diff --git "a/weights/model-idf/trainer_state.json" "b/weights/model-idf/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/weights/model-idf/trainer_state.json"
@@ -0,0 +1,12019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.266666666666667,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.7267,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.527570621338671e-07,
+      "loss": 2.056,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5100842157794359e-06,
+      "loss": 2.5016,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9055141242677342e-06,
+      "loss": 2.0518,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.21223339017097e-06,
+      "loss": 2.2693,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6747272279066545e-06,
+      "loss": 1.8096,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.858271186401601e-06,
+      "loss": 1.9145,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0201684315588718e-06,
+      "loss": 1.8638,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1649904523048373e-06,
+      "loss": 1.9306,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.29599790562588e-06,
+      "loss": 1.5686,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4155983400471703e-06,
+      "loss": 1.5126,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.525620074459582e-06,
+      "loss": 1.4775,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.627484290040521e-06,
+      "loss": 1.4659,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7223176059504063e-06,
+      "loss": 1.3156,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8110282485354684e-06,
+      "loss": 1.2849,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.972925493692739e-06,
+      "loss": 1.1538,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.047242937866133e-06,
+      "loss": 1.2709,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.117747514438704e-06,
+      "loss": 1.175,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.18481144368609e-06,
+      "loss": 1.1417,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2487549677597476e-06,
+      "loss": 1.2041,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.309855599633408e-06,
+      "loss": 1.1791,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.368355402181037e-06,
+      "loss": 1.2157,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.42446678034194e-06,
+      "loss": 0.962,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.478377136593449e-06,
+      "loss": 1.2447,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.580241352174388e-06,
+      "loss": 1.0624,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.628475700819906e-06,
+      "loss": 1.0937,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.675074668084273e-06,
+      "loss": 1.1156,
+      "step": 27
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.720145521918643e-06,
+      "loss": 1.0324,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.763785310669335e-06,
+      "loss": 1.1197,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.806082121405316e-06,
+      "loss": 1.1028,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.84711615034489e-06,
+      "loss": 1.0163,
+      "step": 31
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886960618077624e-06,
+      "loss": 1.0116,
+      "step": 32
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9256825558266055e-06,
+      "loss": 1.0057,
+      "step": 33
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9633434839600245e-06,
+      "loss": 0.952,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5e-06,
+      "loss": 1.0232,
+      "step": 35
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5e-06,
+      "loss": 0.9853,
+      "step": 36
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5e-06,
+      "loss": 1.0767,
+      "step": 37
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5e-06,
+      "loss": 0.9957,
+      "step": 38
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5e-06,
+      "loss": 0.961,
+      "step": 39
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5e-06,
+      "loss": 0.9556,
+      "step": 40
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5e-06,
+      "loss": 1.0288,
+      "step": 41
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5e-06,
+      "loss": 1.0222,
+      "step": 42
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5e-06,
+      "loss": 1.04,
+      "step": 43
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5e-06,
+      "loss": 1.0883,
+      "step": 44
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 0.928,
+      "step": 45
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 1.0055,
+      "step": 46
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 1.0105,
+      "step": 47
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 0.9709,
+      "step": 48
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 0.9756,
+      "step": 49
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5e-06,
+      "loss": 0.9484,
+      "step": 50
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5e-06,
+      "loss": 0.8421,
+      "step": 51
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5e-06,
+      "loss": 0.909,
+      "step": 52
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5e-06,
+      "loss": 0.9494,
+      "step": 53
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5e-06,
+      "loss": 0.8658,
+      "step": 54
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5e-06,
+      "loss": 0.9049,
+      "step": 55
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5e-06,
+      "loss": 0.9367,
+      "step": 56
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5e-06,
+      "loss": 0.988,
+      "step": 57
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5e-06,
+      "loss": 0.764,
+      "step": 58
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-06,
+      "loss": 0.9022,
+      "step": 59
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-06,
+      "loss": 0.9307,
+      "step": 60
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-06,
+      "loss": 0.946,
+      "step": 61
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-06,
+      "loss": 0.8029,
+      "step": 62
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-06,
+      "loss": 1.0022,
+      "step": 63
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-06,
+      "loss": 0.8834,
+      "step": 64
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-06,
+      "loss": 0.9628,
+      "step": 65
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-06,
+      "loss": 0.9793,
+      "step": 66
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-06,
+      "loss": 0.8853,
+      "step": 67
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-06,
+      "loss": 0.9443,
+      "step": 68
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-06,
+      "loss": 0.927,
+      "step": 69
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-06,
+      "loss": 0.8621,
+      "step": 70
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-06,
+      "loss": 0.9842,
+      "step": 71
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-06,
+      "loss": 0.7814,
+      "step": 72
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 0.8225,
+      "step": 73
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 0.8403,
+      "step": 74
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 0.8624,
+      "step": 75
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 0.9089,
+      "step": 76
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 0.851,
+      "step": 77
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5e-06,
+      "loss": 0.8483,
+      "step": 78
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5e-06,
+      "loss": 1.0347,
+      "step": 79
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5e-06,
+      "loss": 0.8825,
+      "step": 80
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5e-06,
+      "loss": 0.8992,
+      "step": 81
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5e-06,
+      "loss": 0.9462,
+      "step": 82
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5e-06,
+      "loss": 0.8295,
+      "step": 83
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5e-06,
+      "loss": 0.9095,
+      "step": 84
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5e-06,
+      "loss": 0.8888,
+      "step": 85
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5e-06,
+      "loss": 0.9823,
+      "step": 86
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5e-06,
+      "loss": 0.8133,
+      "step": 87
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5e-06,
+      "loss": 0.8844,
+      "step": 88
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5e-06,
+      "loss": 0.8388,
+      "step": 89
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5e-06,
