+      "step": 411
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5e-06,
+      "loss": 0.8321,
+      "step": 412
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5e-06,
+      "loss": 0.7992,
+      "step": 413
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5e-06,
+      "loss": 0.7052,
+      "step": 414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5e-06,
+      "loss": 0.7529,
+      "step": 415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5e-06,
+      "loss": 0.8075,
+      "step": 416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5e-06,
+      "loss": 0.7658,
+      "step": 417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5e-06,
+      "loss": 0.7378,
+      "step": 418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5e-06,
+      "loss": 0.8025,
+      "step": 419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.7537,
+      "step": 420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.7587,
+      "step": 421
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.6687,
+      "step": 422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.8163,
+      "step": 423
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.8552,
+      "step": 424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5e-06,
+      "loss": 0.8588,
+      "step": 425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5e-06,
+      "loss": 0.7736,
+      "step": 426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5e-06,
+      "loss": 0.7352,
+      "step": 427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5e-06,
+      "loss": 0.8178,
+      "step": 428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5e-06,
+      "loss": 0.7304,
+      "step": 429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5e-06,
+      "loss": 0.7813,
+      "step": 430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5e-06,
+      "loss": 0.7663,
+      "step": 431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5e-06,
+      "loss": 0.7887,
+      "step": 432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5e-06,
+      "loss": 0.7523,
+      "step": 433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5e-06,
+      "loss": 0.8281,
+      "step": 434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5e-06,
+      "loss": 0.7809,
+      "step": 435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5e-06,
+      "loss": 0.7834,
+      "step": 436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5e-06,
+      "loss": 0.6889,
+      "step": 437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5e-06,
+      "loss": 0.7652,
+      "step": 438
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5e-06,
+      "loss": 0.7642,
+      "step": 439
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5e-06,
+      "loss": 0.7627,
+      "step": 440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5e-06,
+      "loss": 0.7787,
+      "step": 441
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5e-06,
+      "loss": 0.7556,
+      "step": 442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5e-06,
+      "loss": 0.8139,
+      "step": 443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5e-06,
+      "loss": 0.8316,
+      "step": 444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5e-06,
+      "loss": 0.8217,
+      "step": 445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5e-06,
+      "loss": 0.8204,
+      "step": 446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5e-06,
+      "loss": 0.7111,
+      "step": 447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5e-06,
+      "loss": 0.7921,
+      "step": 448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5e-06,
+      "loss": 0.8636,
+      "step": 449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5e-06,
+      "loss": 0.8228,
+      "step": 450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5e-06,
+      "loss": 0.7645,
+      "step": 451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5e-06,
+      "loss": 0.8344,
+      "step": 452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5e-06,
+      "loss": 0.7324,
+      "step": 453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5e-06,
+      "loss": 0.7615,
+      "step": 454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5e-06,
+      "loss": 0.7093,
+      "step": 455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5e-06,
+      "loss": 0.8116,
+      "step": 456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5e-06,
+      "loss": 0.8422,
+      "step": 457
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5e-06,
+      "loss": 0.869,
+      "step": 458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5e-06,
+      "loss": 0.8294,
+      "step": 459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5e-06,
+      "loss": 0.8462,
+      "step": 460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5e-06,
+      "loss": 0.8183,
+      "step": 461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5e-06,
+      "loss": 0.7341,
+      "step": 462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5e-06,
+      "loss": 0.8502,
+      "step": 463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5e-06,
+      "loss": 0.7854,
+      "step": 464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5e-06,
+      "loss": 0.8084,
+      "step": 465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5e-06,
+      "loss": 0.8429,
+      "step": 466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 0.7495,
+      "step": 467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 0.8141,
+      "step": 468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 0.7564,
+      "step": 469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 0.7033,
+      "step": 470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6497,
+      "step": 471
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5e-06,
+      "loss": 0.6674,
+      "step": 472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5e-06,
+      "loss": 0.7164,
+      "step": 473
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5e-06,
+      "loss": 0.6981,
+      "step": 474
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5e-06,
+      "loss": 0.7134,
+      "step": 475
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5e-06,
+      "loss": 0.7195,
+      "step": 476
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5e-06,
+      "loss": 0.6963,
+      "step": 477
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5e-06,
+      "loss": 0.6705,
+      "step": 478
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5e-06,
+      "loss": 0.7127,
+      "step": 479
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5e-06,
+      "loss": 0.6365,
+      "step": 480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6688,
+      "step": 481
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6255,
+      "step": 482
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6971,
+      "step": 483
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6423,
+      "step": 484
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6201,
+      "step": 485
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5e-06,
+      "loss": 0.6894,
+      "step": 486
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5e-06,
+      "loss": 0.7204,
+      "step": 487
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5e-06,
