diff --git "a/checkpoint-1550/trainer_state.json" "b/checkpoint-1550/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1550/trainer_state.json"
@@ -0,0 +1,9319 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1550,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.993548387096775e-05,
+      "loss": 1.716,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.987096774193549e-05,
+      "loss": 1.7745,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.980645161290323e-05,
+      "loss": 1.6054,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.974193548387097e-05,
+      "loss": 1.6363,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.967741935483872e-05,
+      "loss": 1.5976,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.961290322580645e-05,
+      "loss": 1.7749,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.95483870967742e-05,
+      "loss": 1.5595,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.948387096774195e-05,
+      "loss": 1.6096,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.941935483870968e-05,
+      "loss": 1.6574,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.935483870967742e-05,
+      "loss": 1.617,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.929032258064517e-05,
+      "loss": 1.7268,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.922580645161291e-05,
+      "loss": 1.5113,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.916129032258064e-05,
+      "loss": 1.7497,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.90967741935484e-05,
+      "loss": 1.5716,
+      "step": 14
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.903225806451614e-05,
+      "loss": 1.4849,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.896774193548386e-05,
+      "loss": 1.4985,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.890322580645162e-05,
+      "loss": 1.5288,
+      "step": 17
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.883870967741936e-05,
+      "loss": 1.5506,
+      "step": 18
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.87741935483871e-05,
+      "loss": 1.6588,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.870967741935484e-05,
+      "loss": 1.6245,
+      "step": 20
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.864516129032258e-05,
+      "loss": 1.4441,
+      "step": 21
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.858064516129033e-05,
+      "loss": 1.5697,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.851612903225807e-05,
+      "loss": 1.4353,
+      "step": 23
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845161290322581e-05,
+      "loss": 1.3954,
+      "step": 24
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838709677419355e-05,
+      "loss": 1.6193,
+      "step": 25
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83225806451613e-05,
+      "loss": 1.5966,
+      "step": 26
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825806451612903e-05,
+      "loss": 1.5386,
+      "step": 27
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819354838709677e-05,
+      "loss": 1.4932,
+      "step": 28
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812903225806453e-05,
+      "loss": 1.5357,
+      "step": 29
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.806451612903226e-05,
+      "loss": 1.5584,
+      "step": 30
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8e-05,
+      "loss": 1.5107,
+      "step": 31
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.793548387096775e-05,
+      "loss": 1.4353,
+      "step": 32
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.787096774193549e-05,
+      "loss": 1.4709,
+      "step": 33
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.780645161290322e-05,
+      "loss": 1.4193,
+      "step": 34
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.774193548387098e-05,
+      "loss": 1.3577,
+      "step": 35
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774193548387098e-05,
+      "loss": 1.7503,
+      "step": 36
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767741935483872e-05,
+      "loss": 1.652,
+      "step": 37
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.761290322580646e-05,
+      "loss": 1.577,
+      "step": 38
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75483870967742e-05,
+      "loss": 1.4753,
+      "step": 39
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748387096774194e-05,
+      "loss": 1.5025,
+      "step": 40
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741935483870968e-05,
+      "loss": 1.5508,
+      "step": 41
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.735483870967742e-05,
+      "loss": 1.3235,
+      "step": 42
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.729032258064516e-05,
+      "loss": 1.3986,
+      "step": 43
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.72258064516129e-05,
+      "loss": 1.4616,
+      "step": 44
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.716129032258065e-05,
+      "loss": 1.5728,
+      "step": 45
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.709677419354839e-05,
+      "loss": 1.5358,
+      "step": 46
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.703225806451613e-05,
+      "loss": 1.4407,
+      "step": 47
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.696774193548388e-05,
+      "loss": 1.3705,
+      "step": 48
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.690322580645161e-05,
+      "loss": 1.44,
+      "step": 49
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.683870967741935e-05,
+      "loss": 1.4533,
+      "step": 50
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.677419354838711e-05,
+      "loss": 1.5213,
+      "step": 51
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.670967741935484e-05,
+      "loss": 1.4516,
+      "step": 52
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.664516129032259e-05,
+      "loss": 1.3548,
+      "step": 53
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.658064516129033e-05,
+      "loss": 1.4199,
+      "step": 54
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.651612903225807e-05,
+      "loss": 1.4489,
+      "step": 55
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.645161290322581e-05,
+      "loss": 1.4327,
+      "step": 56
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.638709677419355e-05,
+      "loss": 1.4834,
+      "step": 57
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.63225806451613e-05,
+      "loss": 1.4057,
+      "step": 58
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.625806451612904e-05,
+      "loss": 1.4227,
+      "step": 59
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.619354838709678e-05,
+      "loss": 1.3265,
+      "step": 60
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.612903225806452e-05,
+      "loss": 1.4446,
+      "step": 61
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.606451612903226e-05,
+      "loss": 1.4619,
+      "step": 62
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.6e-05,
+      "loss": 1.3791,
+      "step": 63
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.593548387096774e-05,
+      "loss": 1.4303,
+      "step": 64
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.587096774193548e-05,
+      "loss": 1.4665,
+      "step": 65
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.580645161290323e-05,
+      "loss": 1.2667,
+      "step": 66
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.574193548387097e-05,
+      "loss": 1.4362,
+      "step": 67
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.567741935483872e-05,
+      "loss": 1.4245,
+      "step": 68
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.561290322580646e-05,
+      "loss": 1.3996,
+      "step": 69
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.554838709677419e-05,
+      "loss": 1.3897,
+      "step": 70
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.554838709677419e-05,
+      "loss": 1.3155,
+      "step": 71
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.548387096774195e-05,
+      "loss": 1.4341,
+      "step": 72
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.541935483870969e-05,
+      "loss": 1.3562,
+      "step": 73
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.535483870967741e-05,
+      "loss": 1.5022,
+      "step": 74
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.529032258064517e-05,
+      "loss": 1.4112,
+      "step": 75
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.522580645161291e-05,
+      "loss": 1.4515,
+      "step": 76
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.516129032258065e-05,
+      "loss": 1.2948,
+      "step": 77
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.50967741935484e-05,
+      "loss": 1.3857,
+      "step": 78
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.503225806451613e-05,
+      "loss": 1.3419,
+      "step": 79
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.496774193548388e-05,
+      "loss": 1.3807,
+      "step": 80
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.490322580645162e-05,
+      "loss": 1.3096,
+      "step": 81
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.483870967741936e-05,
+      "loss": 1.3366,
+      "step": 82
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.47741935483871e-05,
+      "loss": 1.3849,
+      "step": 83
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.470967741935484e-05,
+      "loss": 1.3757,
+      "step": 84
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.464516129032258e-05,
+      "loss": 1.4037,
+      "step": 85
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.458064516129032e-05,
+      "loss": 1.4029,
+      "step": 86
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.451612903225808e-05,
+      "loss": 1.2547,
+      "step": 87
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.44516129032258e-05,
+      "loss": 1.3634,
+      "step": 88
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.438709677419355e-05,
+      "loss": 1.339,
+      "step": 89
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.43225806451613e-05,
+      "loss": 1.4094,
+      "step": 90
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.425806451612904e-05,
+      "loss": 1.4527,
+      "step": 91
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.419354838709677e-05,
+      "loss": 1.4851,
+      "step": 92
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.412903225806453e-05,
+      "loss": 1.291,
+      "step": 93
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.406451612903227e-05,
+      "loss": 1.2541,
+      "step": 94
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.4e-05,
+      "loss": 1.4211,
+      "step": 95
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.393548387096775e-05,
+      "loss": 1.3906,
+      "step": 96
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.387096774193549e-05,
+      "loss": 1.2921,
+      "step": 97
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.380645161290323e-05,
+      "loss": 1.5347,
+      "step": 98
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.374193548387097e-05,
+      "loss": 1.2438,
+      "step": 99
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.367741935483871e-05,
+      "loss": 1.3963,
+      "step": 100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.361290322580646e-05,
+      "loss": 1.5133,
+      "step": 101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.35483870967742e-05,
+      "loss": 1.4138,
+      "step": 102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.348387096774194e-05,
+      "loss": 1.2888,
+      "step": 103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.341935483870968e-05,
+      "loss": 1.4008,
+      "step": 104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.335483870967743e-05,
+      "loss": 1.3859,
+      "step": 105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.329032258064516e-05,
+      "loss": 1.4192,
+      "step": 106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.32258064516129e-05,
+      "loss": 1.3406,
+      "step": 107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.316129032258066e-05,
+      "loss": 1.467,
+      "step": 108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.309677419354839e-05,
+      "loss": 1.2791,
+      "step": 109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.303225806451613e-05,
+      "loss": 1.3281,
+      "step": 110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.296774193548388e-05,
+      "loss": 1.4177,
+      "step": 111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.290322580645162e-05,
+      "loss": 1.3119,
+      "step": 112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.283870967741935e-05,
+      "loss": 1.5024,
+      "step": 113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.27741935483871e-05,
+      "loss": 1.3078,
+      "step": 114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.270967741935485e-05,
+      "loss": 1.5213,
+      "step": 115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.264516129032257e-05,
+      "loss": 1.4349,
+      "step": 116
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.258064516129033e-05,
+      "loss": 1.4269,
+      "step": 117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.251612903225807e-05,
+      "loss": 1.3975,
+      "step": 118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.245161290322581e-05,
+      "loss": 1.4177,
+      "step": 119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.238709677419355e-05,
+      "loss": 1.365,
+      "step": 120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.23225806451613e-05,
+      "loss": 1.3248,
+      "step": 121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.225806451612904e-05,
+      "loss": 1.4621,
+      "step": 122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.219354838709678e-05,
+      "loss": 1.472,
+      "step": 123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.212903225806452e-05,
+      "loss": 1.1869,
+      "step": 124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.206451612903226e-05,
+      "loss": 1.3621,
+      "step": 125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.4642,
+      "step": 126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.193548387096774e-05,
+      "loss": 1.4772,
+      "step": 127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.187096774193548e-05,
+      "loss": 1.3086,
+      "step": 128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.180645161290324e-05,
+      "loss": 1.6788,
+      "step": 129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.174193548387097e-05,
+      "loss": 1.3363,
+      "step": 130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.16774193548387e-05,
+      "loss": 1.2904,
+      "step": 131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.161290322580646e-05,
+      "loss": 1.3506,
+      "step": 132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.15483870967742e-05,
+      "loss": 1.3903,
+      "step": 133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.148387096774193e-05,
+      "loss": 1.4298,
+      "step": 134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.141935483870968e-05,
+      "loss": 1.5464,
+      "step": 135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.135483870967743e-05,
+      "loss": 1.4275,
+      "step": 136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.129032258064517e-05,
+      "loss": 1.4347,
+      "step": 137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.122580645161291e-05,
+      "loss": 1.4139,
+      "step": 138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.116129032258065e-05,
+      "loss": 1.4364,
+      "step": 139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.109677419354839e-05,
+      "loss": 1.3984,
+      "step": 140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.103225806451613e-05,
+      "loss": 1.2279,
+      "step": 141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.096774193548387e-05,
+      "loss": 1.325,
+      "step": 142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.090322580645161e-05,
+      "loss": 1.2434,
+      "step": 143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.083870967741937e-05,
+      "loss": 1.2724,
+      "step": 144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.07741935483871e-05,
+      "loss": 1.3751,
+      "step": 145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.070967741935484e-05,
+      "loss": 1.3238,
+      "step": 146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.06451612903226e-05,
+      "loss": 1.3634,
+      "step": 147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.058064516129032e-05,
+      "loss": 1.3366,
+      "step": 148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.051612903225806e-05,
+      "loss": 1.4424,
+      "step": 149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.045161290322582e-05,
+      "loss": 1.4755,
+      "step": 150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.038709677419356e-05,
+      "loss": 1.4592,
+      "step": 151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.032258064516129e-05,
+      "loss": 1.2932,
+      "step": 152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.025806451612904e-05,
+      "loss": 1.1448,
+      "step": 153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.019354838709678e-05,
+      "loss": 1.3369,
+      "step": 154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.012903225806451e-05,
+      "loss": 1.1981,
+      "step": 155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.006451612903226e-05,
+      "loss": 1.3879,
+      "step": 156
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9e-05,
+      "loss": 1.2789,
+      "step": 157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.993548387096775e-05,
+      "loss": 1.3595,
+      "step": 158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.987096774193549e-05,
+      "loss": 1.2933,
+      "step": 159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.980645161290323e-05,
+      "loss": 1.5589,
+      "step": 160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.974193548387097e-05,
+      "loss": 1.5362,
+      "step": 161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.967741935483871e-05,
+      "loss": 1.4832,
+      "step": 162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.961290322580645e-05,
+      "loss": 1.3837,
+      "step": 163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.95483870967742e-05,
+      "loss": 1.3511,
+      "step": 164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.948387096774195e-05,
+      "loss": 1.4746,
+      "step": 165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.941935483870968e-05,
+      "loss": 1.4352,
+      "step": 166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.935483870967742e-05,
+      "loss": 1.4561,
+      "step": 167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.929032258064517e-05,
+      "loss": 1.462,
+      "step": 168
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92258064516129e-05,
+      "loss": 1.3375,
+      "step": 169
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916129032258064e-05,
+      "loss": 1.3243,
+      "step": 170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90967741935484e-05,
+      "loss": 1.3681,
+      "step": 171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903225806451614e-05,
+      "loss": 1.3448,
+      "step": 172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.896774193548388e-05,
+      "loss": 1.4693,
+      "step": 173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.890322580645162e-05,
+      "loss": 1.4128,
+      "step": 174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.883870967741936e-05,
+      "loss": 1.3185,
+      "step": 175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.87741935483871e-05,
+      "loss": 1.3531,
+      "step": 176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.870967741935484e-05,
+      "loss": 1.4634,
+      "step": 177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.864516129032259e-05,
+      "loss": 1.456,
+      "step": 178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.858064516129033e-05,
+      "loss": 1.3668,
+      "step": 179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.851612903225807e-05,
+      "loss": 1.3219,
+      "step": 180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.845161290322581e-05,
+      "loss": 1.2563,
+      "step": 181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.838709677419355e-05,
+      "loss": 1.3552,
+      "step": 182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.832258064516129e-05,
+      "loss": 1.2516,
+      "step": 183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.825806451612903e-05,
+      "loss": 1.3944,
+      "step": 184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.819354838709677e-05,
+      "loss": 1.4108,
+      "step": 185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.812903225806453e-05,
+      "loss": 1.2879,
+      "step": 186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.806451612903226e-05,
+      "loss": 1.3993,
+      "step": 187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.5187,
+      "step": 188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.793548387096775e-05,
+      "loss": 1.3581,
+      "step": 189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.787096774193548e-05,
+      "loss": 1.494,
+      "step": 190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.780645161290323e-05,
+      "loss": 1.4936,
+      "step": 191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.774193548387098e-05,
+      "loss": 1.3858,
+      "step": 192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.767741935483872e-05,
+      "loss": 1.2974,
+      "step": 193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.761290322580646e-05,