+      "loss": 0.8632,
+      "step": 90
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5e-06,
+      "loss": 0.965,
+      "step": 91
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-06,
+      "loss": 0.8739,
+      "step": 92
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-06,
+      "loss": 0.8668,
+      "step": 93
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-06,
+      "loss": 0.8455,
+      "step": 94
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-06,
+      "loss": 0.8494,
+      "step": 95
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-06,
+      "loss": 0.9611,
+      "step": 96
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5e-06,
+      "loss": 0.8742,
+      "step": 97
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5e-06,
+      "loss": 0.8517,
+      "step": 98
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5e-06,
+      "loss": 0.8398,
+      "step": 99
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5e-06,
+      "loss": 0.9681,
+      "step": 100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5e-06,
+      "loss": 0.9002,
+      "step": 101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5e-06,
+      "loss": 0.9722,
+      "step": 102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5e-06,
+      "loss": 0.9018,
+      "step": 103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5e-06,
+      "loss": 0.9836,
+      "step": 104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5e-06,
+      "loss": 0.8942,
+      "step": 105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5e-06,
+      "loss": 0.896,
+      "step": 106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5e-06,
+      "loss": 0.843,
+      "step": 107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5e-06,
+      "loss": 0.8508,
+      "step": 108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5e-06,
+      "loss": 0.8787,
+      "step": 109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5e-06,
+      "loss": 0.8242,
+      "step": 110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5e-06,
+      "loss": 0.7959,
+      "step": 111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5e-06,
+      "loss": 0.855,
+      "step": 112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5e-06,
+      "loss": 0.9027,
+      "step": 113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5e-06,
+      "loss": 0.9768,
+      "step": 114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5e-06,
+      "loss": 0.8659,
+      "step": 115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5e-06,
+      "loss": 0.8683,
+      "step": 116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5e-06,
+      "loss": 0.8709,
+      "step": 117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5e-06,
+      "loss": 0.9427,
+      "step": 118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5e-06,
+      "loss": 0.899,
+      "step": 119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5e-06,
+      "loss": 0.8841,
+      "step": 120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5e-06,
+      "loss": 0.8517,
+      "step": 121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5e-06,
+      "loss": 0.7986,
+      "step": 122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5e-06,
+      "loss": 0.7729,
+      "step": 123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5e-06,
+      "loss": 0.8228,
+      "step": 124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5e-06,
+      "loss": 0.8342,
+      "step": 125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5e-06,
+      "loss": 0.8871,
+      "step": 126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5e-06,
+      "loss": 0.8949,
+      "step": 127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5e-06,
+      "loss": 0.8547,
+      "step": 128
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5e-06,
+      "loss": 0.8014,
+      "step": 129
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5e-06,
+      "loss": 0.8116,
+      "step": 130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5e-06,
+      "loss": 0.8674,
+      "step": 131
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5e-06,
+      "loss": 0.7412,
+      "step": 132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5e-06,
+      "loss": 0.8609,
+      "step": 133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5e-06,
+      "loss": 0.8149,
+      "step": 134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5e-06,
+      "loss": 0.8834,
+      "step": 135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5e-06,
+      "loss": 0.8782,
+      "step": 136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5e-06,
+      "loss": 0.8793,
+      "step": 137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5e-06,
+      "loss": 0.8776,
+      "step": 138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5e-06,
+      "loss": 0.8055,
+      "step": 139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5e-06,
+      "loss": 0.8162,
+      "step": 140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5e-06,
+      "loss": 0.7813,
+      "step": 141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5e-06,
+      "loss": 0.9051,
+      "step": 142
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5e-06,
+      "loss": 0.8108,
+      "step": 143
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5e-06,
+      "loss": 0.8084,
+      "step": 144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5e-06,
+      "loss": 0.8745,
+      "step": 145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5e-06,
+      "loss": 0.8192,
+      "step": 146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5e-06,
+      "loss": 0.8342,
+      "step": 147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5e-06,
+      "loss": 0.7906,
+      "step": 148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5e-06,
+      "loss": 0.8166,
+      "step": 149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5e-06,
+      "loss": 0.8481,
+      "step": 150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5e-06,
+      "loss": 0.807,
+      "step": 151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5e-06,
+      "loss": 0.8484,
+      "step": 152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.82,
+      "step": 153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.848,
+      "step": 154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.8249,
+      "step": 155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.8516,
+      "step": 156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5e-06,
+      "loss": 0.7795,
+      "step": 157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5e-06,
+      "loss": 0.8061,
+      "step": 158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5e-06,
+      "loss": 0.8955,
+      "step": 159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5e-06,
+      "loss": 0.8908,
+      "step": 160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5e-06,
+      "loss": 0.8966,
+      "step": 161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5e-06,
+      "loss": 0.857,
+      "step": 162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5e-06,
+      "loss": 0.902,
+      "step": 163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5e-06,
+      "loss": 0.8399,
+      "step": 164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5e-06,
+      "loss": 0.8449,
+      "step": 165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5e-06,
+      "loss": 0.8647,
+      "step": 166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5e-06,
+      "loss": 0.8149,
+      "step": 167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5e-06,
+      "loss": 0.8221,
+      "step": 168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5e-06,
+      "loss": 0.9005,
+      "step": 169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5e-06,
+      "loss": 0.8071,
+      "step": 170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5e-06,
+      "loss": 0.7991,
+      "step": 171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5e-06,
+      "loss": 0.8526,
+      "step": 172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5e-06,
+      "loss": 0.814,
+      "step": 173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5e-06,
+      "loss": 0.8492,
+      "step": 174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5e-06,
+      "loss": 0.8521,
+      "step": 175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5e-06,
+      "loss": 0.9527,
+      "step": 176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5e-06,
+      "loss": 0.7697,
+      "step": 177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5e-06,
+      "loss": 0.7803,
+      "step": 178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5e-06,
+      "loss": 0.8186,