+      "loss": 0.6654,
+      "step": 488
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5e-06,
+      "loss": 0.5703,
+      "step": 489
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5e-06,
+      "loss": 0.6826,
+      "step": 490
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5e-06,
+      "loss": 0.6701,
+      "step": 491
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5e-06,
+      "loss": 0.6855,
+      "step": 492
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5e-06,
+      "loss": 0.7323,
+      "step": 493
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5e-06,
+      "loss": 0.6687,
+      "step": 494
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5e-06,
+      "loss": 0.7448,
+      "step": 495
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5e-06,
+      "loss": 0.6393,
+      "step": 496
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5e-06,
+      "loss": 0.6808,
+      "step": 497
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5e-06,
+      "loss": 0.641,
+      "step": 498
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5e-06,
+      "loss": 0.6673,
+      "step": 499
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5e-06,
+      "loss": 0.7313,
+      "step": 500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5e-06,
+      "loss": 0.6547,
+      "step": 501
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5e-06,
+      "loss": 0.655,
+      "step": 502
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5e-06,
+      "loss": 0.5551,
+      "step": 503
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5e-06,
+      "loss": 0.6753,
+      "step": 504
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5e-06,
+      "loss": 0.7242,
+      "step": 505
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5e-06,
+      "loss": 0.6142,
+      "step": 506
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5e-06,
+      "loss": 0.648,
+      "step": 507
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5e-06,
+      "loss": 0.6885,
+      "step": 508
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5e-06,
+      "loss": 0.7129,
+      "step": 509
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5e-06,
+      "loss": 0.6982,
+      "step": 510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5e-06,
+      "loss": 0.6454,
+      "step": 511
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5e-06,
+      "loss": 0.6995,
+      "step": 512
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5e-06,
+      "loss": 0.7527,
+      "step": 513
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5e-06,
+      "loss": 0.6924,
+      "step": 514
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5e-06,
+      "loss": 0.6735,
+      "step": 515
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5e-06,
+      "loss": 0.6562,
+      "step": 516
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5e-06,
+      "loss": 0.6838,
+      "step": 517
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5e-06,
+      "loss": 0.6035,
+      "step": 518
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5e-06,
+      "loss": 0.6733,
+      "step": 519
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5e-06,
+      "loss": 0.6837,
+      "step": 520
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5e-06,
+      "loss": 0.6671,
+      "step": 521
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5e-06,
+      "loss": 0.6022,
+      "step": 522
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5e-06,
+      "loss": 0.667,
+      "step": 523
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5e-06,
+      "loss": 0.6588,
+      "step": 524
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5e-06,
+      "loss": 0.7055,
+      "step": 525
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5e-06,
+      "loss": 0.6032,
+      "step": 526
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5e-06,
+      "loss": 0.6499,
+      "step": 527
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5e-06,
+      "loss": 0.7225,
+      "step": 528
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5e-06,
+      "loss": 0.6977,
+      "step": 529
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5e-06,
+      "loss": 0.6654,
+      "step": 530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5e-06,
+      "loss": 0.6506,
+      "step": 531
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5e-06,
+      "loss": 0.6521,
+      "step": 532
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5e-06,
+      "loss": 0.6549,
+      "step": 533
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5e-06,
+      "loss": 0.69,
+      "step": 534
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5e-06,
+      "loss": 0.5976,
+      "step": 535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5e-06,
+      "loss": 0.6404,
+      "step": 536
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5e-06,
+      "loss": 0.62,
+      "step": 537
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5e-06,
+      "loss": 0.6177,
+      "step": 538
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5e-06,
+      "loss": 0.6351,
+      "step": 539
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5e-06,
+      "loss": 0.6842,
+      "step": 540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5e-06,
+      "loss": 0.6395,
+      "step": 541
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5e-06,
+      "loss": 0.6735,
+      "step": 542
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5e-06,
+      "loss": 0.6535,
+      "step": 543
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5e-06,
+      "loss": 0.6653,
+      "step": 544
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5e-06,
+      "loss": 0.6591,
+      "step": 545
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5e-06,
+      "loss": 0.6358,
+      "step": 546
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5e-06,
+      "loss": 0.6487,
+      "step": 547
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5e-06,
+      "loss": 0.6429,
+      "step": 548
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5e-06,
+      "loss": 0.6356,
+      "step": 549
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5e-06,
+      "loss": 0.6677,
+      "step": 550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5e-06,
+      "loss": 0.5777,
+      "step": 551
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5e-06,
+      "loss": 0.7143,
+      "step": 552
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5e-06,
+      "loss": 0.6865,
+      "step": 553
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5e-06,
+      "loss": 0.5947,
+      "step": 554
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5e-06,
+      "loss": 0.6978,
+      "step": 555
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5e-06,
+      "loss": 0.6896,
+      "step": 556
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5e-06,
+      "loss": 0.6635,
+      "step": 557
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5e-06,
+      "loss": 0.6268,
+      "step": 558
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5e-06,
+      "loss": 0.6805,
+      "step": 559
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5e-06,
+      "loss": 0.6478,
+      "step": 560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5e-06,
+      "loss": 0.6484,