+      "loss": 1.266,
+      "step": 194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.75483870967742e-05,
+      "loss": 1.3635,
+      "step": 195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.748387096774194e-05,
+      "loss": 1.4649,
+      "step": 196
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.741935483870968e-05,
+      "loss": 1.2769,
+      "step": 197
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.735483870967742e-05,
+      "loss": 1.3293,
+      "step": 198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.729032258064516e-05,
+      "loss": 1.2614,
+      "step": 199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.72258064516129e-05,
+      "loss": 1.2375,
+      "step": 200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.716129032258065e-05,
+      "loss": 1.464,
+      "step": 201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.709677419354839e-05,
+      "loss": 1.3782,
+      "step": 202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.703225806451614e-05,
+      "loss": 1.4477,
+      "step": 203
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.696774193548387e-05,
+      "loss": 1.2875,
+      "step": 204
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.690322580645161e-05,
+      "loss": 1.3477,
+      "step": 205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.683870967741937e-05,
+      "loss": 1.4381,
+      "step": 206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.677419354838711e-05,
+      "loss": 1.3831,
+      "step": 207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.670967741935484e-05,
+      "loss": 1.2368,
+      "step": 208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.664516129032259e-05,
+      "loss": 1.3584,
+      "step": 209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.658064516129033e-05,
+      "loss": 1.4164,
+      "step": 210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.651612903225806e-05,
+      "loss": 1.2897,
+      "step": 211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.645161290322581e-05,
+      "loss": 1.3487,
+      "step": 212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.638709677419356e-05,
+      "loss": 1.386,
+      "step": 213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.63225806451613e-05,
+      "loss": 1.4164,
+      "step": 214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.625806451612904e-05,
+      "loss": 1.5031,
+      "step": 215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.619354838709678e-05,
+      "loss": 1.3228,
+      "step": 216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.612903225806452e-05,
+      "loss": 1.3883,
+      "step": 217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.606451612903226e-05,
+      "loss": 1.5248,
+      "step": 218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.6e-05,
+      "loss": 1.3222,
+      "step": 219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.593548387096774e-05,
+      "loss": 1.3043,
+      "step": 220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.58709677419355e-05,
+      "loss": 1.3316,
+      "step": 221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.580645161290323e-05,
+      "loss": 1.4733,
+      "step": 222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.574193548387097e-05,
+      "loss": 1.3395,
+      "step": 223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.567741935483872e-05,
+      "loss": 1.2229,
+      "step": 224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.561290322580645e-05,
+      "loss": 1.4358,
+      "step": 225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.554838709677419e-05,
+      "loss": 1.2766,
+      "step": 226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.548387096774195e-05,
+      "loss": 1.4023,
+      "step": 227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.541935483870969e-05,
+      "loss": 1.3758,
+      "step": 228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.535483870967742e-05,
+      "loss": 1.4499,
+      "step": 229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.529032258064517e-05,
+      "loss": 1.2924,
+      "step": 230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.522580645161291e-05,
+      "loss": 1.3566,
+      "step": 231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.516129032258064e-05,
+      "loss": 1.3357,
+      "step": 232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.50967741935484e-05,
+      "loss": 1.3551,
+      "step": 233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.503225806451614e-05,
+      "loss": 1.4122,
+      "step": 234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.496774193548388e-05,
+      "loss": 1.4719,
+      "step": 235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.490322580645162e-05,
+      "loss": 1.4777,
+      "step": 236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.483870967741936e-05,
+      "loss": 1.6068,
+      "step": 237
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.47741935483871e-05,
+      "loss": 1.2998,
+      "step": 238
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.470967741935484e-05,
+      "loss": 1.2984,
+      "step": 239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.464516129032258e-05,
+      "loss": 1.3259,
+      "step": 240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.458064516129032e-05,
+      "loss": 1.3474,
+      "step": 241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.451612903225808e-05,
+      "loss": 1.2663,
+      "step": 242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.44516129032258e-05,
+      "loss": 1.4719,
+      "step": 243
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.438709677419355e-05,
+      "loss": 1.2865,
+      "step": 244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.43225806451613e-05,
+      "loss": 1.3435,
+      "step": 245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.425806451612903e-05,
+      "loss": 1.4548,
+      "step": 246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.419354838709677e-05,
+      "loss": 1.3205,
+      "step": 247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.412903225806453e-05,
+      "loss": 1.3323,
+      "step": 248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.406451612903227e-05,
+      "loss": 1.4428,
+      "step": 249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.4e-05,
+      "loss": 1.2907,
+      "step": 250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.393548387096775e-05,
+      "loss": 1.2152,
+      "step": 251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.387096774193549e-05,
+      "loss": 1.3223,
+      "step": 252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.380645161290322e-05,
+      "loss": 1.3879,
+      "step": 253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.374193548387097e-05,
+      "loss": 1.2742,
+      "step": 254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.367741935483872e-05,
+      "loss": 1.6644,
+      "step": 255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.361290322580646e-05,
+      "loss": 1.3722,
+      "step": 256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.35483870967742e-05,
+      "loss": 1.3397,
+      "step": 257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.348387096774194e-05,
+      "loss": 1.4307,
+      "step": 258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.341935483870968e-05,
+      "loss": 1.3314,
+      "step": 259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.335483870967742e-05,
+      "loss": 1.2586,
+      "step": 260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.329032258064516e-05,
+      "loss": 1.3073,
+      "step": 261
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.32258064516129e-05,
+      "loss": 1.3043,
+      "step": 262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.316129032258066e-05,
+      "loss": 1.3565,
+      "step": 263
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.309677419354839e-05,
+      "loss": 1.386,
+      "step": 264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.303225806451613e-05,
+      "loss": 1.3378,
+      "step": 265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.296774193548388e-05,
+      "loss": 1.386,
+      "step": 266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.290322580645161e-05,
+      "loss": 1.4444,
+      "step": 267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.283870967741935e-05,
+      "loss": 1.3499,
+      "step": 268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.27741935483871e-05,
+      "loss": 1.4361,
+      "step": 269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.270967741935485e-05,
+      "loss": 1.37,
+      "step": 270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.264516129032257e-05,
+      "loss": 1.3339,
+      "step": 271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.258064516129033e-05,
+      "loss": 1.3722,
+      "step": 272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.251612903225807e-05,
+      "loss": 1.4191,
+      "step": 273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.245161290322581e-05,
+      "loss": 1.4213,
+      "step": 274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.238709677419355e-05,
+      "loss": 1.2428,
+      "step": 275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.23225806451613e-05,
+      "loss": 1.34,
+      "step": 276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.225806451612904e-05,
+      "loss": 1.2769,
+      "step": 277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.219354838709678e-05,
+      "loss": 1.1893,
+      "step": 278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.212903225806452e-05,
+      "loss": 1.3477,
+      "step": 279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.206451612903226e-05,
+      "loss": 1.4416,
+      "step": 280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.2e-05,
+      "loss": 1.4063,
+      "step": 281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.193548387096774e-05,
+      "loss": 1.322,
+      "step": 282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.187096774193548e-05,
+      "loss": 1.358,
+      "step": 283
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.180645161290324e-05,
+      "loss": 1.4087,
+      "step": 284
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.174193548387097e-05,
+      "loss": 1.4503,
+      "step": 285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.167741935483871e-05,
+      "loss": 1.4052,
+      "step": 286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.161290322580646e-05,
+      "loss": 1.3996,
+      "step": 287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.154838709677419e-05,
+      "loss": 1.3519,
+      "step": 288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.148387096774193e-05,
+      "loss": 1.3025,
+      "step": 289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.141935483870969e-05,
+      "loss": 1.3033,
+      "step": 290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.135483870967743e-05,
+      "loss": 1.2895,
+      "step": 291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.129032258064517e-05,
+      "loss": 1.351,
+      "step": 292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.122580645161291e-05,
+      "loss": 1.2342,
+      "step": 293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.116129032258065e-05,
+      "loss": 1.5016,
+      "step": 294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.109677419354839e-05,
+      "loss": 1.3194,
+      "step": 295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.103225806451613e-05,
+      "loss": 1.2655,
+      "step": 296
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.096774193548387e-05,
+      "loss": 1.3279,
+      "step": 297
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.090322580645162e-05,
+      "loss": 1.3496,
+      "step": 298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.083870967741936e-05,
+      "loss": 1.3972,
+      "step": 299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.07741935483871e-05,
+      "loss": 1.2868,
+      "step": 300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.070967741935484e-05,
+      "loss": 1.2803,
+      "step": 301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.064516129032258e-05,
+      "loss": 1.4244,
+      "step": 302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.058064516129032e-05,
+      "loss": 1.2935,
+      "step": 303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.051612903225806e-05,
+      "loss": 1.2726,
+      "step": 304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.045161290322582e-05,
+      "loss": 1.29,
+      "step": 305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.038709677419355e-05,
+      "loss": 1.3333,
+      "step": 306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.03225806451613e-05,
+      "loss": 1.2197,
+      "step": 307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.025806451612904e-05,
+      "loss": 1.4033,
+      "step": 308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.019354838709677e-05,
+      "loss": 1.2889,
+      "step": 309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.012903225806452e-05,
+      "loss": 1.2519,
+      "step": 310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.006451612903227e-05,
+      "loss": 1.3255,
+      "step": 311
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8e-05,
+      "loss": 1.1336,
+      "step": 312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.993548387096775e-05,
+      "loss": 1.1947,
+      "step": 313
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.987096774193549e-05,
+      "loss": 1.2247,
+      "step": 314
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.980645161290323e-05,
+      "loss": 1.3316,
+      "step": 315
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.974193548387097e-05,
+      "loss": 1.1266,
+      "step": 316
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.967741935483871e-05,
+      "loss": 1.0444,
+      "step": 317
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.961290322580645e-05,
+      "loss": 1.2242,
+      "step": 318
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.95483870967742e-05,
+      "loss": 1.1784,
+      "step": 319
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.948387096774194e-05,
+      "loss": 1.2946,
+      "step": 320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.941935483870968e-05,
+      "loss": 1.3006,
+      "step": 321
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.935483870967743e-05,
+      "loss": 1.2457,
+      "step": 322
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.929032258064516e-05,
+      "loss": 1.2297,
+      "step": 323
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.92258064516129e-05,
+      "loss": 1.2912,
+      "step": 324
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.916129032258066e-05,
+      "loss": 1.2593,
+      "step": 325
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.90967741935484e-05,
+      "loss": 1.1773,
+      "step": 326
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.903225806451613e-05,
+      "loss": 1.2831,
+      "step": 327
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.896774193548388e-05,
+      "loss": 1.2368,
+      "step": 328
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.890322580645162e-05,
+      "loss": 1.2555,
+      "step": 329
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.883870967741935e-05,
+      "loss": 1.3443,
+      "step": 330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.87741935483871e-05,
+      "loss": 1.0449,
+      "step": 331
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.870967741935484e-05,
+      "loss": 1.1819,
+      "step": 332
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.864516129032259e-05,
+      "loss": 1.3371,
+      "step": 333
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.858064516129033e-05,
+      "loss": 1.2173,
+      "step": 334
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.851612903225807e-05,
+      "loss": 1.341,
+      "step": 335
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.845161290322581e-05,
+      "loss": 1.1811,
+      "step": 336
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.838709677419355e-05,
+      "loss": 1.3238,
+      "step": 337
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.832258064516129e-05,
+      "loss": 1.4116,
+      "step": 338
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.825806451612903e-05,
+      "loss": 1.257,
+      "step": 339
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.819354838709679e-05,
+      "loss": 1.2128,
+      "step": 340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.812903225806452e-05,
+      "loss": 1.2872,
+      "step": 341
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.806451612903226e-05,
+      "loss": 1.2865,
+      "step": 342
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.2735,
+      "step": 343
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.793548387096774e-05,
+      "loss": 1.1877,
+      "step": 344
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.787096774193548e-05,
+      "loss": 1.2987,
+      "step": 345
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.780645161290324e-05,
+      "loss": 1.1791,
+      "step": 346
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.774193548387098e-05,
+      "loss": 1.2194,
+      "step": 347
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.76774193548387e-05,
+      "loss": 1.2898,
+      "step": 348
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.761290322580646e-05,
+      "loss": 1.2191,
+      "step": 349
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.75483870967742e-05,
+      "loss": 1.1902,
+      "step": 350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.748387096774193e-05,
+      "loss": 1.2546,
+      "step": 351
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.741935483870968e-05,
+      "loss": 1.2232,
+      "step": 352
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.735483870967742e-05,
+      "loss": 1.1641,
+      "step": 353
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.729032258064517e-05,
+      "loss": 1.247,
+      "step": 354
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.722580645161291e-05,
+      "loss": 1.1752,
+      "step": 355
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.716129032258065e-05,
+      "loss": 1.1836,
+      "step": 356
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.709677419354839e-05,
+      "loss": 1.2564,
+      "step": 357
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.703225806451613e-05,
+      "loss": 1.2807,
+      "step": 358
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.696774193548387e-05,
+      "loss": 1.267,
+      "step": 359
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.690322580645161e-05,
+      "loss": 1.2612,
+      "step": 360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.683870967741937e-05,
+      "loss": 1.2218,
+      "step": 361
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.67741935483871e-05,
+      "loss": 1.1365,
+      "step": 362
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.670967741935484e-05,
+      "loss": 1.0789,
+      "step": 363
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.664516129032259e-05,
+      "loss": 1.2635,
+      "step": 364
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.658064516129032e-05,
+      "loss": 1.22,
+      "step": 365
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.651612903225806e-05,
+      "loss": 1.1883,
+      "step": 366
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.645161290322582e-05,
+      "loss": 1.3253,
+      "step": 367
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.638709677419356e-05,
+      "loss": 1.3024,
+      "step": 368
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.632258064516128e-05,
+      "loss": 1.2581,
+      "step": 369
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.625806451612904e-05,
+      "loss": 0.972,
+      "step": 370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.619354838709678e-05,
+      "loss": 1.2729,
+      "step": 371
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.612903225806451e-05,
+      "loss": 1.3093,
+      "step": 372
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.606451612903226e-05,
+      "loss": 1.1775,
+      "step": 373
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.6e-05,
+      "loss": 1.1554,
+      "step": 374
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.593548387096775e-05,
+      "loss": 1.2968,
+      "step": 375
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.587096774193549e-05,
+      "loss": 1.3013,
+      "step": 376
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.580645161290323e-05,
+      "loss": 1.32,
+      "step": 377
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.574193548387097e-05,
+      "loss": 1.1921,
+      "step": 378
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.567741935483871e-05,
+      "loss": 1.38,
+      "step": 379
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.561290322580645e-05,
+      "loss": 1.1201,
+      "step": 380
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.554838709677419e-05,
+      "loss": 1.2216,
+      "step": 381
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.548387096774195e-05,
+      "loss": 1.2189,
+      "step": 382
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.541935483870968e-05,
+      "loss": 1.3504,
+      "step": 383
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.535483870967742e-05,
+      "loss": 1.2688,
+      "step": 384
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.529032258064517e-05,
+      "loss": 1.1411,
+      "step": 385
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.52258064516129e-05,