+      "step": 179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5e-06,
+      "loss": 0.8415,
+      "step": 180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5e-06,
+      "loss": 0.7976,
+      "step": 181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5e-06,
+      "loss": 0.7741,
+      "step": 182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5e-06,
+      "loss": 0.8984,
+      "step": 183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5e-06,
+      "loss": 0.9106,
+      "step": 184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5e-06,
+      "loss": 0.8973,
+      "step": 185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5e-06,
+      "loss": 0.8228,
+      "step": 186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5e-06,
+      "loss": 0.9127,
+      "step": 187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5e-06,
+      "loss": 0.8722,
+      "step": 188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5e-06,
+      "loss": 0.847,
+      "step": 189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5e-06,
+      "loss": 0.8634,
+      "step": 190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5e-06,
+      "loss": 0.8204,
+      "step": 191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5e-06,
+      "loss": 0.867,
+      "step": 192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5e-06,
+      "loss": 0.8517,
+      "step": 193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5e-06,
+      "loss": 0.8658,
+      "step": 194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5e-06,
+      "loss": 0.8293,
+      "step": 195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5e-06,
+      "loss": 0.796,
+      "step": 196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5e-06,
+      "loss": 0.8854,
+      "step": 197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5e-06,
+      "loss": 0.858,
+      "step": 198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5e-06,
+      "loss": 0.8093,
+      "step": 199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5e-06,
+      "loss": 0.7751,
+      "step": 200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5e-06,
+      "loss": 0.8852,
+      "step": 201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5e-06,
+      "loss": 0.7735,
+      "step": 202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5e-06,
+      "loss": 0.8289,
+      "step": 203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 0.8218,
+      "step": 204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 0.8307,
+      "step": 205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 0.8871,
+      "step": 206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 0.7681,
+      "step": 207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 0.7679,
+      "step": 208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-06,
+      "loss": 0.7806,
+      "step": 209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-06,
+      "loss": 0.8665,
+      "step": 210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-06,
+      "loss": 0.765,
+      "step": 211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-06,
+      "loss": 0.9078,
+      "step": 212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-06,
+      "loss": 0.8837,
+      "step": 213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5e-06,
+      "loss": 0.8407,
+      "step": 214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5e-06,
+      "loss": 0.8257,
+      "step": 215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5e-06,
+      "loss": 0.7901,
+      "step": 216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5e-06,
+      "loss": 0.783,
+      "step": 217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5e-06,
+      "loss": 0.7965,
+      "step": 218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5e-06,
+      "loss": 0.7798,
+      "step": 219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5e-06,
+      "loss": 0.9033,
+      "step": 220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5e-06,
+      "loss": 0.7715,
+      "step": 221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5e-06,
+      "loss": 0.8501,
+      "step": 222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5e-06,
+      "loss": 0.8147,
+      "step": 223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5e-06,
+      "loss": 0.8583,
+      "step": 224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5e-06,
+      "loss": 0.8312,
+      "step": 225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5e-06,
+      "loss": 0.8037,
+      "step": 226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5e-06,
+      "loss": 0.8618,
+      "step": 227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5e-06,
+      "loss": 0.8174,
+      "step": 228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5e-06,
+      "loss": 0.9186,
+      "step": 229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5e-06,
+      "loss": 0.7927,
+      "step": 230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5e-06,
+      "loss": 0.8669,
+      "step": 231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5e-06,
+      "loss": 0.7804,
+      "step": 232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5e-06,
+      "loss": 0.7574,
+      "step": 233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5e-06,
+      "loss": 0.827,
+      "step": 234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5e-06,
+      "loss": 0.8394,
+      "step": 235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5e-06,
+      "loss": 0.7799,
+      "step": 236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 0.7752,
+      "step": 237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 0.7785,
+      "step": 238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 0.8163,
+      "step": 239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 0.8873,
+      "step": 240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 0.762,
+      "step": 241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.7799,
+      "step": 242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.8361,
+      "step": 243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.8452,
+      "step": 244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.8922,
+      "step": 245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.8483,
+      "step": 246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 0.8506,
+      "step": 247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 0.8605,
+      "step": 248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 0.7764,
+      "step": 249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 0.7985,
+      "step": 250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5e-06,
+      "loss": 0.7833,
+      "step": 251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5e-06,
+      "loss": 0.8415,
+      "step": 252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5e-06,
+      "loss": 0.8506,
+      "step": 253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5e-06,
+      "loss": 0.8643,
+      "step": 254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5e-06,
+      "loss": 0.8264,
+      "step": 255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.7925,
+      "step": 256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.7997,
+      "step": 257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.8371,
+      "step": 258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.8164,
+      "step": 259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 0.8678,
+      "step": 260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5e-06,
+      "loss": 0.8202,
+      "step": 261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5e-06,
+      "loss": 0.7688,
+      "step": 262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5e-06,
+      "loss": 0.8932,
+      "step": 263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5e-06,
+      "loss": 0.7976,
+      "step": 264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7852,
+      "step": 265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7746,
+      "step": 266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7961,
+      "step": 267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7203,
+      "step": 268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7479,
+      "step": 269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5e-06,
+      "loss": 0.7464,
+      "step": 270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5e-06,
+      "loss": 0.8631,
+      "step": 271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5e-06,
+      "loss": 0.7908,
+      "step": 272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5e-06,
+      "loss": 0.7126,
+      "step": 273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5e-06,
+      "loss": 0.779,