+      "step": 561
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5e-06,
+      "loss": 0.7277,
+      "step": 562
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5e-06,
+      "loss": 0.6331,
+      "step": 563
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5e-06,
+      "loss": 0.697,
+      "step": 564
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5e-06,
+      "loss": 0.642,
+      "step": 565
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5e-06,
+      "loss": 0.6726,
+      "step": 566
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5e-06,
+      "loss": 0.6104,
+      "step": 567
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5e-06,
+      "loss": 0.6138,
+      "step": 568
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5e-06,
+      "loss": 0.6973,
+      "step": 569
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5e-06,
+      "loss": 0.7026,
+      "step": 570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5e-06,
+      "loss": 0.6633,
+      "step": 571
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5e-06,
+      "loss": 0.6828,
+      "step": 572
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5e-06,
+      "loss": 0.681,
+      "step": 573
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5e-06,
+      "loss": 0.6458,
+      "step": 574
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5e-06,
+      "loss": 0.6628,
+      "step": 575
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5e-06,
+      "loss": 0.6173,
+      "step": 576
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5e-06,
+      "loss": 0.6867,
+      "step": 577
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5e-06,
+      "loss": 0.716,
+      "step": 578
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5e-06,
+      "loss": 0.6416,
+      "step": 579
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5e-06,
+      "loss": 0.6347,
+      "step": 580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5e-06,
+      "loss": 0.6404,
+      "step": 581
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5e-06,
+      "loss": 0.56,
+      "step": 582
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5e-06,
+      "loss": 0.6515,
+      "step": 583
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5e-06,
+      "loss": 0.5962,
+      "step": 584
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5e-06,
+      "loss": 0.6112,
+      "step": 585
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5e-06,
+      "loss": 0.6787,
+      "step": 586
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5e-06,
+      "loss": 0.6346,
+      "step": 587
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5e-06,
+      "loss": 0.6425,
+      "step": 588
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5e-06,
+      "loss": 0.606,
+      "step": 589
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5e-06,
+      "loss": 0.6786,
+      "step": 590
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5e-06,
+      "loss": 0.6199,
+      "step": 591
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5e-06,
+      "loss": 0.6363,
+      "step": 592
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5e-06,
+      "loss": 0.6251,
+      "step": 593
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5e-06,
+      "loss": 0.7168,
+      "step": 594
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5e-06,
+      "loss": 0.6583,
+      "step": 595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5e-06,
+      "loss": 0.6378,
+      "step": 596
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5e-06,
+      "loss": 0.6597,
+      "step": 597
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 0.6476,
+      "step": 598
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 0.6476,
+      "step": 599
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 0.6243,
+      "step": 600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 0.6337,
+      "step": 601
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 0.6993,
+      "step": 602
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5e-06,
+      "loss": 0.7012,
+      "step": 603
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5e-06,
+      "loss": 0.7175,
+      "step": 604
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5e-06,
+      "loss": 0.6601,
+      "step": 605
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5e-06,
+      "loss": 0.682,
+      "step": 606
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5e-06,
+      "loss": 0.7013,
+      "step": 607
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5e-06,
+      "loss": 0.5771,
+      "step": 608
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5e-06,
+      "loss": 0.6549,
+      "step": 609
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5e-06,
+      "loss": 0.6877,
+      "step": 610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5e-06,
+      "loss": 0.6741,
+      "step": 611
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5e-06,
+      "loss": 0.6715,
+      "step": 612
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5e-06,
+      "loss": 0.6248,
+      "step": 613
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5e-06,
+      "loss": 0.7372,
+      "step": 614
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5e-06,
+      "loss": 0.6817,
+      "step": 615
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5e-06,
+      "loss": 0.6559,
+      "step": 616
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5e-06,
+      "loss": 0.6401,
+      "step": 617
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5e-06,
+      "loss": 0.619,
+      "step": 618
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5e-06,
+      "loss": 0.6453,
+      "step": 619
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5e-06,
+      "loss": 0.6353,
+      "step": 620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5e-06,
+      "loss": 0.608,
+      "step": 621
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5e-06,
+      "loss": 0.6197,
+      "step": 622
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5e-06,
+      "loss": 0.6352,
+      "step": 623
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5e-06,
+      "loss": 0.6752,
+      "step": 624
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5e-06,
+      "loss": 0.7124,
+      "step": 625
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5e-06,
+      "loss": 0.6074,
+      "step": 626
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5e-06,
+      "loss": 0.5985,
+      "step": 627
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5e-06,
+      "loss": 0.6955,
+      "step": 628
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5e-06,
+      "loss": 0.6544,
+      "step": 629
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5e-06,
+      "loss": 0.6722,
+      "step": 630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5e-06,
+      "loss": 0.6422,
+      "step": 631
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5e-06,
+      "loss": 0.6611,
+      "step": 632
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5e-06,
+      "loss": 0.6698,
+      "step": 633
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5e-06,
+      "loss": 0.6299,
+      "step": 634
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5e-06,
+      "loss": 0.6641,
+      "step": 635