+      "loss": 1.3266,
+      "step": 386
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.516129032258064e-05,
+      "loss": 1.2878,
+      "step": 387
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.50967741935484e-05,
+      "loss": 1.2512,
+      "step": 388
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.503225806451614e-05,
+      "loss": 1.1967,
+      "step": 389
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.496774193548386e-05,
+      "loss": 1.3149,
+      "step": 390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.490322580645162e-05,
+      "loss": 1.2203,
+      "step": 391
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.483870967741936e-05,
+      "loss": 1.2826,
+      "step": 392
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.47741935483871e-05,
+      "loss": 1.2981,
+      "step": 393
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.470967741935484e-05,
+      "loss": 1.2771,
+      "step": 394
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.464516129032258e-05,
+      "loss": 1.2138,
+      "step": 395
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.458064516129033e-05,
+      "loss": 1.1726,
+      "step": 396
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.451612903225807e-05,
+      "loss": 1.2656,
+      "step": 397
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.445161290322581e-05,
+      "loss": 1.2628,
+      "step": 398
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.438709677419355e-05,
+      "loss": 1.2797,
+      "step": 399
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.43225806451613e-05,
+      "loss": 1.2491,
+      "step": 400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.425806451612903e-05,
+      "loss": 1.2502,
+      "step": 401
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.419354838709677e-05,
+      "loss": 1.1875,
+      "step": 402
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.412903225806453e-05,
+      "loss": 1.2236,
+      "step": 403
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.406451612903225e-05,
+      "loss": 1.1255,
+      "step": 404
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.4e-05,
+      "loss": 1.2813,
+      "step": 405
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.393548387096775e-05,
+      "loss": 1.1283,
+      "step": 406
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.387096774193549e-05,
+      "loss": 1.2351,
+      "step": 407
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.380645161290323e-05,
+      "loss": 1.2723,
+      "step": 408
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.374193548387097e-05,
+      "loss": 1.2865,
+      "step": 409
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.367741935483872e-05,
+      "loss": 1.2589,
+      "step": 410
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.361290322580646e-05,
+      "loss": 1.2197,
+      "step": 411
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.35483870967742e-05,
+      "loss": 1.2036,
+      "step": 412
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.348387096774194e-05,
+      "loss": 1.1727,
+      "step": 413
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.341935483870968e-05,
+      "loss": 1.2835,
+      "step": 414
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.335483870967742e-05,
+      "loss": 1.1364,
+      "step": 415
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.329032258064516e-05,
+      "loss": 1.1284,
+      "step": 416
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.32258064516129e-05,
+      "loss": 1.1517,
+      "step": 417
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.316129032258065e-05,
+      "loss": 1.1535,
+      "step": 418
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.309677419354839e-05,
+      "loss": 1.1734,
+      "step": 419
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.303225806451613e-05,
+      "loss": 1.1108,
+      "step": 420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.296774193548388e-05,
+      "loss": 1.2159,
+      "step": 421
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.290322580645161e-05,
+      "loss": 1.1729,
+      "step": 422
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.283870967741937e-05,
+      "loss": 1.1992,
+      "step": 423
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.277419354838711e-05,
+      "loss": 1.239,
+      "step": 424
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.270967741935483e-05,
+      "loss": 1.3547,
+      "step": 425
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.264516129032259e-05,
+      "loss": 1.3013,
+      "step": 426
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.258064516129033e-05,
+      "loss": 1.2471,
+      "step": 427
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.251612903225807e-05,
+      "loss": 1.2169,
+      "step": 428
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.245161290322581e-05,
+      "loss": 1.2501,
+      "step": 429
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.238709677419355e-05,
+      "loss": 1.1312,
+      "step": 430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.23225806451613e-05,
+      "loss": 1.2524,
+      "step": 431
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.225806451612904e-05,
+      "loss": 1.3236,
+      "step": 432
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.219354838709678e-05,
+      "loss": 1.0989,
+      "step": 433
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.212903225806452e-05,
+      "loss": 1.2075,
+      "step": 434
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.206451612903226e-05,
+      "loss": 1.3255,
+      "step": 435
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.2e-05,
+      "loss": 1.1495,
+      "step": 436
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.193548387096774e-05,
+      "loss": 1.0892,
+      "step": 437
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.187096774193548e-05,
+      "loss": 1.2068,
+      "step": 438
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.180645161290323e-05,
+      "loss": 1.1091,
+      "step": 439
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.174193548387097e-05,
+      "loss": 1.0941,
+      "step": 440
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.167741935483872e-05,
+      "loss": 1.2172,
+      "step": 441
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.161290322580646e-05,
+      "loss": 1.3094,
+      "step": 442
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.154838709677419e-05,
+      "loss": 1.2521,
+      "step": 443
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.148387096774195e-05,
+      "loss": 1.1626,
+      "step": 444
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.141935483870969e-05,
+      "loss": 1.2176,
+      "step": 445
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.135483870967741e-05,
+      "loss": 1.1943,
+      "step": 446
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.129032258064517e-05,
+      "loss": 1.1103,
+      "step": 447
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.122580645161291e-05,
+      "loss": 1.2457,
+      "step": 448
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.116129032258065e-05,
+      "loss": 1.1279,
+      "step": 449
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.109677419354839e-05,
+      "loss": 1.217,
+      "step": 450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.103225806451613e-05,
+      "loss": 1.2433,
+      "step": 451
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.096774193548388e-05,
+      "loss": 1.2683,
+      "step": 452
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.090322580645162e-05,
+      "loss": 1.0663,
+      "step": 453
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.083870967741936e-05,
+      "loss": 1.175,
+      "step": 454
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.07741935483871e-05,
+      "loss": 1.1702,
+      "step": 455
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.070967741935485e-05,
+      "loss": 1.2893,
+      "step": 456
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.064516129032258e-05,
+      "loss": 1.2822,
+      "step": 457
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.058064516129032e-05,
+      "loss": 1.012,
+      "step": 458
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.051612903225808e-05,
+      "loss": 1.2837,
+      "step": 459
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.04516129032258e-05,
+      "loss": 1.1838,
+      "step": 460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.038709677419355e-05,
+      "loss": 1.1814,
+      "step": 461
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.03225806451613e-05,
+      "loss": 1.1655,
+      "step": 462
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.025806451612904e-05,
+      "loss": 1.1503,
+      "step": 463
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.019354838709677e-05,
+      "loss": 1.2196,
+      "step": 464
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.012903225806452e-05,
+      "loss": 1.2698,
+      "step": 465
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.006451612903227e-05,
+      "loss": 1.1131,
+      "step": 466
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7e-05,
+      "loss": 1.2198,
+      "step": 467
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.993548387096775e-05,
+      "loss": 1.1703,
+      "step": 468
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.987096774193549e-05,
+      "loss": 1.1876,
+      "step": 469
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.980645161290323e-05,
+      "loss": 1.2142,
+      "step": 470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.974193548387097e-05,
+      "loss": 1.1484,
+      "step": 471
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.967741935483871e-05,
+      "loss": 1.1529,
+      "step": 472
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.961290322580645e-05,
+      "loss": 1.4009,
+      "step": 473
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.95483870967742e-05,
+      "loss": 1.2733,
+      "step": 474
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.948387096774194e-05,
+      "loss": 1.2198,
+      "step": 475
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.948387096774194e-05,
+      "loss": 1.2224,
+      "step": 476
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.941935483870968e-05,
+      "loss": 1.0744,
+      "step": 477
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.935483870967743e-05,
+      "loss": 1.2484,
+      "step": 478
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.929032258064516e-05,
+      "loss": 1.2018,
+      "step": 479
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.92258064516129e-05,
+      "loss": 1.0806,
+      "step": 480
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.916129032258066e-05,
+      "loss": 1.1088,
+      "step": 481
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.909677419354838e-05,
+      "loss": 1.3505,
+      "step": 482
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.903225806451613e-05,
+      "loss": 1.113,
+      "step": 483
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.896774193548388e-05,
+      "loss": 1.3283,
+      "step": 484
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.890322580645162e-05,
+      "loss": 1.2982,
+      "step": 485
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.883870967741935e-05,
+      "loss": 1.307,
+      "step": 486
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.87741935483871e-05,
+      "loss": 1.1495,
+      "step": 487
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.870967741935485e-05,
+      "loss": 1.1041,
+      "step": 488
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.864516129032257e-05,
+      "loss": 1.1484,
+      "step": 489
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.858064516129033e-05,
+      "loss": 1.2887,
+      "step": 490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.851612903225807e-05,
+      "loss": 1.1657,
+      "step": 491
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.845161290322581e-05,
+      "loss": 1.2411,
+      "step": 492
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.838709677419355e-05,
+      "loss": 1.1875,
+      "step": 493
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.83225806451613e-05,
+      "loss": 1.3288,
+      "step": 494
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.825806451612903e-05,
+      "loss": 1.0947,
+      "step": 495
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.819354838709678e-05,
+      "loss": 1.2137,
+      "step": 496
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.812903225806452e-05,
+      "loss": 1.2108,
+      "step": 497
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.806451612903226e-05,
+      "loss": 1.194,
+      "step": 498
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.1875,
+      "step": 499
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.793548387096774e-05,
+      "loss": 1.325,
+      "step": 500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.787096774193548e-05,
+      "loss": 1.1707,
+      "step": 501
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.780645161290324e-05,
+      "loss": 1.3017,
+      "step": 502
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.774193548387096e-05,
+      "loss": 1.4463,
+      "step": 503
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.76774193548387e-05,
+      "loss": 1.1761,
+      "step": 504
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.761290322580646e-05,
+      "loss": 1.3094,
+      "step": 505
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.75483870967742e-05,
+      "loss": 1.19,
+      "step": 506
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.748387096774193e-05,
+      "loss": 1.2318,
+      "step": 507
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.741935483870968e-05,
+      "loss": 1.2646,
+      "step": 508
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.735483870967743e-05,
+      "loss": 1.2207,
+      "step": 509
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.729032258064515e-05,
+      "loss": 1.1883,
+      "step": 510
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.722580645161291e-05,
+      "loss": 1.1763,
+      "step": 511
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.716129032258065e-05,
+      "loss": 1.2934,
+      "step": 512
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.709677419354839e-05,
+      "loss": 1.2894,
+      "step": 513
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.703225806451613e-05,
+      "loss": 1.2498,
+      "step": 514
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.696774193548387e-05,
+      "loss": 1.183,
+      "step": 515
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.690322580645161e-05,
+      "loss": 1.2735,
+      "step": 516
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.683870967741936e-05,
+      "loss": 1.1541,
+      "step": 517
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.67741935483871e-05,
+      "loss": 1.2153,
+      "step": 518
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.670967741935484e-05,
+      "loss": 1.2869,
+      "step": 519
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.664516129032259e-05,
+      "loss": 1.2448,
+      "step": 520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.658064516129032e-05,
+      "loss": 1.1966,
+      "step": 521
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.651612903225806e-05,
+      "loss": 1.2238,
+      "step": 522
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.645161290322582e-05,
+      "loss": 1.2975,
+      "step": 523
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.638709677419354e-05,
+      "loss": 1.3073,
+      "step": 524
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.632258064516129e-05,
+      "loss": 1.1785,
+      "step": 525
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.625806451612904e-05,
+      "loss": 1.2413,
+      "step": 526
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.619354838709678e-05,
+      "loss": 1.2806,
+      "step": 527
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.612903225806452e-05,
+      "loss": 1.14,
+      "step": 528
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.606451612903226e-05,
+      "loss": 1.233,
+      "step": 529
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.6e-05,
+      "loss": 1.2962,
+      "step": 530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.593548387096775e-05,
+      "loss": 1.0797,
+      "step": 531
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.587096774193549e-05,
+      "loss": 1.2569,
+      "step": 532
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.580645161290323e-05,
+      "loss": 1.1821,
+      "step": 533
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.574193548387097e-05,
+      "loss": 1.2141,
+      "step": 534
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.567741935483871e-05,
+      "loss": 1.2667,
+      "step": 535
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.561290322580645e-05,
+      "loss": 1.1583,
+      "step": 536
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.55483870967742e-05,
+      "loss": 1.2265,
+      "step": 537
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.548387096774193e-05,
+      "loss": 1.2142,
+      "step": 538
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.541935483870968e-05,
+      "loss": 1.2357,
+      "step": 539
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.535483870967742e-05,
+      "loss": 1.2936,
+      "step": 540
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.529032258064517e-05,
+      "loss": 1.1729,
+      "step": 541
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.52258064516129e-05,
+      "loss": 1.1928,
+      "step": 542
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.516129032258065e-05,
+      "loss": 1.3035,
+      "step": 543
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.50967741935484e-05,
+      "loss": 1.3385,
+      "step": 544
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.503225806451612e-05,
+      "loss": 1.3105,
+      "step": 545
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.496774193548388e-05,
+      "loss": 1.1637,
+      "step": 546
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.490322580645162e-05,
+      "loss": 1.1861,
+      "step": 547
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.483870967741936e-05,
+      "loss": 1.2912,
+      "step": 548
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.47741935483871e-05,
+      "loss": 1.2126,
+      "step": 549
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.470967741935484e-05,
+      "loss": 1.2942,
+      "step": 550
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.464516129032258e-05,
+      "loss": 1.2012,
+      "step": 551
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.458064516129033e-05,
+      "loss": 1.2668,
+      "step": 552
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.451612903225807e-05,
+      "loss": 1.2788,
+      "step": 553
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.445161290322581e-05,
+      "loss": 1.2599,
+      "step": 554
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.438709677419355e-05,
+      "loss": 1.0709,
+      "step": 555
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.432258064516129e-05,
+      "loss": 1.3085,
+      "step": 556
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.425806451612903e-05,
+      "loss": 1.2126,
+      "step": 557
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.419354838709679e-05,
+      "loss": 1.2078,
+      "step": 558
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.412903225806451e-05,
+      "loss": 1.1256,
+      "step": 559
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.406451612903226e-05,
+      "loss": 1.2807,
+      "step": 560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.2953,
+      "step": 561
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.393548387096775e-05,
+      "loss": 1.3366,
+      "step": 562
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.387096774193548e-05,
+      "loss": 1.1301,
+      "step": 563
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.380645161290323e-05,
+      "loss": 1.2603,
+      "step": 564
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.374193548387098e-05,
+      "loss": 1.2094,
+      "step": 565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.36774193548387e-05,
+      "loss": 1.2262,
+      "step": 566
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.361290322580646e-05,
+      "loss": 1.1948,
+      "step": 567
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.35483870967742e-05,
+      "loss": 1.1709,
+      "step": 568
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.348387096774194e-05,
+      "loss": 1.3069,
+      "step": 569
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.341935483870968e-05,
+      "loss": 1.1746,
+      "step": 570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.335483870967742e-05,
+      "loss": 1.1771,
+      "step": 571
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.329032258064516e-05,
+      "loss": 1.1965,
+      "step": 572
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.32258064516129e-05,
+      "loss": 1.2352,
+      "step": 573
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.316129032258065e-05,
+      "loss": 1.3588,
+      "step": 574
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.309677419354839e-05,
+      "loss": 1.3035,
+      "step": 575
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.303225806451614e-05,
+      "loss": 1.102,
+      "step": 576
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.296774193548387e-05,