+      "step": 274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5e-06,
+      "loss": 0.7716,
+      "step": 275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5e-06,
+      "loss": 0.7714,
+      "step": 276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5e-06,
+      "loss": 0.785,
+      "step": 277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5e-06,
+      "loss": 0.8476,
+      "step": 278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5e-06,
+      "loss": 0.7805,
+      "step": 279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5e-06,
+      "loss": 0.7421,
+      "step": 280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5e-06,
+      "loss": 0.7945,
+      "step": 281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5e-06,
+      "loss": 0.8065,
+      "step": 282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5e-06,
+      "loss": 0.7397,
+      "step": 283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-06,
+      "loss": 0.8748,
+      "step": 284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-06,
+      "loss": 0.7774,
+      "step": 285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-06,
+      "loss": 0.8375,
+      "step": 286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-06,
+      "loss": 0.8304,
+      "step": 287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-06,
+      "loss": 0.8527,
+      "step": 288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5e-06,
+      "loss": 0.8307,
+      "step": 289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5e-06,
+      "loss": 0.8221,
+      "step": 290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5e-06,
+      "loss": 0.754,
+      "step": 291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5e-06,
+      "loss": 0.8281,
+      "step": 292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5e-06,
+      "loss": 0.82,
+      "step": 293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5e-06,
+      "loss": 0.7871,
+      "step": 294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5e-06,
+      "loss": 0.8627,
+      "step": 295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5e-06,
+      "loss": 0.8357,
+      "step": 296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5e-06,
+      "loss": 0.7652,
+      "step": 297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 0.8342,
+      "step": 298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 0.826,
+      "step": 299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 0.7935,
+      "step": 300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 0.876,
+      "step": 301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5e-06,
+      "loss": 0.8231,
+      "step": 302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5e-06,
+      "loss": 0.8711,
+      "step": 303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5e-06,
+      "loss": 0.8295,
+      "step": 304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5e-06,
+      "loss": 0.6819,
+      "step": 305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5e-06,
+      "loss": 0.8086,
+      "step": 306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5e-06,
+      "loss": 0.7869,
+      "step": 307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5e-06,
+      "loss": 0.9321,
+      "step": 308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5e-06,
+      "loss": 0.8835,
+      "step": 309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5e-06,
+      "loss": 0.8117,
+      "step": 310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5e-06,
+      "loss": 0.7907,
+      "step": 311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5e-06,
+      "loss": 0.7854,
+      "step": 312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5e-06,
+      "loss": 0.824,
+      "step": 313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5e-06,
+      "loss": 0.8162,
+      "step": 314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5e-06,
+      "loss": 0.7934,
+      "step": 315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5e-06,
+      "loss": 0.8409,
+      "step": 316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7804,
+      "step": 317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7581,
+      "step": 318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7583,
+      "step": 319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7925,
+      "step": 320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7735,
+      "step": 321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5e-06,
+      "loss": 0.7384,
+      "step": 322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5e-06,
+      "loss": 0.7675,
+      "step": 323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5e-06,
+      "loss": 0.7893,
+      "step": 324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5e-06,
+      "loss": 0.832,
+      "step": 325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5e-06,
+      "loss": 0.874,
+      "step": 326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5e-06,
+      "loss": 0.798,
+      "step": 327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5e-06,
+      "loss": 0.7945,
+      "step": 328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5e-06,
+      "loss": 0.8026,
+      "step": 329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5e-06,
+      "loss": 0.7747,
+      "step": 330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5e-06,
+      "loss": 0.8223,
+      "step": 331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5e-06,
+      "loss": 0.7928,
+      "step": 332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5e-06,
+      "loss": 0.8238,
+      "step": 333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5e-06,
+      "loss": 0.8359,
+      "step": 334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5e-06,
+      "loss": 0.8898,
+      "step": 335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5e-06,
+      "loss": 0.8414,
+      "step": 336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5e-06,
+      "loss": 0.7565,
+      "step": 337
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5e-06,
+      "loss": 0.8034,
+      "step": 338
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5e-06,
+      "loss": 0.7576,
+      "step": 339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5e-06,
+      "loss": 0.8405,
+      "step": 340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5e-06,
+      "loss": 0.8232,
+      "step": 341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5e-06,
+      "loss": 0.7523,
+      "step": 342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5e-06,
+      "loss": 0.787,
+      "step": 343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5e-06,
+      "loss": 0.7854,
+      "step": 344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5e-06,
+      "loss": 0.7948,
+      "step": 345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5e-06,
+      "loss": 0.7727,
+      "step": 346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5e-06,
+      "loss": 0.706,
+      "step": 347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5e-06,
+      "loss": 0.7266,
+      "step": 348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5e-06,
+      "loss": 0.7527,
+      "step": 349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5e-06,
+      "loss": 0.7969,
+      "step": 350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5e-06,
+      "loss": 0.796,
+      "step": 351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5e-06,
+      "loss": 0.8334,
+      "step": 352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5e-06,
+      "loss": 0.8263,
+      "step": 353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5e-06,
+      "loss": 0.8632,
+      "step": 354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5e-06,
+      "loss": 0.8125,
+      "step": 355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5e-06,
+      "loss": 0.7745,
+      "step": 356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5e-06,
+      "loss": 0.826,
+      "step": 357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5e-06,
+      "loss": 0.7721,
+      "step": 358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5e-06,
+      "loss": 0.8339,
+      "step": 359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5e-06,
+      "loss": 0.7562,
+      "step": 360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5e-06,
+      "loss": 0.8442,
+      "step": 361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5e-06,
+      "loss": 0.8748,
+      "step": 362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5e-06,
+      "loss": 0.8911,
+      "step": 363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5e-06,
+      "loss": 0.7833,
+      "step": 364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5e-06,