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5e-06,
+      "loss": 0.5547,
+      "step": 636
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5e-06,
+      "loss": 0.6695,
+      "step": 637
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5e-06,
+      "loss": 0.6563,
+      "step": 638
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5e-06,
+      "loss": 0.5838,
+      "step": 639
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5e-06,
+      "loss": 0.7056,
+      "step": 640
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5e-06,
+      "loss": 0.6545,
+      "step": 641
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5e-06,
+      "loss": 0.6671,
+      "step": 642
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5e-06,
+      "loss": 0.6916,
+      "step": 643
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5e-06,
+      "loss": 0.7028,
+      "step": 644
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5e-06,
+      "loss": 0.6374,
+      "step": 645
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5e-06,
+      "loss": 0.7019,
+      "step": 646
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5e-06,
+      "loss": 0.6289,
+      "step": 647
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5e-06,
+      "loss": 0.7065,
+      "step": 648
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5e-06,
+      "loss": 0.6515,
+      "step": 649
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5e-06,
+      "loss": 0.6206,
+      "step": 650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5e-06,
+      "loss": 0.6995,
+      "step": 651
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5e-06,
+      "loss": 0.6437,
+      "step": 652
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5e-06,
+      "loss": 0.7256,
+      "step": 653
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5e-06,
+      "loss": 0.6417,
+      "step": 654
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5e-06,
+      "loss": 0.6212,
+      "step": 655
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5e-06,
+      "loss": 0.679,
+      "step": 656
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5e-06,
+      "loss": 0.7172,
+      "step": 657
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5e-06,
+      "loss": 0.67,
+      "step": 658
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5e-06,
+      "loss": 0.7212,
+      "step": 659
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5e-06,
+      "loss": 0.6787,
+      "step": 660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5e-06,
+      "loss": 0.6826,
+      "step": 661
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5e-06,
+      "loss": 0.6982,
+      "step": 662
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5e-06,
+      "loss": 0.6329,
+      "step": 663
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5e-06,
+      "loss": 0.6791,
+      "step": 664
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5e-06,
+      "loss": 0.6931,
+      "step": 665
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5e-06,
+      "loss": 0.7053,
+      "step": 666
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5e-06,
+      "loss": 0.6428,
+      "step": 667
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5e-06,
+      "loss": 0.6975,
+      "step": 668
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5e-06,
+      "loss": 0.7014,
+      "step": 669
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5e-06,
+      "loss": 0.6394,
+      "step": 670
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5e-06,
+      "loss": 0.674,
+      "step": 671
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5e-06,
+      "loss": 0.7004,
+      "step": 672
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5e-06,
+      "loss": 0.6573,
+      "step": 673
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5e-06,
+      "loss": 0.6101,
+      "step": 674
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5e-06,
+      "loss": 0.5865,
+      "step": 675
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5e-06,
+      "loss": 0.6688,
+      "step": 676
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5e-06,
+      "loss": 0.7086,
+      "step": 677
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5e-06,
+      "loss": 0.6607,
+      "step": 678
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5e-06,
+      "loss": 0.6057,
+      "step": 679
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5e-06,
+      "loss": 0.6317,
+      "step": 680
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5e-06,
+      "loss": 0.6438,
+      "step": 681
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5e-06,
+      "loss": 0.6601,
+      "step": 682
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5e-06,
+      "loss": 0.5958,
+      "step": 683
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5e-06,
+      "loss": 0.7044,
+      "step": 684
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5e-06,
+      "loss": 0.6376,
+      "step": 685
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5e-06,
+      "loss": 0.7003,
+      "step": 686
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5e-06,
+      "loss": 0.6902,
+      "step": 687
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5e-06,
+      "loss": 0.6356,
+      "step": 688
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5e-06,
+      "loss": 0.6639,
+      "step": 689
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5e-06,
+      "loss": 0.6478,
+      "step": 690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5e-06,
+      "loss": 0.6927,
+      "step": 691
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5e-06,
+      "loss": 0.6603,
+      "step": 692
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5e-06,
+      "loss": 0.597,
+      "step": 693
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5e-06,
+      "loss": 0.7114,
+      "step": 694
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5e-06,
+      "loss": 0.6883,
+      "step": 695
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5e-06,
+      "loss": 0.6049,
+      "step": 696
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5e-06,
+      "loss": 0.6745,
+      "step": 697
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5e-06,
+      "loss": 0.7301,
+      "step": 698
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5e-06,
+      "loss": 0.6386,
+      "step": 699
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5e-06,
+      "loss": 0.687,
+      "step": 700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-06,
+      "loss": 0.6799,
+      "step": 701
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-06,
+      "loss": 0.7321,
+      "step": 702
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-06,
+      "loss": 0.6444,
+      "step": 703
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-06,
+      "loss": 0.7091,
+      "step": 704
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-06,
+      "loss": 0.7048,
+      "step": 705
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5e-06,
+      "loss": 0.6787,
+      "step": 706
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5e-06,
+      "loss": 0.7037,
+      "step": 707
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5e-06,
+      "loss": 0.704,
+      "step": 708