+      "loss": 1.2424,
+      "step": 577
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.290322580645161e-05,
+      "loss": 1.1647,
+      "step": 578
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.283870967741937e-05,
+      "loss": 1.2188,
+      "step": 579
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.27741935483871e-05,
+      "loss": 1.3609,
+      "step": 580
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.270967741935484e-05,
+      "loss": 1.2879,
+      "step": 581
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.264516129032259e-05,
+      "loss": 1.3221,
+      "step": 582
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.258064516129033e-05,
+      "loss": 1.1337,
+      "step": 583
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.251612903225806e-05,
+      "loss": 1.1611,
+      "step": 584
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.245161290322581e-05,
+      "loss": 1.2829,
+      "step": 585
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.238709677419356e-05,
+      "loss": 1.2616,
+      "step": 586
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.232258064516128e-05,
+      "loss": 1.2378,
+      "step": 587
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.225806451612904e-05,
+      "loss": 1.2424,
+      "step": 588
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.219354838709678e-05,
+      "loss": 1.3422,
+      "step": 589
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.212903225806452e-05,
+      "loss": 1.1654,
+      "step": 590
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.206451612903226e-05,
+      "loss": 1.2674,
+      "step": 591
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2e-05,
+      "loss": 1.0968,
+      "step": 592
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.193548387096774e-05,
+      "loss": 1.1989,
+      "step": 593
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.187096774193549e-05,
+      "loss": 1.1781,
+      "step": 594
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.180645161290323e-05,
+      "loss": 1.2104,
+      "step": 595
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.174193548387097e-05,
+      "loss": 1.2879,
+      "step": 596
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.167741935483872e-05,
+      "loss": 1.1927,
+      "step": 597
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.161290322580645e-05,
+      "loss": 1.2184,
+      "step": 598
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.154838709677419e-05,
+      "loss": 1.2234,
+      "step": 599
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.148387096774195e-05,
+      "loss": 1.1684,
+      "step": 600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.141935483870967e-05,
+      "loss": 1.0678,
+      "step": 601
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.135483870967742e-05,
+      "loss": 1.1604,
+      "step": 602
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.129032258064517e-05,
+      "loss": 1.1536,
+      "step": 603
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.122580645161291e-05,
+      "loss": 1.2518,
+      "step": 604
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.116129032258064e-05,
+      "loss": 1.1394,
+      "step": 605
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.10967741935484e-05,
+      "loss": 1.238,
+      "step": 606
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.1032258064516135e-05,
+      "loss": 1.2727,
+      "step": 607
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.096774193548387e-05,
+      "loss": 1.2239,
+      "step": 608
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.090322580645161e-05,
+      "loss": 1.1393,
+      "step": 609
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.083870967741936e-05,
+      "loss": 1.1231,
+      "step": 610
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.077419354838711e-05,
+      "loss": 1.2431,
+      "step": 611
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.070967741935484e-05,
+      "loss": 1.1693,
+      "step": 612
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.064516129032258e-05,
+      "loss": 1.1998,
+      "step": 613
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.058064516129033e-05,
+      "loss": 1.2099,
+      "step": 614
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.0516129032258065e-05,
+      "loss": 1.1924,
+      "step": 615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.0451612903225806e-05,
+      "loss": 1.1998,
+      "step": 616
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.0387096774193554e-05,
+      "loss": 1.2079,
+      "step": 617
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.0322580645161295e-05,
+      "loss": 1.2224,
+      "step": 618
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.025806451612903e-05,
+      "loss": 1.2026,
+      "step": 619
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.019354838709678e-05,
+      "loss": 1.1097,
+      "step": 620
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.012903225806452e-05,
+      "loss": 1.0573,
+      "step": 621
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.0064516129032254e-05,
+      "loss": 1.0722,
+      "step": 622
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6e-05,
+      "loss": 1.0389,
+      "step": 623
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.993548387096774e-05,
+      "loss": 0.9134,
+      "step": 624
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.987096774193549e-05,
+      "loss": 1.0954,
+      "step": 625
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.9806451612903225e-05,
+      "loss": 1.0856,
+      "step": 626
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.974193548387097e-05,
+      "loss": 0.9766,
+      "step": 627
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.9677419354838715e-05,
+      "loss": 1.0056,
+      "step": 628
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.961290322580645e-05,
+      "loss": 0.9547,
+      "step": 629
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.95483870967742e-05,
+      "loss": 1.023,
+      "step": 630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.948387096774194e-05,
+      "loss": 1.0786,
+      "step": 631
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.9419354838709686e-05,
+      "loss": 1.0435,
+      "step": 632
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.935483870967742e-05,
+      "loss": 1.0513,
+      "step": 633
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.929032258064516e-05,
+      "loss": 0.9474,
+      "step": 634
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.922580645161291e-05,
+      "loss": 1.0442,
+      "step": 635
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.9161290322580644e-05,
+      "loss": 0.9425,
+      "step": 636
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.9096774193548386e-05,
+      "loss": 1.0467,
+      "step": 637
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.9032258064516134e-05,
+      "loss": 1.0366,
+      "step": 638
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.8967741935483875e-05,
+      "loss": 1.0204,
+      "step": 639
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.890322580645161e-05,
+      "loss": 0.9771,
+      "step": 640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.883870967741936e-05,
+      "loss": 1.0935,
+      "step": 641
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.87741935483871e-05,
+      "loss": 0.9953,
+      "step": 642
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.870967741935483e-05,
+      "loss": 0.9362,
+      "step": 643
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.864516129032258e-05,
+      "loss": 0.8792,
+      "step": 644
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.858064516129033e-05,
+      "loss": 0.9622,
+      "step": 645
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.851612903225807e-05,
+      "loss": 0.9641,
+      "step": 646
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.8451612903225805e-05,
+      "loss": 1.028,
+      "step": 647
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.838709677419355e-05,
+      "loss": 1.0005,
+      "step": 648
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.8322580645161294e-05,
+      "loss": 0.9237,
+      "step": 649
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.825806451612904e-05,
+      "loss": 1.0742,
+      "step": 650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.819354838709678e-05,
+      "loss": 0.9908,
+      "step": 651
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.812903225806452e-05,
+      "loss": 1.0778,
+      "step": 652
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.8064516129032266e-05,
+      "loss": 1.0134,
+      "step": 653
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.8e-05,
+      "loss": 0.991,
+      "step": 654
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.793548387096774e-05,
+      "loss": 1.1276,
+      "step": 655
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.787096774193549e-05,
+      "loss": 0.8141,
+      "step": 656
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.780645161290323e-05,
+      "loss": 0.9493,
+      "step": 657
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.7741935483870965e-05,
+      "loss": 1.0376,
+      "step": 658
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.767741935483871e-05,
+      "loss": 1.2083,
+      "step": 659
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.761290322580646e-05,
+      "loss": 1.0257,
+      "step": 660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.754838709677419e-05,
+      "loss": 0.9465,
+      "step": 661
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.748387096774194e-05,
+      "loss": 1.0043,
+      "step": 662
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7419354838709685e-05,
+      "loss": 1.1537,
+      "step": 663
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7354838709677426e-05,
+      "loss": 1.0005,
+      "step": 664
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.729032258064516e-05,
+      "loss": 0.9794,
+      "step": 665
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.722580645161291e-05,
+      "loss": 0.9691,
+      "step": 666
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.716129032258065e-05,
+      "loss": 1.0826,
+      "step": 667
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.7096774193548384e-05,
+      "loss": 0.8696,
+      "step": 668
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.703225806451613e-05,
+      "loss": 1.1422,
+      "step": 669
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6967741935483874e-05,
+      "loss": 0.9148,
+      "step": 670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.690322580645162e-05,
+      "loss": 1.0487,
+      "step": 671
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.6838709677419356e-05,
+      "loss": 1.0589,
+      "step": 672
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.67741935483871e-05,
+      "loss": 1.096,
+      "step": 673
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.6709677419354846e-05,
+      "loss": 1.0005,
+      "step": 674
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.664516129032258e-05,
+      "loss": 0.9411,
+      "step": 675
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.658064516129032e-05,
+      "loss": 1.1504,
+      "step": 676
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.651612903225807e-05,
+      "loss": 1.0322,
+      "step": 677
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.645161290322582e-05,
+      "loss": 0.9941,
+      "step": 678
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.638709677419355e-05,
+      "loss": 1.1151,
+      "step": 679
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.632258064516129e-05,
+      "loss": 1.0767,
+      "step": 680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.625806451612904e-05,
+      "loss": 1.0894,
+      "step": 681
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.6193548387096775e-05,
+      "loss": 1.0369,
+      "step": 682
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.612903225806452e-05,
+      "loss": 1.0255,
+      "step": 683
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.6064516129032265e-05,
+      "loss": 0.9577,
+      "step": 684
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.0897,
+      "step": 685
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.593548387096774e-05,
+      "loss": 0.8976,
+      "step": 686
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.587096774193549e-05,
+      "loss": 1.0415,
+      "step": 687
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.580645161290323e-05,
+      "loss": 0.9691,
+      "step": 688
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.5741935483870964e-05,
+      "loss": 1.1505,
+      "step": 689
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.567741935483871e-05,
+      "loss": 0.9205,
+      "step": 690
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.561290322580645e-05,
+      "loss": 0.9398,
+      "step": 691
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.55483870967742e-05,
+      "loss": 0.9206,
+      "step": 692
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.5483870967741936e-05,
+      "loss": 1.0684,
+      "step": 693
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.5419354838709684e-05,
+      "loss": 1.0075,
+      "step": 694
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.5354838709677425e-05,
+      "loss": 1.013,
+      "step": 695
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.529032258064516e-05,
+      "loss": 0.9477,
+      "step": 696
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.522580645161291e-05,
+      "loss": 1.0021,
+      "step": 697
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.516129032258065e-05,
+      "loss": 1.0695,
+      "step": 698
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.50967741935484e-05,
+      "loss": 0.9361,
+      "step": 699
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.503225806451613e-05,
+      "loss": 1.0084,
+      "step": 700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.496774193548387e-05,
+      "loss": 1.1262,
+      "step": 701
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.490322580645162e-05,
+      "loss": 1.1294,
+      "step": 702
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.4838709677419355e-05,
+      "loss": 0.9938,
+      "step": 703
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.4774193548387096e-05,
+      "loss": 1.0987,
+      "step": 704
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.4709677419354844e-05,
+      "loss": 0.9692,
+      "step": 705
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.4645161290322586e-05,
+      "loss": 1.0963,
+      "step": 706
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.458064516129032e-05,
+      "loss": 1.0611,
+      "step": 707
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.451612903225807e-05,
+      "loss": 1.0953,
+      "step": 708
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.445161290322581e-05,
+      "loss": 1.108,
+      "step": 709
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.4387096774193544e-05,
+      "loss": 0.9869,
+      "step": 710
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.432258064516129e-05,
+      "loss": 0.9568,
+      "step": 711
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.425806451612904e-05,
+      "loss": 1.1244,
+      "step": 712
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.419354838709678e-05,
+      "loss": 1.1645,
+      "step": 713
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.4129032258064515e-05,
+      "loss": 1.0168,
+      "step": 714
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.4064516129032264e-05,
+      "loss": 0.9239,
+      "step": 715
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 1.0431,
+      "step": 716
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.393548387096774e-05,
+      "loss": 1.0123,
+      "step": 717
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.387096774193549e-05,
+      "loss": 1.0819,
+      "step": 718
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.380645161290323e-05,
+      "loss": 1.0884,
+      "step": 719
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.3741935483870976e-05,
+      "loss": 0.9856,
+      "step": 720
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.367741935483871e-05,
+      "loss": 0.9892,
+      "step": 721
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.361290322580645e-05,
+      "loss": 1.0186,
+      "step": 722
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.35483870967742e-05,
+      "loss": 1.0671,
+      "step": 723
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.3483870967741935e-05,
+      "loss": 1.1127,
+      "step": 724
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.3419354838709676e-05,
+      "loss": 1.0388,
+      "step": 725
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.3354838709677424e-05,
+      "loss": 0.9752,
+      "step": 726
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.329032258064517e-05,
+      "loss": 1.0875,
+      "step": 727
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.32258064516129e-05,
+      "loss": 0.9491,
+      "step": 728
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.316129032258065e-05,
+      "loss": 1.1455,
+      "step": 729
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.3096774193548396e-05,
+      "loss": 0.9851,
+      "step": 730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.303225806451613e-05,
+      "loss": 0.9875,
+      "step": 731
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.296774193548387e-05,
+      "loss": 1.0846,
+      "step": 732
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.290322580645162e-05,
+      "loss": 1.051,
+      "step": 733
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.283870967741936e-05,
+      "loss": 1.0024,
+      "step": 734
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.2774193548387095e-05,
+      "loss": 1.0766,
+      "step": 735
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.270967741935484e-05,
+      "loss": 0.9315,
+      "step": 736
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.2645161290322584e-05,
+      "loss": 1.0206,
+      "step": 737
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.258064516129032e-05,
+      "loss": 0.9521,
+      "step": 738
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.251612903225807e-05,
+      "loss": 1.0473,
+      "step": 739
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.245161290322581e-05,
+      "loss": 1.0606,
+      "step": 740
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.2387096774193556e-05,
+      "loss": 1.1122,
+      "step": 741
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.232258064516129e-05,
+      "loss": 1.0409,
+      "step": 742
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.225806451612903e-05,
+      "loss": 1.0086,
+      "step": 743
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.219354838709678e-05,
+      "loss": 1.003,
+      "step": 744
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.2129032258064514e-05,
+      "loss": 1.015,
+      "step": 745
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.206451612903226e-05,
+      "loss": 1.0287,
+      "step": 746
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.1539,
+      "step": 747
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.193548387096775e-05,
+      "loss": 1.1138,
+      "step": 748
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.1870967741935486e-05,
+      "loss": 1.1595,
+      "step": 749
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.180645161290323e-05,
+      "loss": 1.0326,
+      "step": 750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.1741935483870975e-05,
+      "loss": 1.0772,
+      "step": 751
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.167741935483871e-05,
+      "loss": 0.9341,
+      "step": 752
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.161290322580645e-05,
+      "loss": 1.0321,
+      "step": 753
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.15483870967742e-05,
+      "loss": 1.2155,
+      "step": 754
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.148387096774194e-05,
+      "loss": 1.0748,
+      "step": 755
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.1419354838709675e-05,
+      "loss": 1.0376,
+      "step": 756
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.135483870967742e-05,
+      "loss": 1.0336,
+      "step": 757
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.1290322580645164e-05,
+      "loss": 0.9883,
+      "step": 758
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.12258064516129e-05,
+      "loss": 0.9669,
+      "step": 759
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.1161290322580646e-05,
+      "loss": 0.9783,
+      "step": 760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.1096774193548395e-05,
+      "loss": 0.9964,
+      "step": 761
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.1032258064516136e-05,
+      "loss": 1.1133,
+      "step": 762
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.096774193548387e-05,
+      "loss": 1.0046,
+      "step": 763
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.090322580645162e-05,
+      "loss": 1.0321,
+      "step": 764
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.083870967741936e-05,
+      "loss": 1.036,
+      "step": 765
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.0774193548387094e-05,
+      "loss": 1.017,
+      "step": 766
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.070967741935484e-05,
+      "loss": 0.9967,
+      "step": 767