+      "loss": 0.7545,
+      "step": 365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5e-06,
+      "loss": 0.806,
+      "step": 366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5e-06,
+      "loss": 0.7486,
+      "step": 367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5e-06,
+      "loss": 0.8192,
+      "step": 368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5e-06,
+      "loss": 0.7854,
+      "step": 369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5e-06,
+      "loss": 0.7517,
+      "step": 370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5e-06,
+      "loss": 0.7561,
+      "step": 371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5e-06,
+      "loss": 0.7919,
+      "step": 372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5e-06,
+      "loss": 0.8029,
+      "step": 373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5e-06,
+      "loss": 0.8719,
+      "step": 374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5e-06,
+      "loss": 0.8472,
+      "step": 375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5e-06,
+      "loss": 0.8205,
+      "step": 376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5e-06,
+      "loss": 0.7317,
+      "step": 377
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5e-06,
+      "loss": 0.8098,
+      "step": 378
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5e-06,
+      "loss": 0.8471,
+      "step": 379
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5e-06,
+      "loss": 0.7641,
+      "step": 380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5e-06,
+      "loss": 0.7932,
+      "step": 381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5e-06,
+      "loss": 0.7417,
+      "step": 382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5e-06,
+      "loss": 0.7451,
+      "step": 383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5e-06,
+      "loss": 0.8708,
+      "step": 384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5e-06,
+      "loss": 0.7414,
+      "step": 385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5e-06,
+      "loss": 0.7832,
+      "step": 386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5e-06,
+      "loss": 0.8284,
+      "step": 387
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5e-06,
+      "loss": 0.7331,
+      "step": 388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5e-06,
+      "loss": 0.8574,
+      "step": 389
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5e-06,
+      "loss": 0.7274,
+      "step": 390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5e-06,
+      "loss": 0.829,
+      "step": 391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5e-06,
+      "loss": 0.7692,
+      "step": 392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5e-06,
+      "loss": 0.7103,
+      "step": 393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5e-06,
+      "loss": 0.6986,
+      "step": 394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5e-06,
+      "loss": 0.7823,
+      "step": 395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5e-06,
+      "loss": 0.8224,
+      "step": 396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5e-06,
+      "loss": 0.776,
+      "step": 397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5e-06,
+      "loss": 0.8145,
+      "step": 398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5e-06,
+      "loss": 0.8366,
+      "step": 399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5e-06,
+      "loss": 0.7337,
+      "step": 400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5e-06,
+      "loss": 0.7277,
+      "step": 401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5e-06,
+      "loss": 0.8287,
+      "step": 402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5e-06,
+      "loss": 0.7477,
+      "step": 403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5e-06,
+      "loss": 0.8215,
+      "step": 404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5e-06,
+      "loss": 0.7958,
+      "step": 405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5e-06,
+      "loss": 0.8034,
+      "step": 406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5e-06,
+      "loss": 0.814,
+      "step": 407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5e-06,
+      "loss": 0.8338,
+      "step": 408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5e-06,
+      "loss": 0.7919,
+      "step": 409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5e-06,
+      "loss": 0.7046,
+      "step": 410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5e-06,
+      "loss": 0.7669,
+      "step": 411
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5e-06,
+      "loss": 0.8322,
+      "step": 412
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5e-06,
+      "loss": 0.7972,
+      "step": 413
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5e-06,
+      "loss": 0.7056,
+      "step": 414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5e-06,
+      "loss": 0.7538,
+      "step": 415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5e-06,
+      "loss": 0.8051,
+      "step": 416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5e-06,
+      "loss": 0.7669,
+      "step": 417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5e-06,
+      "loss": 0.7365,
+      "step": 418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5e-06,
+      "loss": 0.8048,
+      "step": 419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.7535,
+      "step": 420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.7603,
+      "step": 421
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.6702,
+      "step": 422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.8161,
+      "step": 423
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.8558,
+      "step": 424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5e-06,
+      "loss": 0.8582,
+      "step": 425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5e-06,
+      "loss": 0.7737,
+      "step": 426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5e-06,
+      "loss": 0.7348,
+      "step": 427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5e-06,
+      "loss": 0.8166,
+      "step": 428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5e-06,
+      "loss": 0.7272,
+      "step": 429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5e-06,
+      "loss": 0.7803,
+      "step": 430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5e-06,
+      "loss": 0.7663,
+      "step": 431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5e-06,
+      "loss": 0.7876,
+      "step": 432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5e-06,
+      "loss": 0.7525,
+      "step": 433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5e-06,
+      "loss": 0.8254,
+      "step": 434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5e-06,
+      "loss": 0.7806,
+      "step": 435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5e-06,
+      "loss": 0.7832,
+      "step": 436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5e-06,
+      "loss": 0.6899,
+      "step": 437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5e-06,
+      "loss": 0.7649,
+      "step": 438
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5e-06,
+      "loss": 0.7634,
+      "step": 439
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5e-06,
+      "loss": 0.7601,
+      "step": 440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5e-06,
+      "loss": 0.7773,
+      "step": 441
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5e-06,
+      "loss": 0.7549,
+      "step": 442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5e-06,
+      "loss": 0.8136,
+      "step": 443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5e-06,
+      "loss": 0.8326,
+      "step": 444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5e-06,
+      "loss": 0.8208,
+      "step": 445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5e-06,
+      "loss": 0.8215,
+      "step": 446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5e-06,
+      "loss": 0.7129,
+      "step": 447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5e-06,
+      "loss": 0.7912,
+      "step": 448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5e-06,
+      "loss": 0.8633,
+      "step": 449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5e-06,
+      "loss": 0.8228,
+      "step": 450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5e-06,
+      "loss": 0.7657,
+      "step": 451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5e-06,
+      "loss": 0.8326,
+      "step": 452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5e-06,
+      "loss": 0.7321,
+      "step": 453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5e-06,
+      "loss": 0.7589,