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5e-06,
+      "loss": 0.6792,
+      "step": 709
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5e-06,
+      "loss": 0.6202,
+      "step": 710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5e-06,
+      "loss": 0.707,
+      "step": 711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5e-06,
+      "loss": 0.6656,
+      "step": 712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5e-06,
+      "loss": 0.7435,
+      "step": 713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5e-06,
+      "loss": 0.6696,
+      "step": 714
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5e-06,
+      "loss": 0.6318,
+      "step": 715
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5e-06,
+      "loss": 0.6409,
+      "step": 716
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5e-06,
+      "loss": 0.6747,
+      "step": 717
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5e-06,
+      "loss": 0.6485,
+      "step": 718
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5e-06,
+      "loss": 0.6788,
+      "step": 719
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5e-06,
+      "loss": 0.6411,
+      "step": 720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5e-06,
+      "loss": 0.617,
+      "step": 721
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5e-06,
+      "loss": 0.648,
+      "step": 722
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5e-06,
+      "loss": 0.6382,
+      "step": 723
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5e-06,
+      "loss": 0.5992,
+      "step": 724
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5e-06,
+      "loss": 0.6955,
+      "step": 725
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5e-06,
+      "loss": 0.7319,
+      "step": 726
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5e-06,
+      "loss": 0.7246,
+      "step": 727
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5e-06,
+      "loss": 0.7442,
+      "step": 728
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5e-06,
+      "loss": 0.7032,
+      "step": 729
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5e-06,
+      "loss": 0.6187,
+      "step": 730
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5e-06,
+      "loss": 0.6505,
+      "step": 731
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5e-06,
+      "loss": 0.6716,
+      "step": 732
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5e-06,
+      "loss": 0.6385,
+      "step": 733
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5e-06,
+      "loss": 0.685,
+      "step": 734
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5e-06,
+      "loss": 0.6434,
+      "step": 735
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5e-06,
+      "loss": 0.6803,
+      "step": 736
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5e-06,
+      "loss": 0.6362,
+      "step": 737
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7028,
+      "step": 738
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5e-06,
+      "loss": 0.6736,
+      "step": 739
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5e-06,
+      "loss": 0.7004,
+      "step": 740
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5e-06,
+      "loss": 0.688,
+      "step": 741
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5e-06,
+      "loss": 0.6202,
+      "step": 742
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5e-06,
+      "loss": 0.6794,
+      "step": 743
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5e-06,
+      "loss": 0.6959,
+      "step": 744
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5e-06,
+      "loss": 0.6825,
+      "step": 745
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5e-06,
+      "loss": 0.6811,
+      "step": 746
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5e-06,
+      "loss": 0.6419,
+      "step": 747
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5e-06,
+      "loss": 0.6952,
+      "step": 748
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5e-06,
+      "loss": 0.5878,
+      "step": 749
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5e-06,
+      "loss": 0.6662,
+      "step": 750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5e-06,
+      "loss": 0.633,
+      "step": 751
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5e-06,
+      "loss": 0.6257,
+      "step": 752
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5e-06,
+      "loss": 0.6676,
+      "step": 753
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5e-06,
+      "loss": 0.6126,
+      "step": 754
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5e-06,
+      "loss": 0.7211,
+      "step": 755
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5e-06,
+      "loss": 0.6555,
+      "step": 756
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5e-06,
+      "loss": 0.6537,
+      "step": 757
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5e-06,
+      "loss": 0.7181,
+      "step": 758
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5e-06,
+      "loss": 0.6001,
+      "step": 759
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5e-06,
+      "loss": 0.6627,
+      "step": 760
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5e-06,
+      "loss": 0.6386,
+      "step": 761
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5e-06,
+      "loss": 0.613,
+      "step": 762
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5e-06,
+      "loss": 0.7076,
+      "step": 763
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5e-06,
+      "loss": 0.6191,
+      "step": 764
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5e-06,
+      "loss": 0.6729,
+      "step": 765
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5e-06,
+      "loss": 0.6373,
+      "step": 766
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-06,
+      "loss": 0.6052,
+      "step": 767
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-06,
+      "loss": 0.6561,
+      "step": 768
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-06,
+      "loss": 0.619,
+      "step": 769
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-06,
+      "loss": 0.7382,
+      "step": 770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-06,
+      "loss": 0.6352,
+      "step": 771
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5e-06,
+      "loss": 0.6813,
+      "step": 772
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5e-06,
+      "loss": 0.667,
+      "step": 773
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5e-06,
+      "loss": 0.7465,
+      "step": 774
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5e-06,
+      "loss": 0.6699,
+      "step": 775
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5e-06,
+      "loss": 0.6768,
+      "step": 776
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5e-06,
+      "loss": 0.6493,
+      "step": 777
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5e-06,
+      "loss": 0.6459,
+      "step": 778
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5e-06,
+      "loss": 0.6794,
+      "step": 779
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5e-06,
+      "loss": 0.6765,
+      "step": 780
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-06,
+      "loss": 0.654,
+      "step": 781
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-06,