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.064516129032258e-05,
+      "loss": 0.9045,
+      "step": 768
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.058064516129033e-05,
+      "loss": 1.1871,
+      "step": 769
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.0516129032258066e-05,
+      "loss": 1.0803,
+      "step": 770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.045161290322581e-05,
+      "loss": 1.2301,
+      "step": 771
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.0387096774193555e-05,
+      "loss": 0.917,
+      "step": 772
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.032258064516129e-05,
+      "loss": 0.9631,
+      "step": 773
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.025806451612903e-05,
+      "loss": 0.9958,
+      "step": 774
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.019354838709678e-05,
+      "loss": 0.9283,
+      "step": 775
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.012903225806453e-05,
+      "loss": 1.027,
+      "step": 776
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.0064516129032254e-05,
+      "loss": 0.9122,
+      "step": 777
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5e-05,
+      "loss": 1.1081,
+      "step": 778
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.9935483870967744e-05,
+      "loss": 1.0485,
+      "step": 779
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.9870967741935485e-05,
+      "loss": 1.1539,
+      "step": 780
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.9806451612903226e-05,
+      "loss": 1.1736,
+      "step": 781
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.9741935483870974e-05,
+      "loss": 0.9283,
+      "step": 782
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.967741935483871e-05,
+      "loss": 1.1044,
+      "step": 783
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.961290322580646e-05,
+      "loss": 1.0973,
+      "step": 784
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.95483870967742e-05,
+      "loss": 1.0377,
+      "step": 785
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.948387096774193e-05,
+      "loss": 1.027,
+      "step": 786
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.941935483870968e-05,
+      "loss": 1.002,
+      "step": 787
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.935483870967742e-05,
+      "loss": 1.0909,
+      "step": 788
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.929032258064516e-05,
+      "loss": 0.9661,
+      "step": 789
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.9225806451612904e-05,
+      "loss": 1.0551,
+      "step": 790
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.916129032258065e-05,
+      "loss": 1.029,
+      "step": 791
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.9096774193548387e-05,
+      "loss": 1.1676,
+      "step": 792
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.903225806451613e-05,
+      "loss": 1.1336,
+      "step": 793
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.8967741935483876e-05,
+      "loss": 0.9343,
+      "step": 794
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.890322580645161e-05,
+      "loss": 0.9677,
+      "step": 795
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.883870967741936e-05,
+      "loss": 1.0292,
+      "step": 796
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.87741935483871e-05,
+      "loss": 1.0626,
+      "step": 797
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.870967741935484e-05,
+      "loss": 1.0186,
+      "step": 798
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.864516129032258e-05,
+      "loss": 0.9717,
+      "step": 799
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.858064516129032e-05,
+      "loss": 1.1345,
+      "step": 800
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8516129032258065e-05,
+      "loss": 1.0375,
+      "step": 801
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.8451612903225806e-05,
+      "loss": 0.9368,
+      "step": 802
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.8387096774193554e-05,
+      "loss": 1.0571,
+      "step": 803
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.8322580645161295e-05,
+      "loss": 1.0389,
+      "step": 804
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.8258064516129036e-05,
+      "loss": 1.0602,
+      "step": 805
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.819354838709678e-05,
+      "loss": 1.0863,
+      "step": 806
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.812903225806452e-05,
+      "loss": 1.0699,
+      "step": 807
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.806451612903226e-05,
+      "loss": 1.0114,
+      "step": 808
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.8e-05,
+      "loss": 0.9719,
+      "step": 809
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.793548387096774e-05,
+      "loss": 0.9617,
+      "step": 810
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.7870967741935484e-05,
+      "loss": 1.2139,
+      "step": 811
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.780645161290323e-05,
+      "loss": 0.9263,
+      "step": 812
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.774193548387097e-05,
+      "loss": 0.9778,
+      "step": 813
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.767741935483871e-05,
+      "loss": 1.0187,
+      "step": 814
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.7612903225806455e-05,
+      "loss": 1.0138,
+      "step": 815
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.75483870967742e-05,
+      "loss": 0.9936,
+      "step": 816
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.748387096774194e-05,
+      "loss": 0.9786,
+      "step": 817
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.741935483870968e-05,
+      "loss": 1.158,
+      "step": 818
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.735483870967742e-05,
+      "loss": 1.1623,
+      "step": 819
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.729032258064516e-05,
+      "loss": 1.0366,
+      "step": 820
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.72258064516129e-05,
+      "loss": 1.0105,
+      "step": 821
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.716129032258065e-05,
+      "loss": 0.9102,
+      "step": 822
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.7096774193548385e-05,
+      "loss": 1.0105,
+      "step": 823
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.7032258064516133e-05,
+      "loss": 1.1636,
+      "step": 824
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.6967741935483875e-05,
+      "loss": 0.9957,
+      "step": 825
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.6903225806451616e-05,
+      "loss": 1.1424,
+      "step": 826
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.683870967741936e-05,
+      "loss": 1.001,
+      "step": 827
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.67741935483871e-05,
+      "loss": 0.9862,
+      "step": 828
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.670967741935484e-05,
+      "loss": 1.073,
+      "step": 829
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.664516129032258e-05,
+      "loss": 1.0237,
+      "step": 830
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.658064516129033e-05,
+      "loss": 1.0196,
+      "step": 831
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.651612903225806e-05,
+      "loss": 0.961,
+      "step": 832
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.645161290322581e-05,
+      "loss": 1.004,
+      "step": 833
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.638709677419355e-05,
+      "loss": 1.1842,
+      "step": 834
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.632258064516129e-05,
+      "loss": 0.9939,
+      "step": 835
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.6258064516129035e-05,
+      "loss": 1.0968,
+      "step": 836
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.6193548387096776e-05,
+      "loss": 1.0201,
+      "step": 837
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.612903225806452e-05,
+      "loss": 1.0598,
+      "step": 838
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.606451612903226e-05,
+      "loss": 1.024,
+      "step": 839
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.0289,
+      "step": 840
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.593548387096774e-05,
+      "loss": 1.0121,
+      "step": 841
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.587096774193548e-05,
+      "loss": 0.9903,
+      "step": 842
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.580645161290323e-05,
+      "loss": 0.9317,
+      "step": 843
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5741935483870965e-05,
+      "loss": 0.9314,
+      "step": 844
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.567741935483871e-05,
+      "loss": 0.9828,
+      "step": 845
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5612903225806454e-05,
+      "loss": 1.1432,
+      "step": 846
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5548387096774196e-05,
+      "loss": 1.0787,
+      "step": 847
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.548387096774194e-05,
+      "loss": 1.1064,
+      "step": 848
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5419354838709685e-05,
+      "loss": 0.9467,
+      "step": 849
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.535483870967742e-05,
+      "loss": 1.0755,
+      "step": 850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.529032258064516e-05,
+      "loss": 1.0729,
+      "step": 851
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.522580645161291e-05,
+      "loss": 0.8151,
+      "step": 852
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.516129032258064e-05,
+      "loss": 1.0593,
+      "step": 853
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.509677419354839e-05,
+      "loss": 1.0355,
+      "step": 854
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.503225806451613e-05,
+      "loss": 1.0367,
+      "step": 855
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.4967741935483873e-05,
+      "loss": 0.9592,
+      "step": 856
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.4903225806451615e-05,
+      "loss": 0.9615,
+      "step": 857
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.4838709677419356e-05,
+      "loss": 1.1759,
+      "step": 858
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.47741935483871e-05,
+      "loss": 1.0294,
+      "step": 859
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.470967741935484e-05,
+      "loss": 0.9455,
+      "step": 860
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.4645161290322586e-05,
+      "loss": 1.0467,
+      "step": 861
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.458064516129032e-05,
+      "loss": 1.1854,
+      "step": 862
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.451612903225807e-05,
+      "loss": 1.1561,
+      "step": 863
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.445161290322581e-05,
+      "loss": 1.1202,
+      "step": 864
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.438709677419355e-05,
+      "loss": 1.0063,
+      "step": 865
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.432258064516129e-05,
+      "loss": 1.2304,
+      "step": 866
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.4258064516129034e-05,
+      "loss": 1.1192,
+      "step": 867
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.4193548387096775e-05,
+      "loss": 1.0051,
+      "step": 868
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.4129032258064516e-05,
+      "loss": 1.1803,
+      "step": 869
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.4064516129032264e-05,
+      "loss": 0.9707,
+      "step": 870
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.0776,
+      "step": 871
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.393548387096774e-05,
+      "loss": 1.036,
+      "step": 872
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.387096774193549e-05,
+      "loss": 0.9284,
+      "step": 873
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.380645161290323e-05,
+      "loss": 1.0881,
+      "step": 874
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.374193548387097e-05,
+      "loss": 0.9334,
+      "step": 875
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.367741935483871e-05,
+      "loss": 1.0243,
+      "step": 876
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.361290322580645e-05,
+      "loss": 1.0736,
+      "step": 877
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.3548387096774194e-05,
+      "loss": 1.0218,
+      "step": 878
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.3483870967741936e-05,
+      "loss": 1.0716,
+      "step": 879
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.3419354838709684e-05,
+      "loss": 1.0509,
+      "step": 880
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.335483870967742e-05,
+      "loss": 0.9829,
+      "step": 881
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.3290322580645166e-05,
+      "loss": 1.0356,
+      "step": 882
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.322580645161291e-05,
+      "loss": 1.1909,
+      "step": 883
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.316129032258065e-05,
+      "loss": 1.0799,
+      "step": 884
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.309677419354839e-05,
+      "loss": 1.0101,
+      "step": 885
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.303225806451613e-05,
+      "loss": 1.0404,
+      "step": 886
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.296774193548387e-05,
+      "loss": 0.921,
+      "step": 887
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.2903225806451614e-05,
+      "loss": 1.0621,
+      "step": 888
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.283870967741936e-05,
+      "loss": 1.1195,
+      "step": 889
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.2774193548387096e-05,
+      "loss": 0.9656,
+      "step": 890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.2709677419354844e-05,
+      "loss": 0.9574,
+      "step": 891
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.2645161290322585e-05,
+      "loss": 1.0445,
+      "step": 892
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.258064516129032e-05,
+      "loss": 0.9791,
+      "step": 893
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.251612903225807e-05,
+      "loss": 0.959,
+      "step": 894
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.245161290322581e-05,
+      "loss": 0.912,
+      "step": 895
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.238709677419355e-05,
+      "loss": 1.0182,
+      "step": 896
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.232258064516129e-05,
+      "loss": 1.115,
+      "step": 897
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.225806451612904e-05,
+      "loss": 0.9096,
+      "step": 898
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.2193548387096774e-05,
+      "loss": 1.0564,
+      "step": 899
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.2129032258064515e-05,
+      "loss": 0.9584,
+      "step": 900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.206451612903226e-05,
+      "loss": 1.004,
+      "step": 901
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2e-05,
+      "loss": 1.0381,
+      "step": 902
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.1935483870967746e-05,
+      "loss": 1.1434,
+      "step": 903
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.187096774193549e-05,
+      "loss": 0.9617,
+      "step": 904
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.180645161290323e-05,
+      "loss": 1.1639,
+      "step": 905
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.174193548387097e-05,
+      "loss": 0.9943,
+      "step": 906
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.167741935483871e-05,
+      "loss": 1.0384,
+      "step": 907
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.161290322580645e-05,
+      "loss": 0.9414,
+      "step": 908
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.154838709677419e-05,
+      "loss": 1.1443,
+      "step": 909
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.148387096774194e-05,
+      "loss": 1.0652,
+      "step": 910
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1419354838709676e-05,
+      "loss": 1.1806,
+      "step": 911
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1354838709677424e-05,
+      "loss": 1.0389,
+      "step": 912
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.1290322580645165e-05,
+      "loss": 0.9994,
+      "step": 913
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.1225806451612906e-05,
+      "loss": 1.0076,
+      "step": 914
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.116129032258065e-05,
+      "loss": 0.9742,
+      "step": 915
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.109677419354839e-05,
+      "loss": 0.9685,
+      "step": 916
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.103225806451613e-05,
+      "loss": 1.0056,
+      "step": 917
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.096774193548387e-05,
+      "loss": 0.9958,
+      "step": 918
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.090322580645162e-05,
+      "loss": 1.2102,
+      "step": 919
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0838709677419354e-05,
+      "loss": 0.9691,
+      "step": 920
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0774193548387095e-05,
+      "loss": 0.9423,
+      "step": 921
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.070967741935484e-05,
+      "loss": 0.9508,
+      "step": 922
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0645161290322584e-05,
+      "loss": 0.9396,
+      "step": 923
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0580645161290325e-05,
+      "loss": 0.8818,
+      "step": 924
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0516129032258067e-05,
+      "loss": 1.0582,
+      "step": 925
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.045161290322581e-05,
+      "loss": 1.0241,
+      "step": 926
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.038709677419355e-05,
+      "loss": 1.0078,
+      "step": 927
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.032258064516129e-05,
+      "loss": 1.0195,
+      "step": 928
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.025806451612903e-05,
+      "loss": 1.0415,
+      "step": 929
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.019354838709677e-05,
+      "loss": 0.9715,
+      "step": 930
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.012903225806452e-05,
+      "loss": 0.8668,
+      "step": 931
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.006451612903226e-05,
+      "loss": 0.7918,
+      "step": 932
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4e-05,
+      "loss": 0.7938,
+      "step": 933
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.9935483870967745e-05,
+      "loss": 0.7962,
+      "step": 934
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9870967741935486e-05,
+      "loss": 0.8585,
+      "step": 935
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.980645161290323e-05,
+      "loss": 0.7677,
+      "step": 936
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.974193548387097e-05,
+      "loss": 0.7192,
+      "step": 937
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.7077,
+      "step": 938
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.961290322580645e-05,
+      "loss": 0.9244,
+      "step": 939
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.95483870967742e-05,
+      "loss": 0.8375,
+      "step": 940
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.948387096774194e-05,
+      "loss": 0.8264,
+      "step": 941
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.9419354838709674e-05,
+      "loss": 0.7559,
+      "step": 942
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.71,
+      "step": 943
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9290322580645164e-05,
+      "loss": 0.8278,
+      "step": 944
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9225806451612905e-05,
+      "loss": 0.816,
+      "step": 945
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9161290322580646e-05,
+      "loss": 0.8577,
+      "step": 946
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9096774193548394e-05,
+      "loss": 0.8066,
+      "step": 947
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.903225806451613e-05,
+      "loss": 0.8142,
+      "step": 948
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.896774193548387e-05,
+      "loss": 0.9036,
+      "step": 949
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.890322580645162e-05,
+      "loss": 0.6882,
+      "step": 950
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.883870967741935e-05,
+      "loss": 0.7452,
+      "step": 951
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.87741935483871e-05,
+      "loss": 0.7839,
+      "step": 952
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.8035,
+      "step": 953
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.864516129032258e-05,
+      "loss": 0.8279,
+      "step": 954
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.8580645161290324e-05,
+      "loss": 0.8376,
+      "step": 955
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.8516129032258065e-05,
+      "loss": 0.7078,
+      "step": 956
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.845161290322581e-05,
+      "loss": 0.9274,
+      "step": 957
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.838709677419355e-05,
+      "loss": 0.837,
+      "step": 958
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.8322580645161296e-05,
+      "loss": 0.7034,