+      "step": 454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5e-06,
+      "loss": 0.7066,
+      "step": 455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5e-06,
+      "loss": 0.8139,
+      "step": 456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5e-06,
+      "loss": 0.8437,
+      "step": 457
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5e-06,
+      "loss": 0.8668,
+      "step": 458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5e-06,
+      "loss": 0.8295,
+      "step": 459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5e-06,
+      "loss": 0.8448,
+      "step": 460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5e-06,
+      "loss": 0.8171,
+      "step": 461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5e-06,
+      "loss": 0.732,
+      "step": 462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5e-06,
+      "loss": 0.851,
+      "step": 463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5e-06,
+      "loss": 0.7838,
+      "step": 464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5e-06,
+      "loss": 0.8084,
+      "step": 465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5e-06,
+      "loss": 0.8418,
+      "step": 466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 0.7488,
+      "step": 467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 0.8164,
+      "step": 468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 0.7547,
+      "step": 469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 0.7027,
+      "step": 470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6493,
+      "step": 471
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5e-06,
+      "loss": 0.6637,
+      "step": 472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5e-06,
+      "loss": 0.7166,
+      "step": 473
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5e-06,
+      "loss": 0.695,
+      "step": 474
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5e-06,
+      "loss": 0.7113,
+      "step": 475
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5e-06,
+      "loss": 0.7167,
+      "step": 476
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5e-06,
+      "loss": 0.6969,
+      "step": 477
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5e-06,
+      "loss": 0.6721,
+      "step": 478
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5e-06,
+      "loss": 0.7113,
+      "step": 479
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5e-06,
+      "loss": 0.6327,
+      "step": 480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6675,
+      "step": 481
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6252,
+      "step": 482
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6951,
+      "step": 483
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6406,
+      "step": 484
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6212,
+      "step": 485
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5e-06,
+      "loss": 0.688,
+      "step": 486
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5e-06,
+      "loss": 0.721,
+      "step": 487
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5e-06,
+      "loss": 0.6638,
+      "step": 488
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5e-06,
+      "loss": 0.5704,
+      "step": 489
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5e-06,
+      "loss": 0.6826,
+      "step": 490
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5e-06,
+      "loss": 0.6729,
+      "step": 491
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5e-06,
+      "loss": 0.6846,
+      "step": 492
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5e-06,
+      "loss": 0.7347,
+      "step": 493
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5e-06,
+      "loss": 0.6682,
+      "step": 494
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5e-06,
+      "loss": 0.7449,
+      "step": 495
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5e-06,
+      "loss": 0.6397,
+      "step": 496
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5e-06,
+      "loss": 0.6795,
+      "step": 497
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5e-06,
+      "loss": 0.641,
+      "step": 498
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5e-06,
+      "loss": 0.6674,
+      "step": 499
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5e-06,
+      "loss": 0.7313,
+      "step": 500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5e-06,
+      "loss": 0.6553,
+      "step": 501
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5e-06,
+      "loss": 0.6551,
+      "step": 502
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5e-06,
+      "loss": 0.5606,
+      "step": 503
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5e-06,
+      "loss": 0.676,
+      "step": 504
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5e-06,
+      "loss": 0.7232,
+      "step": 505
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5e-06,
+      "loss": 0.6153,
+      "step": 506
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5e-06,
+      "loss": 0.6494,
+      "step": 507
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5e-06,
+      "loss": 0.6926,
+      "step": 508
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5e-06,
+      "loss": 0.7134,
+      "step": 509
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5e-06,
+      "loss": 0.6983,
+      "step": 510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5e-06,
+      "loss": 0.6461,
+      "step": 511
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5e-06,
+      "loss": 0.6971,
+      "step": 512
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5e-06,
+      "loss": 0.7545,
+      "step": 513
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5e-06,
+      "loss": 0.6954,
+      "step": 514
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5e-06,
+      "loss": 0.675,
+      "step": 515
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5e-06,
+      "loss": 0.66,
+      "step": 516
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5e-06,
+      "loss": 0.6876,
+      "step": 517
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5e-06,
+      "loss": 0.6067,
+      "step": 518
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5e-06,
+      "loss": 0.6764,
+      "step": 519
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5e-06,
+      "loss": 0.6842,
+      "step": 520
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5e-06,
+      "loss": 0.6717,
+      "step": 521
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5e-06,
+      "loss": 0.6042,
+      "step": 522
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5e-06,
+      "loss": 0.668,
+      "step": 523
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5e-06,
+      "loss": 0.6601,
+      "step": 524
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5e-06,
+      "loss": 0.7059,
+      "step": 525
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5e-06,
+      "loss": 0.6043,
+      "step": 526
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5e-06,
+      "loss": 0.6525,
+      "step": 527
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5e-06,
+      "loss": 0.7233,
+      "step": 528
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5e-06,
+      "loss": 0.7,
+      "step": 529
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5e-06,
+      "loss": 0.6649,
+      "step": 530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5e-06,
+      "loss": 0.6509,
+      "step": 531
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5e-06,
+      "loss": 0.653,
+      "step": 532
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5e-06,
+      "loss": 0.6571,
+      "step": 533
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5e-06,
+      "loss": 0.6933,
+      "step": 534
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5e-06,
+      "loss": 0.598,
+      "step": 535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5e-06,
+      "loss": 0.642,
+      "step": 536
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5e-06,
+      "loss": 0.6209,
+      "step": 537
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5e-06,
+      "loss": 0.6201,
+      "step": 538
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5e-06,
+      "loss": 0.6363,
+      "step": 539
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5e-06,
+      "loss": 0.6848,
+      "step": 540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5e-06,
+      "loss": 0.6422,
+      "step": 541
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5e-06,
+      "loss": 0.6751,
+      "step": 542
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5e-06,
+      "loss": 0.6553,
+      "step": 543
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5e-06,
+      "loss": 0.6706,
+      "step": 544