+      "loss": 0.6116,
+      "step": 782
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-06,
+      "loss": 0.6778,
+      "step": 783
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-06,
+      "loss": 0.6864,
+      "step": 784
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-06,
+      "loss": 0.7155,
+      "step": 785
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5e-06,
+      "loss": 0.6075,
+      "step": 786
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7276,
+      "step": 787
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5e-06,
+      "loss": 0.6361,
+      "step": 788
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5e-06,
+      "loss": 0.6892,
+      "step": 789
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5e-06,
+      "loss": 0.6267,
+      "step": 790
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5e-06,
+      "loss": 0.7546,
+      "step": 791
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5e-06,
+      "loss": 0.6671,
+      "step": 792
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5e-06,
+      "loss": 0.6924,
+      "step": 793
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5e-06,
+      "loss": 0.7384,
+      "step": 794
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5e-06,
+      "loss": 0.6468,
+      "step": 795
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5e-06,
+      "loss": 0.6513,
+      "step": 796
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5e-06,
+      "loss": 0.6565,
+      "step": 797
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5e-06,
+      "loss": 0.7515,
+      "step": 798
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5e-06,
+      "loss": 0.6423,
+      "step": 799
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5e-06,
+      "loss": 0.6139,
+      "step": 800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5e-06,
+      "loss": 0.6632,
+      "step": 801
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5e-06,
+      "loss": 0.6741,
+      "step": 802
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5e-06,
+      "loss": 0.6358,
+      "step": 803
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5e-06,
+      "loss": 0.6303,
+      "step": 804
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5e-06,
+      "loss": 0.6295,
+      "step": 805
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5e-06,
+      "loss": 0.6687,
+      "step": 806
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5e-06,
+      "loss": 0.5648,
+      "step": 807
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5e-06,
+      "loss": 0.6721,
+      "step": 808
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5e-06,
+      "loss": 0.6527,
+      "step": 809
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5e-06,
+      "loss": 0.7354,
+      "step": 810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5e-06,
+      "loss": 0.6099,
+      "step": 811
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5e-06,
+      "loss": 0.7045,
+      "step": 812
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5e-06,
+      "loss": 0.661,
+      "step": 813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5e-06,
+      "loss": 0.6575,
+      "step": 814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5e-06,
+      "loss": 0.6982,
+      "step": 815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5e-06,
+      "loss": 0.6495,
+      "step": 816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5e-06,
+      "loss": 0.626,
+      "step": 817
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5e-06,
+      "loss": 0.6524,
+      "step": 818
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5e-06,
+      "loss": 0.6283,
+      "step": 819
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5e-06,
+      "loss": 0.7343,
+      "step": 820
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5e-06,
+      "loss": 0.6777,
+      "step": 821
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5e-06,
+      "loss": 0.6922,
+      "step": 822
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5e-06,
+      "loss": 0.6522,
+      "step": 823
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5e-06,
+      "loss": 0.5868,
+      "step": 824
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5e-06,
+      "loss": 0.6213,
+      "step": 825
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5e-06,
+      "loss": 0.5686,
+      "step": 826
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5e-06,
+      "loss": 0.6395,
+      "step": 827
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5e-06,
+      "loss": 0.583,
+      "step": 828
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5e-06,
+      "loss": 0.7285,
+      "step": 829
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5e-06,
+      "loss": 0.6323,
+      "step": 830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5e-06,
+      "loss": 0.6937,
+      "step": 831
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5e-06,
+      "loss": 0.6779,
+      "step": 832
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5e-06,
+      "loss": 0.6587,
+      "step": 833
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5e-06,
+      "loss": 0.6876,
+      "step": 834
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5e-06,
+      "loss": 0.624,
+      "step": 835
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5e-06,
+      "loss": 0.5833,
+      "step": 836
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5e-06,
+      "loss": 0.6241,
+      "step": 837
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5e-06,
+      "loss": 0.6034,
+      "step": 838
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5e-06,
+      "loss": 0.6813,
+      "step": 839
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5e-06,
+      "loss": 0.6554,
+      "step": 840
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5e-06,
+      "loss": 0.6334,
+      "step": 841
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5e-06,
+      "loss": 0.6501,
+      "step": 842
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5e-06,
+      "loss": 0.6495,
+      "step": 843
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5e-06,
+      "loss": 0.6817,
+      "step": 844
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5e-06,
+      "loss": 0.6359,
+      "step": 845
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5e-06,
+      "loss": 0.709,
+      "step": 846
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5e-06,
+      "loss": 0.7104,
+      "step": 847
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5e-06,
+      "loss": 0.6433,
+      "step": 848
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5e-06,
+      "loss": 0.7394,
+      "step": 849
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5e-06,
+      "loss": 0.6313,
+      "step": 850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5e-06,
+      "loss": 0.714,
+      "step": 851
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5e-06,
+      "loss": 0.6563,
+      "step": 852
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5e-06,
+      "loss": 0.6482,
+      "step": 853
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5e-06,
+      "loss": 0.6478,
+      "step": 854
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5e-06,
+      "loss": 0.681,
+      "step": 855