+      "step": 959
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.825806451612903e-05,
+      "loss": 0.8264,
+      "step": 960
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.819354838709678e-05,
+      "loss": 0.9064,
+      "step": 961
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.812903225806452e-05,
+      "loss": 0.6782,
+      "step": 962
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.8064516129032254e-05,
+      "loss": 0.8262,
+      "step": 963
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.8e-05,
+      "loss": 0.9089,
+      "step": 964
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.793548387096774e-05,
+      "loss": 0.7265,
+      "step": 965
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.7870967741935485e-05,
+      "loss": 0.8651,
+      "step": 966
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.7806451612903226e-05,
+      "loss": 0.8483,
+      "step": 967
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6734,
+      "step": 968
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.767741935483871e-05,
+      "loss": 0.7586,
+      "step": 969
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.761290322580645e-05,
+      "loss": 0.7803,
+      "step": 970
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.75483870967742e-05,
+      "loss": 0.8179,
+      "step": 971
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.748387096774193e-05,
+      "loss": 0.8383,
+      "step": 972
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.9211,
+      "step": 973
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.735483870967742e-05,
+      "loss": 0.8113,
+      "step": 974
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.729032258064516e-05,
+      "loss": 0.8633,
+      "step": 975
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.7225806451612904e-05,
+      "loss": 0.8162,
+      "step": 976
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.716129032258065e-05,
+      "loss": 0.7596,
+      "step": 977
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.691,
+      "step": 978
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.703225806451613e-05,
+      "loss": 0.7505,
+      "step": 979
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.6967741935483876e-05,
+      "loss": 0.8569,
+      "step": 980
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.690322580645162e-05,
+      "loss": 0.8245,
+      "step": 981
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.683870967741936e-05,
+      "loss": 0.727,
+      "step": 982
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.8954,
+      "step": 983
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.670967741935484e-05,
+      "loss": 0.7326,
+      "step": 984
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.664516129032258e-05,
+      "loss": 0.75,
+      "step": 985
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.658064516129032e-05,
+      "loss": 0.7942,
+      "step": 986
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.6516129032258064e-05,
+      "loss": 0.8427,
+      "step": 987
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.7232,
+      "step": 988
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.6387096774193553e-05,
+      "loss": 0.8862,
+      "step": 989
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.6322580645161295e-05,
+      "loss": 0.7916,
+      "step": 990
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.6258064516129036e-05,
+      "loss": 0.8367,
+      "step": 991
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.619354838709678e-05,
+      "loss": 0.7564,
+      "step": 992
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.8702,
+      "step": 993
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.606451612903226e-05,
+      "loss": 0.9461,
+      "step": 994
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.6e-05,
+      "loss": 0.8006,
+      "step": 995
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.593548387096774e-05,
+      "loss": 0.8714,
+      "step": 996
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.5870967741935483e-05,
+      "loss": 0.8469,
+      "step": 997
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.580645161290323e-05,
+      "loss": 0.8389,
+      "step": 998
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.574193548387097e-05,
+      "loss": 0.7487,
+      "step": 999
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.567741935483871e-05,
+      "loss": 0.8682,
+      "step": 1000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.5612903225806455e-05,
+      "loss": 0.7122,
+      "step": 1001
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.5548387096774196e-05,
+      "loss": 0.9075,
+      "step": 1002
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.7634,
+      "step": 1003
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.541935483870968e-05,
+      "loss": 0.838,
+      "step": 1004
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.535483870967743e-05,
+      "loss": 0.8543,
+      "step": 1005
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.529032258064516e-05,
+      "loss": 0.7787,
+      "step": 1006
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.52258064516129e-05,
+      "loss": 0.7279,
+      "step": 1007
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.9031,
+      "step": 1008
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.5096774193548385e-05,
+      "loss": 0.8419,
+      "step": 1009
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.503225806451613e-05,
+      "loss": 0.7533,
+      "step": 1010
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.4967741935483874e-05,
+      "loss": 0.8498,
+      "step": 1011
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.4903225806451616e-05,
+      "loss": 0.8153,
+      "step": 1012
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.8292,
+      "step": 1013
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.47741935483871e-05,
+      "loss": 0.7558,
+      "step": 1014
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.470967741935484e-05,
+      "loss": 0.7849,
+      "step": 1015
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.464516129032258e-05,
+      "loss": 0.8507,
+      "step": 1016
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.458064516129033e-05,
+      "loss": 0.8206,
+      "step": 1017
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.451612903225806e-05,
+      "loss": 0.9087,
+      "step": 1018
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.445161290322581e-05,
+      "loss": 0.704,
+      "step": 1019
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.438709677419355e-05,
+      "loss": 0.8087,
+      "step": 1020
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.432258064516129e-05,
+      "loss": 0.7544,
+      "step": 1021
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.4258064516129035e-05,
+      "loss": 0.7863,
+      "step": 1022
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.8317,
+      "step": 1023
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.412903225806452e-05,
+      "loss": 0.6947,
+      "step": 1024
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.406451612903226e-05,
+      "loss": 0.6632,
+      "step": 1025
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.7629,
+      "step": 1026
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.393548387096774e-05,
+      "loss": 0.8403,
+      "step": 1027
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.387096774193548e-05,
+      "loss": 0.8377,
+      "step": 1028
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.380645161290323e-05,
+      "loss": 0.6528,
+      "step": 1029
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3741935483870965e-05,
+      "loss": 0.8235,
+      "step": 1030
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.367741935483871e-05,
+      "loss": 0.8121,
+      "step": 1031
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3612903225806454e-05,
+      "loss": 0.8653,
+      "step": 1032
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6668,
+      "step": 1033
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3483870967741936e-05,
+      "loss": 0.8866,
+      "step": 1034
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.341935483870968e-05,
+      "loss": 0.8178,
+      "step": 1035
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.335483870967742e-05,
+      "loss": 0.8656,
+      "step": 1036
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.329032258064516e-05,
+      "loss": 0.9814,
+      "step": 1037
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.7848,
+      "step": 1038
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.316129032258064e-05,
+      "loss": 0.791,
+      "step": 1039
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.309677419354839e-05,
+      "loss": 0.7412,
+      "step": 1040
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.303225806451613e-05,
+      "loss": 0.8625,
+      "step": 1041
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.296774193548387e-05,
+      "loss": 0.7696,
+      "step": 1042
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.7589,
+      "step": 1043
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2838709677419356e-05,
+      "loss": 0.8567,
+      "step": 1044
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.27741935483871e-05,
+      "loss": 0.8983,
+      "step": 1045
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.270967741935484e-05,
+      "loss": 0.8608,
+      "step": 1046
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2645161290322586e-05,
+      "loss": 0.8221,
+      "step": 1047
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.8011,
+      "step": 1048
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.251612903225806e-05,
+      "loss": 0.9114,
+      "step": 1049
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.245161290322581e-05,
+      "loss": 0.8008,
+      "step": 1050
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.238709677419355e-05,
+      "loss": 0.8842,
+      "step": 1051
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.232258064516129e-05,
+      "loss": 0.8412,
+      "step": 1052
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.7607,
+      "step": 1053
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2193548387096775e-05,
+      "loss": 0.8043,
+      "step": 1054
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2129032258064516e-05,
+      "loss": 0.8355,
+      "step": 1055
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.206451612903226e-05,
+      "loss": 0.8703,
+      "step": 1056
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.9075,
+      "step": 1057
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.7907,
+      "step": 1058
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.187096774193549e-05,
+      "loss": 0.8413,
+      "step": 1059
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.180645161290323e-05,
+      "loss": 0.8163,
+      "step": 1060
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.174193548387097e-05,
+      "loss": 0.8375,
+      "step": 1061
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.167741935483871e-05,
+      "loss": 0.8036,
+      "step": 1062
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.8322,
+      "step": 1063
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.1548387096774194e-05,
+      "loss": 0.8598,
+      "step": 1064
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1483870967741935e-05,
+      "loss": 0.8217,
+      "step": 1065
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.141935483870968e-05,
+      "loss": 0.8563,
+      "step": 1066
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.135483870967742e-05,
+      "loss": 0.7079,
+      "step": 1067
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.858,
+      "step": 1068
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.122580645161291e-05,
+      "loss": 0.7648,
+      "step": 1069
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.116129032258064e-05,
+      "loss": 0.8679,
+      "step": 1070
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.109677419354839e-05,
+      "loss": 0.7823,
+      "step": 1071
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.103225806451613e-05,
+      "loss": 0.7459,
+      "step": 1072
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.9583,
+      "step": 1073
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.090322580645161e-05,
+      "loss": 0.9355,
+      "step": 1074
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.083870967741936e-05,
+      "loss": 0.8174,
+      "step": 1075
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0774193548387096e-05,
+      "loss": 0.8091,
+      "step": 1076
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.070967741935484e-05,
+      "loss": 0.8532,
+      "step": 1077
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.7285,
+      "step": 1078
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.058064516129032e-05,
+      "loss": 0.7843,
+      "step": 1079
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0516129032258067e-05,
+      "loss": 0.8159,
+      "step": 1080
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0451612903225805e-05,
+      "loss": 0.8285,
+      "step": 1081
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0387096774193553e-05,
+      "loss": 0.7619,
+      "step": 1082
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.8265,
+      "step": 1083
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0258064516129032e-05,
+      "loss": 0.8809,
+      "step": 1084
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0193548387096777e-05,
+      "loss": 0.8698,
+      "step": 1085
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0129032258064515e-05,
+      "loss": 0.7926,
+      "step": 1086
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.006451612903226e-05,
+      "loss": 0.8833,
+      "step": 1087
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3e-05,
+      "loss": 0.8174,
+      "step": 1088
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.9935483870967745e-05,
+      "loss": 0.6907,
+      "step": 1089
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.9870967741935487e-05,
+      "loss": 0.8872,
+      "step": 1090
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.9806451612903224e-05,
+      "loss": 0.9388,
+      "step": 1091
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.974193548387097e-05,
+      "loss": 0.8372,
+      "step": 1092
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.967741935483871e-05,
+      "loss": 0.8376,
+      "step": 1093
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.9612903225806455e-05,
+      "loss": 0.8409,
+      "step": 1094
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.9548387096774193e-05,
+      "loss": 0.8439,
+      "step": 1095
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9483870967741937e-05,
+      "loss": 0.8417,
+      "step": 1096
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.941935483870968e-05,
+      "loss": 0.8299,
+      "step": 1097
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9354838709677417e-05,
+      "loss": 0.8008,
+      "step": 1098
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9290322580645165e-05,
+      "loss": 0.7473,
+      "step": 1099
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9225806451612902e-05,
+      "loss": 0.9038,
+      "step": 1100
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9161290322580647e-05,
+      "loss": 0.8687,
+      "step": 1101
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.909677419354839e-05,
+      "loss": 0.9336,
+      "step": 1102
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.916,
+      "step": 1103
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.896774193548387e-05,
+      "loss": 0.8089,
+      "step": 1104
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.8903225806451615e-05,
+      "loss": 0.8306,
+      "step": 1105
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.8838709677419357e-05,
+      "loss": 0.7816,
+      "step": 1106
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.8774193548387095e-05,
+      "loss": 0.7792,
+      "step": 1107
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.7225,
+      "step": 1108
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.864516129032258e-05,
+      "loss": 0.9085,
+      "step": 1109
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.8580645161290325e-05,
+      "loss": 0.8679,
+      "step": 1110
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.8516129032258066e-05,
+      "loss": 0.9113,
+      "step": 1111
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.845161290322581e-05,
+      "loss": 0.7766,
+      "step": 1112
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.838709677419355e-05,
+      "loss": 0.8579,
+      "step": 1113
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.832258064516129e-05,
+      "loss": 0.7831,
+      "step": 1114
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.8258064516129035e-05,
+      "loss": 0.7211,
+      "step": 1115
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.8193548387096776e-05,
+      "loss": 0.8509,
+      "step": 1116
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.812903225806452e-05,
+      "loss": 0.9841,
+      "step": 1117
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.806451612903226e-05,
+      "loss": 0.7754,
+      "step": 1118
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.8266,
+      "step": 1119
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.7935483870967744e-05,
+      "loss": 1.0096,
+      "step": 1120
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.7870967741935482e-05,
+      "loss": 0.8194,
+      "step": 1121
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.7806451612903227e-05,
+      "loss": 0.8982,
+      "step": 1122
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.7741935483870968e-05,
+      "loss": 0.877,
+      "step": 1123
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.7677419354838713e-05,
+      "loss": 0.8668,
+      "step": 1124
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.7612903225806454e-05,
+      "loss": 0.7396,
+      "step": 1125
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.75483870967742e-05,
+      "loss": 0.8806,
+      "step": 1126
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.7483870967741936e-05,
+      "loss": 0.9158,
+      "step": 1127
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.9653,
+      "step": 1128
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.7354838709677422e-05,
+      "loss": 0.7757,
+      "step": 1129
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.729032258064516e-05,
+      "loss": 0.8242,
+      "step": 1130
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.7225806451612905e-05,
+      "loss": 0.9322,
+      "step": 1131
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.7161290322580646e-05,
+      "loss": 0.769,
+      "step": 1132
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.709677419354839e-05,
+      "loss": 0.907,
+      "step": 1133
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.7032258064516132e-05,
+      "loss": 0.763,
+      "step": 1134
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.696774193548387e-05,
+      "loss": 0.7924,
+      "step": 1135
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.6903225806451614e-05,
+      "loss": 0.848,
+      "step": 1136
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.6838709677419355e-05,
+      "loss": 0.7867,
+      "step": 1137
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.67741935483871e-05,
+      "loss": 0.9541,
+      "step": 1138
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.6709677419354838e-05,
+      "loss": 0.8296,
+      "step": 1139
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.6645161290322586e-05,
+      "loss": 0.8245,
+      "step": 1140
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.6580645161290324e-05,
+      "loss": 0.7879,
+      "step": 1141
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.6516129032258065e-05,
+      "loss": 0.7777,
+      "step": 1142
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.645161290322581e-05,
+      "loss": 0.8862,
+      "step": 1143
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.6387096774193548e-05,
+      "loss": 0.7877,
+      "step": 1144
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.6322580645161292e-05,
+      "loss": 0.7926,
+      "step": 1145
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6258064516129033e-05,
+      "loss": 0.7622,
+      "step": 1146
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6193548387096778e-05,
+      "loss": 0.8447,
+      "step": 1147
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.8921,
+      "step": 1148
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.6064516129032257e-05,
+      "loss": 0.812,
+      "step": 1149
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.8313,
+      "step": 1150
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.5935483870967743e-05,
+      "loss": 0.8052,
+      "step": 1151
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5870967741935488e-05,
+      "loss": 0.7871,
+      "step": 1152
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.8802,
+      "step": 1153
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.574193548387097e-05,
+      "loss": 0.966,
+      "step": 1154
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.567741935483871e-05,
+      "loss": 0.8239,
+      "step": 1155
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.561290322580645e-05,
+      "loss": 0.7747,
+      "step": 1156
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.5548387096774197e-05,
+      "loss": 0.8897,
+      "step": 1157