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5e-06,
+      "loss": 0.6588,
+      "step": 545
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5e-06,
+      "loss": 0.6374,
+      "step": 546
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5e-06,
+      "loss": 0.6515,
+      "step": 547
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5e-06,
+      "loss": 0.6464,
+      "step": 548
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5e-06,
+      "loss": 0.6378,
+      "step": 549
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5e-06,
+      "loss": 0.6681,
+      "step": 550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5e-06,
+      "loss": 0.5781,
+      "step": 551
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5e-06,
+      "loss": 0.7138,
+      "step": 552
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5e-06,
+      "loss": 0.6881,
+      "step": 553
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5e-06,
+      "loss": 0.5963,
+      "step": 554
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5e-06,
+      "loss": 0.702,
+      "step": 555
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5e-06,
+      "loss": 0.6905,
+      "step": 556
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5e-06,
+      "loss": 0.6648,
+      "step": 557
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5e-06,
+      "loss": 0.6278,
+      "step": 558
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5e-06,
+      "loss": 0.6839,
+      "step": 559
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5e-06,
+      "loss": 0.6527,
+      "step": 560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5e-06,
+      "loss": 0.6479,
+      "step": 561
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5e-06,
+      "loss": 0.7339,
+      "step": 562
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5e-06,
+      "loss": 0.6351,
+      "step": 563
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5e-06,
+      "loss": 0.6982,
+      "step": 564
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5e-06,
+      "loss": 0.645,
+      "step": 565
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5e-06,
+      "loss": 0.6757,
+      "step": 566
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5e-06,
+      "loss": 0.6112,
+      "step": 567
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5e-06,
+      "loss": 0.6126,
+      "step": 568
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5e-06,
+      "loss": 0.6979,
+      "step": 569
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5e-06,
+      "loss": 0.7065,
+      "step": 570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5e-06,
+      "loss": 0.6637,
+      "step": 571
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5e-06,
+      "loss": 0.6836,
+      "step": 572
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5e-06,
+      "loss": 0.6816,
+      "step": 573
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5e-06,
+      "loss": 0.6499,
+      "step": 574
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5e-06,
+      "loss": 0.6661,
+      "step": 575
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5e-06,
+      "loss": 0.6163,
+      "step": 576
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5e-06,
+      "loss": 0.6908,
+      "step": 577
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5e-06,
+      "loss": 0.7166,
+      "step": 578
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5e-06,
+      "loss": 0.6433,
+      "step": 579
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5e-06,
+      "loss": 0.6364,
+      "step": 580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5e-06,
+      "loss": 0.6434,
+      "step": 581
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5e-06,
+      "loss": 0.5645,
+      "step": 582
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5e-06,
+      "loss": 0.6546,
+      "step": 583
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5e-06,
+      "loss": 0.5976,
+      "step": 584
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5e-06,
+      "loss": 0.6154,
+      "step": 585
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5e-06,
+      "loss": 0.6784,
+      "step": 586
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5e-06,
+      "loss": 0.636,
+      "step": 587
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5e-06,
+      "loss": 0.6433,
+      "step": 588
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5e-06,
+      "loss": 0.6084,
+      "step": 589
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5e-06,
+      "loss": 0.6789,
+      "step": 590
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5e-06,
+      "loss": 0.6231,
+      "step": 591
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5e-06,
+      "loss": 0.637,
+      "step": 592
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5e-06,
+      "loss": 0.6264,
+      "step": 593
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5e-06,
+      "loss": 0.7176,
+      "step": 594
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5e-06,
+      "loss": 0.6586,
+      "step": 595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5e-06,
+      "loss": 0.6445,
+      "step": 596
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5e-06,
+      "loss": 0.6616,
+      "step": 597
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 0.6489,
+      "step": 598
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 0.6486,
+      "step": 599
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 0.6292,
+      "step": 600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 0.635,
+      "step": 601
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 0.7002,
+      "step": 602
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5e-06,
+      "loss": 0.6997,
+      "step": 603
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5e-06,
+      "loss": 0.7209,
+      "step": 604
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5e-06,
+      "loss": 0.6589,
+      "step": 605
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5e-06,
+      "loss": 0.6831,
+      "step": 606
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5e-06,
+      "loss": 0.7048,
+      "step": 607
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5e-06,
+      "loss": 0.5766,
+      "step": 608
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5e-06,
+      "loss": 0.658,
+      "step": 609
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5e-06,
+      "loss": 0.6901,
+      "step": 610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5e-06,
+      "loss": 0.6785,
+      "step": 611
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5e-06,
+      "loss": 0.6726,
+      "step": 612
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5e-06,
+      "loss": 0.6277,
+      "step": 613
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5e-06,
+      "loss": 0.7384,
+      "step": 614
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5e-06,
+      "loss": 0.6855,
+      "step": 615
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5e-06,
+      "loss": 0.659,
+      "step": 616
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5e-06,
+      "loss": 0.6434,
+      "step": 617
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5e-06,
+      "loss": 0.6208,
+      "step": 618
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5e-06,
+      "loss": 0.6439,
+      "step": 619
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5e-06,
+      "loss": 0.6355,
+      "step": 620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5e-06,
+      "loss": 0.6109,
+      "step": 621
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5e-06,
+      "loss": 0.6185,
+      "step": 622
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5e-06,
+      "loss": 0.6391,
+      "step": 623
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5e-06,
+      "loss": 0.6783,
+      "step": 624
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5e-06,
+      "loss": 0.7121,
+      "step": 625
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5e-06,
+      "loss": 0.6122,
+      "step": 626
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5e-06,
+      "loss": 0.5999,
+      "step": 627
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5e-06,
+      "loss": 0.6984,
+      "step": 628
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5e-06,
+      "loss": 0.6565,
+      "step": 629
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5e-06,
+      "loss": 0.6737,
+      "step": 630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5e-06,
+      "loss": 0.6447,
+      "step": 631
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5e-06,
+      "loss": 0.6637,
+      "step": 632
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5e-06,
+      "loss": 0.6724,
+      "step": 633