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5e-06,
+      "loss": 0.602,
+      "step": 856
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5e-06,
+      "loss": 0.6192,
+      "step": 857
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5e-06,
+      "loss": 0.6472,
+      "step": 858
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5e-06,
+      "loss": 0.6767,
+      "step": 859
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5e-06,
+      "loss": 0.6588,
+      "step": 860
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5e-06,
+      "loss": 0.665,
+      "step": 861
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5e-06,
+      "loss": 0.6688,
+      "step": 862
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5e-06,
+      "loss": 0.5669,
+      "step": 863
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5e-06,
+      "loss": 0.6826,
+      "step": 864
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5e-06,
+      "loss": 0.619,
+      "step": 865
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5e-06,
+      "loss": 0.6268,
+      "step": 866
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5e-06,
+      "loss": 0.6319,
+      "step": 867
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5e-06,
+      "loss": 0.6852,
+      "step": 868
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5e-06,
+      "loss": 0.6543,
+      "step": 869
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5e-06,
+      "loss": 0.5947,
+      "step": 870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5e-06,
+      "loss": 0.668,
+      "step": 871
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5e-06,
+      "loss": 0.6297,
+      "step": 872
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5e-06,
+      "loss": 0.5856,
+      "step": 873
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5e-06,
+      "loss": 0.6924,
+      "step": 874
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5e-06,
+      "loss": 0.6611,
+      "step": 875
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5e-06,
+      "loss": 0.6887,
+      "step": 876
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5e-06,
+      "loss": 0.6502,
+      "step": 877
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5e-06,
+      "loss": 0.6269,
+      "step": 878
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5e-06,
+      "loss": 0.6166,
+      "step": 879
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5e-06,
+      "loss": 0.7394,
+      "step": 880
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5e-06,
+      "loss": 0.6479,
+      "step": 881
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5e-06,
+      "loss": 0.6558,
+      "step": 882
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5e-06,
+      "loss": 0.6866,
+      "step": 883
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5e-06,
+      "loss": 0.6439,
+      "step": 884
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5e-06,
+      "loss": 0.6577,
+      "step": 885
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5e-06,
+      "loss": 0.7388,
+      "step": 886
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5e-06,
+      "loss": 0.7518,
+      "step": 887
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5e-06,
+      "loss": 0.6439,
+      "step": 888
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5e-06,
+      "loss": 0.6509,
+      "step": 889
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5e-06,
+      "loss": 0.658,
+      "step": 890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5e-06,
+      "loss": 0.6729,
+      "step": 891
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5e-06,
+      "loss": 0.6712,
+      "step": 892
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5e-06,
+      "loss": 0.7011,
+      "step": 893
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5e-06,
+      "loss": 0.6966,
+      "step": 894
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5e-06,
+      "loss": 0.6344,
+      "step": 895
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5e-06,
+      "loss": 0.6276,
+      "step": 896
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5e-06,
+      "loss": 0.684,
+      "step": 897
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5e-06,
+      "loss": 0.6803,
+      "step": 898
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5e-06,
+      "loss": 0.68,
+      "step": 899
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5e-06,
+      "loss": 0.6628,
+      "step": 900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5e-06,
+      "loss": 0.6333,
+      "step": 901
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5e-06,
+      "loss": 0.5684,
+      "step": 902
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5e-06,
+      "loss": 0.6721,
+      "step": 903
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5e-06,
+      "loss": 0.69,
+      "step": 904
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5e-06,
+      "loss": 0.6724,
+      "step": 905
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5e-06,
+      "loss": 0.6217,
+      "step": 906
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5e-06,
+      "loss": 0.715,
+      "step": 907
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5e-06,
+      "loss": 0.6323,
+      "step": 908
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5e-06,
+      "loss": 0.5879,
+      "step": 909
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5e-06,
+      "loss": 0.6679,
+      "step": 910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5e-06,
+      "loss": 0.6374,
+      "step": 911
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5e-06,
+      "loss": 0.6286,
+      "step": 912
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5e-06,
+      "loss": 0.5938,
+      "step": 913
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5e-06,
+      "loss": 0.6596,
+      "step": 914
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5e-06,
+      "loss": 0.7006,
+      "step": 915
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5e-06,
+      "loss": 0.6663,
+      "step": 916
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5e-06,
+      "loss": 0.6291,
+      "step": 917
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5e-06,
+      "loss": 0.6638,
+      "step": 918
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5e-06,
+      "loss": 0.6641,
+      "step": 919
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5e-06,
+      "loss": 0.6743,
+      "step": 920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5e-06,
+      "loss": 0.675,
+      "step": 921
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5e-06,
+      "loss": 0.6395,
+      "step": 922
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5e-06,
+      "loss": 0.6489,
+      "step": 923
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5e-06,
+      "loss": 0.6081,
+      "step": 924
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5e-06,
+      "loss": 0.6611,
+      "step": 925
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5e-06,
+      "loss": 0.6118,
+      "step": 926
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5e-06,
+      "loss": 0.5995,
+      "step": 927
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5e-06,
+      "loss": 0.6284,
+      "step": 928