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.5483870967741935e-05,
+      "loss": 0.7448,
+      "step": 1158
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.541935483870968e-05,
+      "loss": 0.8005,
+      "step": 1159
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.535483870967742e-05,
+      "loss": 0.7984,
+      "step": 1160
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.5290322580645166e-05,
+      "loss": 0.8553,
+      "step": 1161
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.5225806451612903e-05,
+      "loss": 0.8268,
+      "step": 1162
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.5161290322580645e-05,
+      "loss": 0.7189,
+      "step": 1163
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.509677419354839e-05,
+      "loss": 0.7613,
+      "step": 1164
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.5032258064516127e-05,
+      "loss": 0.8731,
+      "step": 1165
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.4967741935483872e-05,
+      "loss": 0.9225,
+      "step": 1166
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.4903225806451613e-05,
+      "loss": 0.8617,
+      "step": 1167
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.4838709677419354e-05,
+      "loss": 0.8213,
+      "step": 1168
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.47741935483871e-05,
+      "loss": 0.8245,
+      "step": 1169
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.470967741935484e-05,
+      "loss": 0.8335,
+      "step": 1170
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.464516129032258e-05,
+      "loss": 0.7966,
+      "step": 1171
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.4580645161290326e-05,
+      "loss": 0.7708,
+      "step": 1172
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.4516129032258064e-05,
+      "loss": 0.7484,
+      "step": 1173
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.4451612903225805e-05,
+      "loss": 0.7432,
+      "step": 1174
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.438709677419355e-05,
+      "loss": 0.8345,
+      "step": 1175
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.432258064516129e-05,
+      "loss": 0.8352,
+      "step": 1176
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.4258064516129032e-05,
+      "loss": 0.9303,
+      "step": 1177
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.8325,
+      "step": 1178
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.4129032258064518e-05,
+      "loss": 0.9517,
+      "step": 1179
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.406451612903226e-05,
+      "loss": 0.754,
+      "step": 1180
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.4e-05,
+      "loss": 0.6975,
+      "step": 1181
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.3935483870967742e-05,
+      "loss": 0.7833,
+      "step": 1182
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.8658,
+      "step": 1183
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.3806451612903228e-05,
+      "loss": 0.8329,
+      "step": 1184
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.374193548387097e-05,
+      "loss": 0.7778,
+      "step": 1185
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.367741935483871e-05,
+      "loss": 0.7668,
+      "step": 1186
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.361290322580645e-05,
+      "loss": 0.9689,
+      "step": 1187
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3548387096774193e-05,
+      "loss": 0.9555,
+      "step": 1188
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.3483870967741937e-05,
+      "loss": 0.8192,
+      "step": 1189
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.341935483870968e-05,
+      "loss": 0.8178,
+      "step": 1190
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.335483870967742e-05,
+      "loss": 0.9306,
+      "step": 1191
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.3290322580645164e-05,
+      "loss": 0.9246,
+      "step": 1192
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.8442,
+      "step": 1193
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.3161290322580644e-05,
+      "loss": 0.7844,
+      "step": 1194
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.3096774193548388e-05,
+      "loss": 0.7412,
+      "step": 1195
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.303225806451613e-05,
+      "loss": 0.9944,
+      "step": 1196
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.296774193548387e-05,
+      "loss": 0.816,
+      "step": 1197
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.8803,
+      "step": 1198
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.2838709677419357e-05,
+      "loss": 0.7605,
+      "step": 1199
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.2774193548387098e-05,
+      "loss": 0.8659,
+      "step": 1200
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.2709677419354842e-05,
+      "loss": 0.7926,
+      "step": 1201
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.264516129032258e-05,
+      "loss": 0.7595,
+      "step": 1202
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.258064516129032e-05,
+      "loss": 0.7992,
+      "step": 1203
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.2516129032258066e-05,
+      "loss": 0.8144,
+      "step": 1204
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.2451612903225807e-05,
+      "loss": 0.7966,
+      "step": 1205
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.238709677419355e-05,
+      "loss": 0.8103,
+      "step": 1206
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.2322580645161293e-05,
+      "loss": 0.8729,
+      "step": 1207
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.2258064516129034e-05,
+      "loss": 0.8749,
+      "step": 1208
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.2193548387096776e-05,
+      "loss": 0.6836,
+      "step": 1209
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.2129032258064517e-05,
+      "loss": 0.8969,
+      "step": 1210
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.2064516129032258e-05,
+      "loss": 0.8128,
+      "step": 1211
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.8888,
+      "step": 1212
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.1935483870967744e-05,
+      "loss": 0.7695,
+      "step": 1213
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.1870967741935485e-05,
+      "loss": 0.8197,
+      "step": 1214
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.1806451612903227e-05,
+      "loss": 0.887,
+      "step": 1215
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.1741935483870968e-05,
+      "loss": 0.9445,
+      "step": 1216
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.167741935483871e-05,
+      "loss": 0.8773,
+      "step": 1217
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.1612903225806454e-05,
+      "loss": 0.8289,
+      "step": 1218
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.1548387096774195e-05,
+      "loss": 0.8664,
+      "step": 1219
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.1483870967741936e-05,
+      "loss": 0.8889,
+      "step": 1220
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.141935483870968e-05,
+      "loss": 0.8541,
+      "step": 1221
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.1354838709677422e-05,
+      "loss": 0.8165,
+      "step": 1222
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.129032258064516e-05,
+      "loss": 0.9555,
+      "step": 1223
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.1225806451612904e-05,
+      "loss": 0.7988,
+      "step": 1224
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.1161290322580646e-05,
+      "loss": 0.7686,
+      "step": 1225
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.1096774193548387e-05,
+      "loss": 0.9054,
+      "step": 1226
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.103225806451613e-05,
+      "loss": 0.9356,
+      "step": 1227
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.876,
+      "step": 1228
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.0903225806451614e-05,
+      "loss": 0.8883,
+      "step": 1229
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.0838709677419355e-05,
+      "loss": 0.7536,
+      "step": 1230
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.0774193548387097e-05,
+      "loss": 0.9243,
+      "step": 1231
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.0709677419354838e-05,
+      "loss": 0.8693,
+      "step": 1232
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.7926,
+      "step": 1233
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0580645161290324e-05,
+      "loss": 0.8087,
+      "step": 1234
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0516129032258065e-05,
+      "loss": 0.819,
+      "step": 1235
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.045161290322581e-05,
+      "loss": 0.766,
+      "step": 1236
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0387096774193547e-05,
+      "loss": 1.0422,
+      "step": 1237
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.7943,
+      "step": 1238
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0258064516129033e-05,
+      "loss": 0.8539,
+      "step": 1239
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0193548387096775e-05,
+      "loss": 0.876,
+      "step": 1240
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0129032258064516e-05,
+      "loss": 0.7159,
+      "step": 1241
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.006451612903226e-05,
+      "loss": 0.575,
+      "step": 1242
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7151,
+      "step": 1243
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9935483870967743e-05,
+      "loss": 0.7105,
+      "step": 1244
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9870967741935484e-05,
+      "loss": 0.6494,
+      "step": 1245
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9806451612903225e-05,
+      "loss": 0.7084,
+      "step": 1246
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.974193548387097e-05,
+      "loss": 0.5918,
+      "step": 1247
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.5696,
+      "step": 1248
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9612903225806452e-05,
+      "loss": 0.7026,
+      "step": 1249
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9548387096774197e-05,
+      "loss": 0.5106,
+      "step": 1250
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9483870967741935e-05,
+      "loss": 0.5699,
+      "step": 1251
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9419354838709676e-05,
+      "loss": 0.6299,
+      "step": 1252
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6698,
+      "step": 1253
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.9290322580645162e-05,
+      "loss": 0.5369,
+      "step": 1254
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.9225806451612903e-05,
+      "loss": 0.7671,
+      "step": 1255
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.9161290322580648e-05,
+      "loss": 0.8277,
+      "step": 1256
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.909677419354839e-05,
+      "loss": 0.6565,
+      "step": 1257
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.9032258064516127e-05,
+      "loss": 0.5911,
+      "step": 1258
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.896774193548387e-05,
+      "loss": 0.6535,
+      "step": 1259
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8903225806451613e-05,
+      "loss": 0.6463,
+      "step": 1260
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.8838709677419354e-05,
+      "loss": 0.6633,
+      "step": 1261
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.87741935483871e-05,
+      "loss": 0.6885,
+      "step": 1262
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6571,
+      "step": 1263
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.864516129032258e-05,
+      "loss": 0.6919,
+      "step": 1264
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.8580645161290326e-05,
+      "loss": 0.6211,
+      "step": 1265
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.8516129032258064e-05,
+      "loss": 0.6655,
+      "step": 1266
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.845161290322581e-05,
+      "loss": 0.6175,
+      "step": 1267
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6898,
+      "step": 1268
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.832258064516129e-05,
+      "loss": 0.7191,
+      "step": 1269
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.8258064516129032e-05,
+      "loss": 0.6256,
+      "step": 1270
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.8193548387096777e-05,
+      "loss": 0.5733,
+      "step": 1271
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.8129032258064518e-05,
+      "loss": 0.6177,
+      "step": 1272
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.5948,
+      "step": 1273
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5588,
+      "step": 1274
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7935483870967742e-05,
+      "loss": 0.6483,
+      "step": 1275
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7870967741935486e-05,
+      "loss": 0.6773,
+      "step": 1276
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7806451612903228e-05,
+      "loss": 0.6162,
+      "step": 1277
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.7266,
+      "step": 1278
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7677419354838713e-05,
+      "loss": 0.7562,
+      "step": 1279
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.761290322580645e-05,
+      "loss": 0.6161,
+      "step": 1280
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7548387096774193e-05,
+      "loss": 0.7565,
+      "step": 1281
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7483870967741937e-05,
+      "loss": 0.7224,
+      "step": 1282
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.7232,
+      "step": 1283
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.735483870967742e-05,
+      "loss": 0.5573,
+      "step": 1284
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7290322580645164e-05,
+      "loss": 0.4954,
+      "step": 1285
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7225806451612906e-05,
+      "loss": 0.5973,
+      "step": 1286
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7161290322580643e-05,
+      "loss": 0.5319,
+      "step": 1287
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.7503,
+      "step": 1288
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.703225806451613e-05,
+      "loss": 0.6255,
+      "step": 1289
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.696774193548387e-05,
+      "loss": 0.6426,
+      "step": 1290
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.6903225806451615e-05,
+      "loss": 0.6013,
+      "step": 1291
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6838709677419356e-05,
+      "loss": 0.6989,
+      "step": 1292
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6484,
+      "step": 1293
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.670967741935484e-05,
+      "loss": 0.5811,
+      "step": 1294
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.664516129032258e-05,
+      "loss": 0.707,
+      "step": 1295
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.658064516129032e-05,
+      "loss": 0.6887,
+      "step": 1296
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.6516129032258066e-05,
+      "loss": 0.8003,
+      "step": 1297
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.6404,
+      "step": 1298
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.638709677419355e-05,
+      "loss": 0.5392,
+      "step": 1299
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6322580645161293e-05,
+      "loss": 0.6085,
+      "step": 1300
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.625806451612903e-05,
+      "loss": 0.6861,
+      "step": 1301
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6193548387096776e-05,
+      "loss": 0.6786,
+      "step": 1302
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.6375,
+      "step": 1303
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6064516129032258e-05,
+      "loss": 0.5089,
+      "step": 1304
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6864,
+      "step": 1305
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.5935483870967744e-05,
+      "loss": 0.6734,
+      "step": 1306
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.5870967741935485e-05,
+      "loss": 0.6844,
+      "step": 1307
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.6315,
+      "step": 1308
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.5741935483870968e-05,
+      "loss": 0.6081,
+      "step": 1309
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.567741935483871e-05,
+      "loss": 0.6865,
+      "step": 1310
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.5612903225806454e-05,
+      "loss": 0.5783,
+      "step": 1311
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.5548387096774195e-05,
+      "loss": 0.6796,
+      "step": 1312
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.5058,
+      "step": 1313
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.541935483870968e-05,
+      "loss": 0.6751,
+      "step": 1314
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.535483870967742e-05,
+      "loss": 0.6048,
+      "step": 1315
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.529032258064516e-05,
+      "loss": 0.7052,
+      "step": 1316
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.5225806451612903e-05,
+      "loss": 0.7153,
+      "step": 1317
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.6712,
+      "step": 1318
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.5096774193548389e-05,
+      "loss": 0.7057,
+      "step": 1319
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.503225806451613e-05,
+      "loss": 0.6384,
+      "step": 1320
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.4967741935483873e-05,
+      "loss": 0.6823,
+      "step": 1321
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.4903225806451612e-05,
+      "loss": 0.6034,
+      "step": 1322
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.4838709677419355e-05,
+      "loss": 0.6829,
+      "step": 1323
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.4774193548387096e-05,
+      "loss": 0.7492,
+      "step": 1324
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.470967741935484e-05,
+      "loss": 0.5751,
+      "step": 1325
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.4645161290322582e-05,
+      "loss": 0.7502,
+      "step": 1326
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.4580645161290324e-05,
+      "loss": 0.6922,
+      "step": 1327
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.6798,
+      "step": 1328
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4451612903225808e-05,
+      "loss": 0.6495,
+      "step": 1329
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4387096774193547e-05,
+      "loss": 0.6538,
+      "step": 1330
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.432258064516129e-05,
+      "loss": 0.615,
+      "step": 1331
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4258064516129033e-05,
+      "loss": 0.7179,
+      "step": 1332
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4193548387096774e-05,
+      "loss": 0.6048,
+      "step": 1333
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4129032258064517e-05,
+      "loss": 0.6998,
+      "step": 1334
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.406451612903226e-05,
+      "loss": 0.7111,
+      "step": 1335
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.6123,
+      "step": 1336
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3935483870967741e-05,
+      "loss": 0.6423,
+      "step": 1337
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3870967741935484e-05,
+      "loss": 0.677,
+      "step": 1338
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3806451612903227e-05,
+      "loss": 0.656,
+      "step": 1339
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3741935483870968e-05,
+      "loss": 0.5675,
+      "step": 1340
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3677419354838711e-05,
+      "loss": 0.5713,
+      "step": 1341
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3612903225806452e-05,
+      "loss": 0.7718,
+      "step": 1342
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3548387096774195e-05,
+      "loss": 0.7708,
+      "step": 1343
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3483870967741935e-05,
+      "loss": 0.6362,
+      "step": 1344
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3419354838709678e-05,
+      "loss": 0.7056,
+      "step": 1345
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3354838709677419e-05,
+      "loss": 0.5778,
+      "step": 1346
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3290322580645162e-05,
+      "loss": 0.575,
+      "step": 1347
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3225806451612905e-05,
+      "loss": 0.7064,
+      "step": 1348
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3161290322580646e-05,
+      "loss": 0.7055,
+      "step": 1349
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3096774193548389e-05,
+      "loss": 0.4841,
+      "step": 1350
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.3032258064516129e-05,
+      "loss": 0.5905,
+      "step": 1351
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2967741935483872e-05,
+      "loss": 0.6409,
+      "step": 1352
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.8074,
+      "step": 1353
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2838709677419356e-05,
+      "loss": 0.6642,
+      "step": 1354
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2774193548387099e-05,
+      "loss": 0.7098,
+      "step": 1355
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.270967741935484e-05,
+      "loss": 0.6682,
+      "step": 1356
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.2645161290322583e-05,
+      "loss": 0.6304,
+      "step": 1357
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.2580645161290322e-05,
+      "loss": 0.6668,
+      "step": 1358
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.2516129032258064e-05,
+      "loss": 0.5949,
+      "step": 1359
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.2451612903225807e-05,
+      "loss": 0.5863,
+      "step": 1360
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.238709677419355e-05,
+      "loss": 0.6214,
+      "step": 1361
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.232258064516129e-05,
+      "loss": 0.6252,
+      "step": 1362
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.2258064516129032e-05,
+      "loss": 0.6603,
+      "step": 1363
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.2193548387096775e-05,
+      "loss": 0.7043,
+      "step": 1364
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.2129032258064516e-05,
+      "loss": 0.6552,
+      "step": 1365
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.2064516129032259e-05,
+      "loss": 0.7396,
+      "step": 1366
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.2e-05,
+      "loss": 0.5525,
+      "step": 1367
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.6536,
+      "step": 1368
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1870967741935484e-05,
+      "loss": 0.5933,
+      "step": 1369
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1806451612903226e-05,
+      "loss": 0.6237,
+      "step": 1370
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1741935483870969e-05,
+      "loss": 0.6459,
+      "step": 1371
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.167741935483871e-05,
+      "loss": 0.7637,
+      "step": 1372
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7763,
+      "step": 1373
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1548387096774194e-05,
+      "loss": 0.7071,
+      "step": 1374
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1483870967741935e-05,
+      "loss": 0.6611,
+      "step": 1375
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1419354838709678e-05,
+      "loss": 0.569,
+      "step": 1376
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1354838709677421e-05,
+      "loss": 0.5717,
+      "step": 1377
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.129032258064516e-05,
+      "loss": 0.5598,
+      "step": 1378
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1225806451612904e-05,
+      "loss": 0.5784,
+      "step": 1379
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1161290322580647e-05,
+      "loss": 0.6597,
+      "step": 1380
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1096774193548388e-05,
+      "loss": 0.7039,
+      "step": 1381
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1032258064516129e-05,
+      "loss": 0.6883,
+      "step": 1382
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0967741935483872e-05,
+      "loss": 0.5066,
+      "step": 1383
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0903225806451613e-05,
+      "loss": 0.6968,
+      "step": 1384
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0838709677419355e-05,
+      "loss": 0.6048,
+      "step": 1385
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0774193548387097e-05,
+      "loss": 0.6661,
+      "step": 1386
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.070967741935484e-05,
+      "loss": 0.6046,
+      "step": 1387
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.064516129032258e-05,
+      "loss": 0.7513,
+      "step": 1388
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0580645161290323e-05,
+      "loss": 0.5724,
+      "step": 1389
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0516129032258066e-05,
+      "loss": 0.6955,
+      "step": 1390
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0451612903225807e-05,
+      "loss": 0.6688,
+      "step": 1391
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0387096774193548e-05,
+      "loss": 0.5707,
+      "step": 1392
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.6995,
+      "step": 1393
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0258064516129032e-05,
+      "loss": 0.6562,
+      "step": 1394
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0193548387096774e-05,
+      "loss": 0.6199,
+      "step": 1395
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0129032258064517e-05,
+      "loss": 0.7756,
+      "step": 1396
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0064516129032258e-05,
+      "loss": 0.7084,
+      "step": 1397
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1e-05,
+      "loss": 0.6923,
+      "step": 1398
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.935483870967742e-06,
+      "loss": 0.7607,
+      "step": 1399
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.870967741935485e-06,
+      "loss": 0.5394,
+      "step": 1400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.806451612903226e-06,
+      "loss": 0.5582,
+      "step": 1401
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.741935483870967e-06,
+      "loss": 0.5202,
+      "step": 1402
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.6358,
+      "step": 1403
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.612903225806452e-06,
+      "loss": 0.7541,
+      "step": 1404
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.548387096774195e-06,
+      "loss": 0.4452,
+      "step": 1405
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.483870967741936e-06,
+      "loss": 0.5563,
+      "step": 1406
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.419354838709677e-06,
+      "loss": 0.7024,
+      "step": 1407
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.5694,
+      "step": 1408
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.290322580645163e-06,
+      "loss": 0.6731,
+      "step": 1409
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.225806451612904e-06,
+      "loss": 0.5769,
+      "step": 1410
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.161290322580645e-06,
+      "loss": 0.6554,
+      "step": 1411
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.096774193548388e-06,
+      "loss": 0.4904,
+      "step": 1412
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.6401,
+      "step": 1413
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.967741935483871e-06,
+      "loss": 0.5306,
+      "step": 1414
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.903225806451614e-06,
+      "loss": 0.6432,
+      "step": 1415
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.838709677419357e-06,
+      "loss": 0.6156,
+      "step": 1416
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.774193548387096e-06,
+      "loss": 0.6642,
+      "step": 1417
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.626,
+      "step": 1418
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.645161290322582e-06,
+      "loss": 0.6074,
+      "step": 1419
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.580645161290322e-06,
+      "loss": 0.6699,
+      "step": 1420
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.516129032258065e-06,
+      "loss": 0.7005,
+      "step": 1421
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.451612903225808e-06,
+      "loss": 0.7,
+      "step": 1422
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.5915,
+      "step": 1423
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.32258064516129e-06,
+      "loss": 0.6277,
+      "step": 1424
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.258064516129033e-06,
+      "loss": 0.7321,
+      "step": 1425
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.193548387096774e-06,
+      "loss": 0.6235,
+      "step": 1426
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.129032258064515e-06,
+      "loss": 0.6086,
+      "step": 1427
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7046,
+      "step": 1428
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.5526,
+      "step": 1429
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.935483870967743e-06,
+      "loss": 0.7193,
+      "step": 1430
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.870967741935484e-06,
+      "loss": 0.6326,
+      "step": 1431
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.806451612903227e-06,
+      "loss": 0.5135,
+      "step": 1432
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.6619,
+      "step": 1433
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.67741935483871e-06,
+      "loss": 0.6705,
+      "step": 1434
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.612903225806451e-06,
+      "loss": 0.6258,
+      "step": 1435
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.548387096774194e-06,
+      "loss": 0.5937,
+      "step": 1436
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.483870967741936e-06,
+      "loss": 0.5468,
+      "step": 1437
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.419354838709678e-06,
+      "loss": 0.7105,
+      "step": 1438
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.35483870967742e-06,
+      "loss": 0.6655,
+      "step": 1439
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.290322580645162e-06,
+      "loss": 0.664,
+      "step": 1440
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.225806451612904e-06,
+      "loss": 0.7436,
+      "step": 1441
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.161290322580645e-06,
+      "loss": 0.6306,
+      "step": 1442
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.096774193548387e-06,
+      "loss": 0.6058,
+      "step": 1443
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.03225806451613e-06,
+      "loss": 0.6906,
+      "step": 1444
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.9677419354838705e-06,
+      "loss": 0.61,
+      "step": 1445
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.9032258064516135e-06,
+      "loss": 0.5709,
+      "step": 1446
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.8387096774193555e-06,
+      "loss": 0.7571,
+      "step": 1447
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.774193548387098e-06,
+      "loss": 0.6511,
+      "step": 1448
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.709677419354839e-06,
+      "loss": 0.7669,
+      "step": 1449
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.645161290322581e-06,
+      "loss": 0.6246,
+      "step": 1450
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.580645161290323e-06,
+      "loss": 0.7914,
+      "step": 1451
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.516129032258064e-06,
+      "loss": 0.5579,
+      "step": 1452
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.6791,
+      "step": 1453
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.387096774193549e-06,
+      "loss": 0.6929,
+      "step": 1454
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.322580645161291e-06,
+      "loss": 0.6625,
+      "step": 1455
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.258064516129032e-06,
+      "loss": 0.5717,
+      "step": 1456
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.193548387096775e-06,
+      "loss": 0.5767,
+      "step": 1457
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.129032258064516e-06,
+      "loss": 0.6271,
+      "step": 1458
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.064516129032258e-06,
+      "loss": 0.7021,
+      "step": 1459
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6e-06,
+      "loss": 0.758,
+      "step": 1460
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.935483870967742e-06,
+      "loss": 0.7593,
+      "step": 1461
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.870967741935484e-06,
+      "loss": 0.5879,
+      "step": 1462
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8323,
+      "step": 1463
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.741935483870968e-06,
+      "loss": 0.6431,
+      "step": 1464
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.677419354838711e-06,
+      "loss": 0.5894,
+      "step": 1465
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.612903225806452e-06,
+      "loss": 0.6166,
+      "step": 1466
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.548387096774194e-06,
+      "loss": 0.6267,
+      "step": 1467
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.483870967741936e-06,
+      "loss": 0.6469,
+      "step": 1468
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.419354838709677e-06,
+      "loss": 0.683,
+      "step": 1469
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.35483870967742e-06,
+      "loss": 0.5552,
+      "step": 1470
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.2903225806451614e-06,
+      "loss": 0.6176,
+      "step": 1471
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.2258064516129035e-06,
+      "loss": 0.7323,
+      "step": 1472
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.6605,
+      "step": 1473
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.096774193548387e-06,
+      "loss": 0.5557,
+      "step": 1474
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.032258064516129e-06,
+      "loss": 0.6234,
+      "step": 1475
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.967741935483871e-06,
+      "loss": 0.5927,
+      "step": 1476
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.903225806451613e-06,
+      "loss": 0.5407,
+      "step": 1477
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.7392,
+      "step": 1478
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.774193548387097e-06,
+      "loss": 0.5521,
+      "step": 1479
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.7096774193548385e-06,
+      "loss": 0.615,
+      "step": 1480
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.6451612903225815e-06,
+      "loss": 0.5893,
+      "step": 1481
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.580645161290323e-06,
+      "loss": 0.6718,
+      "step": 1482
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.6906,
+      "step": 1483
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.451612903225807e-06,
+      "loss": 0.5744,
+      "step": 1484
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.387096774193548e-06,
+      "loss": 0.5755,
+      "step": 1485
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.322580645161291e-06,
+      "loss": 0.6555,
+      "step": 1486
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.258064516129032e-06,
+      "loss": 0.5446,
+      "step": 1487
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.7418,
+      "step": 1488
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.1290322580645165e-06,
+      "loss": 0.7693,
+      "step": 1489
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.064516129032258e-06,
+      "loss": 0.71,
+      "step": 1490
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.6208,
+      "step": 1491
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.935483870967742e-06,
+      "loss": 0.6198,
+      "step": 1492
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.6497,
+      "step": 1493
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.8064516129032257e-06,
+      "loss": 0.5976,
+      "step": 1494
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.741935483870968e-06,
+      "loss": 0.7461,
+      "step": 1495
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.67741935483871e-06,
+      "loss": 0.5697,
+      "step": 1496
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.612903225806452e-06,
+      "loss": 0.6824,
+      "step": 1497
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.5483870967741936e-06,
+      "loss": 0.7642,
+      "step": 1498
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.4838709677419353e-06,
+      "loss": 0.7262,
+      "step": 1499
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.4193548387096778e-06,
+      "loss": 0.7624,
+      "step": 1500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.3548387096774194e-06,
+      "loss": 0.6021,
+      "step": 1501
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.2903225806451615e-06,
+      "loss": 0.5585,
+      "step": 1502
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.5601,
+      "step": 1503
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.1612903225806457e-06,
+      "loss": 0.6952,
+      "step": 1504
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.0967741935483874e-06,
+      "loss": 0.6282,
+      "step": 1505
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.032258064516129e-06,
+      "loss": 0.5939,
+      "step": 1506
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.967741935483871e-06,
+      "loss": 0.6182,
+      "step": 1507
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.643,
+      "step": 1508
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.8387096774193553e-06,
+      "loss": 0.5113,
+      "step": 1509
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.774193548387097e-06,
+      "loss": 0.6686,
+      "step": 1510
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.7096774193548386e-06,
+      "loss": 0.6625,
+      "step": 1511
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.6451612903225807e-06,
+      "loss": 0.6973,
+      "step": 1512
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.6,
+      "step": 1513
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5161290322580645e-06,
+      "loss": 0.6578,
+      "step": 1514
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4516129032258066e-06,
+      "loss": 0.563,
+      "step": 1515
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3870967741935486e-06,
+      "loss": 0.5886,
+      "step": 1516
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3225806451612907e-06,
+      "loss": 0.5348,
+      "step": 1517
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.5961,
+      "step": 1518
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.193548387096774e-06,
+      "loss": 0.6572,
+      "step": 1519
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.129032258064516e-06,
+      "loss": 0.7702,
+      "step": 1520
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.0645161290322582e-06,
+      "loss": 0.586,
+      "step": 1521
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.5792,
+      "step": 1522
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.7252,
+      "step": 1523
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.870967741935484e-06,
+      "loss": 0.6667,
+      "step": 1524
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.806451612903226e-06,
+      "loss": 0.6491,
+      "step": 1525
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7419354838709676e-06,
+      "loss": 0.615,
+      "step": 1526
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6774193548387097e-06,
+      "loss": 0.7414,
+      "step": 1527
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.6942,
+      "step": 1528
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5483870967741937e-06,
+      "loss": 0.6807,
+      "step": 1529
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.4838709677419356e-06,
+      "loss": 0.5819,
+      "step": 1530
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.4193548387096776e-06,
+      "loss": 0.6764,
+      "step": 1531
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.3548387096774193e-06,
+      "loss": 0.6497,
+      "step": 1532
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.6226,
+      "step": 1533
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2258064516129033e-06,
+      "loss": 0.7171,
+      "step": 1534
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.1612903225806454e-06,
+      "loss": 0.5307,
+      "step": 1535
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.096774193548387e-06,
+      "loss": 0.5491,
+      "step": 1536
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0322580645161291e-06,
+      "loss": 0.8189,
+      "step": 1537
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.6079,
+      "step": 1538
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.03225806451613e-07,
+      "loss": 0.7023,
+      "step": 1539
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 8.387096774193549e-07,
+      "loss": 0.6359,
+      "step": 1540
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.741935483870968e-07,
+      "loss": 0.674,
+      "step": 1541
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.096774193548388e-07,
+      "loss": 0.631,
+      "step": 1542
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.6267,
+      "step": 1543
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.806451612903227e-07,
+      "loss": 0.6578,
+      "step": 1544
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.161290322580646e-07,
+      "loss": 0.6858,
+      "step": 1545
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.516129032258065e-07,
+      "loss": 0.7275,
+      "step": 1546
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.870967741935484e-07,
+      "loss": 0.7238,
+      "step": 1547
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.7413,
+      "step": 1548
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.580645161290323e-07,
+      "loss": 0.6286,
+      "step": 1549
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.935483870967742e-07,
+      "loss": 0.6622,
+      "step": 1550
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1550,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 7.198127478963118e+18,
+  "trial_name": null,
+  "trial_params": null
+}