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5e-06,
+      "loss": 0.5616,
+      "step": 929
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5e-06,
+      "loss": 0.6686,
+      "step": 930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5e-06,
+      "loss": 0.6133,
+      "step": 931
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5e-06,
+      "loss": 0.6482,
+      "step": 932
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5e-06,
+      "loss": 0.6703,
+      "step": 933
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5e-06,
+      "loss": 0.6462,
+      "step": 934
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5e-06,
+      "loss": 0.5937,
+      "step": 935
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5e-06,
+      "loss": 0.679,
+      "step": 936
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6658,
+      "step": 937
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6225,
+      "step": 938
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5e-06,
+      "loss": 0.5104,
+      "step": 939
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5e-06,
+      "loss": 0.4632,
+      "step": 940
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5e-06,
+      "loss": 0.4713,
+      "step": 941
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5e-06,
+      "loss": 0.4562,
+      "step": 942
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5e-06,
+      "loss": 0.5048,
+      "step": 943
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5e-06,
+      "loss": 0.4529,
+      "step": 944
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5e-06,
+      "loss": 0.5007,
+      "step": 945
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5e-06,
+      "loss": 0.4808,
+      "step": 946
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5e-06,
+      "loss": 0.4504,
+      "step": 947
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5e-06,
+      "loss": 0.4919,
+      "step": 948
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5e-06,
+      "loss": 0.4911,
+      "step": 949
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5e-06,
+      "loss": 0.4432,
+      "step": 950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5e-06,
+      "loss": 0.469,
+      "step": 951
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5e-06,
+      "loss": 0.4923,
+      "step": 952
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5e-06,
+      "loss": 0.4578,
+      "step": 953
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5e-06,
+      "loss": 0.4947,
+      "step": 954
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5e-06,
+      "loss": 0.4264,
+      "step": 955
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5e-06,
+      "loss": 0.4309,
+      "step": 956
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5e-06,
+      "loss": 0.4866,
+      "step": 957
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5e-06,
+      "loss": 0.493,
+      "step": 958
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5e-06,
+      "loss": 0.4002,
+      "step": 959
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5e-06,
+      "loss": 0.4199,
+      "step": 960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5e-06,
+      "loss": 0.4693,
+      "step": 961
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5e-06,
+      "loss": 0.4943,
+      "step": 962
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5e-06,
+      "loss": 0.497,
+      "step": 963
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5e-06,
+      "loss": 0.421,
+      "step": 964
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5e-06,
+      "loss": 0.4072,
+      "step": 965
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5e-06,
+      "loss": 0.4544,
+      "step": 966
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5e-06,
+      "loss": 0.4661,
+      "step": 967
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5e-06,
+      "loss": 0.4485,
+      "step": 968
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5e-06,
+      "loss": 0.438,
+      "step": 969
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5e-06,
+      "loss": 0.4142,
+      "step": 970
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5e-06,
+      "loss": 0.4532,
+      "step": 971
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5e-06,
+      "loss": 0.4182,
+      "step": 972
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5e-06,
+      "loss": 0.4515,
+      "step": 973
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5e-06,
+      "loss": 0.4963,
+      "step": 974
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5e-06,
+      "loss": 0.4475,
+      "step": 975
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5e-06,
+      "loss": 0.4708,
+      "step": 976
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5e-06,
+      "loss": 0.5019,
+      "step": 977
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5e-06,
+      "loss": 0.4483,
+      "step": 978
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5e-06,
+      "loss": 0.4666,
+      "step": 979
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5e-06,
+      "loss": 0.3958,
+      "step": 980
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5e-06,
+      "loss": 0.4056,
+      "step": 981
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5e-06,
+      "loss": 0.4519,
+      "step": 982
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5e-06,
+      "loss": 0.4274,
+      "step": 983
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5e-06,
+      "loss": 0.5147,
+      "step": 984
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5e-06,
+      "loss": 0.4513,
+      "step": 985
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5e-06,
+      "loss": 0.4441,
+      "step": 986
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5e-06,
+      "loss": 0.4566,
+      "step": 987
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5e-06,
+      "loss": 0.4696,
+      "step": 988
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5e-06,
+      "loss": 0.4535,
+      "step": 989
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5e-06,
+      "loss": 0.4788,
+      "step": 990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5e-06,
+      "loss": 0.4302,
+      "step": 991
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5e-06,
+      "loss": 0.4414,
+      "step": 992
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5e-06,
+      "loss": 0.4807,
+      "step": 993
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5e-06,
+      "loss": 0.4952,
+      "step": 994
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5e-06,
+      "loss": 0.4783,
+      "step": 995
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5e-06,
+      "loss": 0.4265,
+      "step": 996
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5e-06,
+      "loss": 0.4712,
+      "step": 997
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5e-06,
+      "loss": 0.4672,
+      "step": 998
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5e-06,
+      "loss": 0.4801,
+      "step": 999
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5e-06,
+      "loss": 0.4601,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3744,
+  "num_train_epochs": 8,
+  "save_steps": 500,
+  "total_flos": 5.75975864622121e+16,
+  "trial_name": null,
+  "trial_params": null
+}