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5e-06,
+      "loss": 0.6337,
+      "step": 634
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5e-06,
+      "loss": 0.6672,
+      "step": 635
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5e-06,
+      "loss": 0.5569,
+      "step": 636
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5e-06,
+      "loss": 0.6711,
+      "step": 637
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5e-06,
+      "loss": 0.66,
+      "step": 638
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5e-06,
+      "loss": 0.5833,
+      "step": 639
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5e-06,
+      "loss": 0.7082,
+      "step": 640
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5e-06,
+      "loss": 0.6548,
+      "step": 641
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5e-06,
+      "loss": 0.6692,
+      "step": 642
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5e-06,
+      "loss": 0.6954,
+      "step": 643
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5e-06,
+      "loss": 0.7016,
+      "step": 644
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5e-06,
+      "loss": 0.6396,
+      "step": 645
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5e-06,
+      "loss": 0.7051,
+      "step": 646
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5e-06,
+      "loss": 0.6304,
+      "step": 647
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5e-06,
+      "loss": 0.712,
+      "step": 648
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5e-06,
+      "loss": 0.6546,
+      "step": 649
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5e-06,
+      "loss": 0.623,
+      "step": 650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5e-06,
+      "loss": 0.6999,
+      "step": 651
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5e-06,
+      "loss": 0.6456,
+      "step": 652
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5e-06,
+      "loss": 0.7269,
+      "step": 653
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5e-06,
+      "loss": 0.6431,
+      "step": 654
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5e-06,
+      "loss": 0.6238,
+      "step": 655
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5e-06,
+      "loss": 0.6795,
+      "step": 656
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5e-06,
+      "loss": 0.717,
+      "step": 657
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5e-06,
+      "loss": 0.672,
+      "step": 658
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5e-06,
+      "loss": 0.7249,
+      "step": 659
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5e-06,
+      "loss": 0.682,
+      "step": 660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5e-06,
+      "loss": 0.6847,
+      "step": 661
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5e-06,
+      "loss": 0.7015,
+      "step": 662
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5e-06,
+      "loss": 0.6356,
+      "step": 663
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5e-06,
+      "loss": 0.68,
+      "step": 664
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5e-06,
+      "loss": 0.6923,
+      "step": 665
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5e-06,
+      "loss": 0.7048,
+      "step": 666
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5e-06,
+      "loss": 0.6439,
+      "step": 667
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5e-06,
+      "loss": 0.7025,
+      "step": 668
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5e-06,
+      "loss": 0.705,
+      "step": 669
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5e-06,
+      "loss": 0.6406,
+      "step": 670
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5e-06,
+      "loss": 0.6755,
+      "step": 671
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5e-06,
+      "loss": 0.7014,
+      "step": 672
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5e-06,
+      "loss": 0.6622,
+      "step": 673
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5e-06,
+      "loss": 0.6148,
+      "step": 674
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5e-06,
+      "loss": 0.5899,
+      "step": 675
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5e-06,
+      "loss": 0.6733,
+      "step": 676
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5e-06,
+      "loss": 0.7083,
+      "step": 677
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5e-06,
+      "loss": 0.6621,
+      "step": 678
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5e-06,
+      "loss": 0.6074,
+      "step": 679
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5e-06,
+      "loss": 0.6343,
+      "step": 680
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5e-06,
+      "loss": 0.6449,
+      "step": 681
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5e-06,
+      "loss": 0.6625,
+      "step": 682
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5e-06,
+      "loss": 0.5941,
+      "step": 683
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5e-06,
+      "loss": 0.7105,
+      "step": 684
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5e-06,
+      "loss": 0.6419,
+      "step": 685
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5e-06,
+      "loss": 0.7047,
+      "step": 686
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5e-06,
+      "loss": 0.6915,
+      "step": 687
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5e-06,
+      "loss": 0.6395,
+      "step": 688
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5e-06,
+      "loss": 0.6658,
+      "step": 689
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5e-06,
+      "loss": 0.65,
+      "step": 690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5e-06,
+      "loss": 0.6938,
+      "step": 691
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5e-06,
+      "loss": 0.6632,
+      "step": 692
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5e-06,
+      "loss": 0.6023,
+      "step": 693
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5e-06,
+      "loss": 0.7095,
+      "step": 694
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5e-06,
+      "loss": 0.6893,
+      "step": 695
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5e-06,
+      "loss": 0.6084,
+      "step": 696
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5e-06,
+      "loss": 0.677,
+      "step": 697
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5e-06,
+      "loss": 0.7342,
+      "step": 698
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5e-06,
+      "loss": 0.6424,
+      "step": 699
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5e-06,
+      "loss": 0.6884,
+      "step": 700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-06,
+      "loss": 0.682,
+      "step": 701
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-06,
+      "loss": 0.7351,
+      "step": 702
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-06,
+      "loss": 0.6483,
+      "step": 703
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-06,
+      "loss": 0.7119,
+      "step": 704
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5e-06,
+      "loss": 0.7062,
+      "step": 705
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5e-06,
+      "loss": 0.6801,
+      "step": 706
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5e-06,
+      "loss": 0.706,
+      "step": 707
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5e-06,
+      "loss": 0.7079,
+      "step": 708
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5e-06,
+      "loss": 0.6788,
+      "step": 709
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5e-06,
+      "loss": 0.6193,
+      "step": 710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5e-06,
+      "loss": 0.7094,
+      "step": 711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5e-06,
+      "loss": 0.6654,
+      "step": 712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5e-06,
+      "loss": 0.7434,
+      "step": 713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5e-06,
+      "loss": 0.6704,
+      "step": 714
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5e-06,
+      "loss": 0.6333,
+      "step": 715
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5e-06,
+      "loss": 0.6433,
+      "step": 716
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5e-06,
+      "loss": 0.6736,
+      "step": 717
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5e-06,
+      "loss": 0.6514,
+      "step": 718
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5e-06,
+      "loss": 0.682,
+      "step": 719
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5e-06,
+      "loss": 0.6433,
+      "step": 720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5e-06,
+      "loss": 0.617,
+      "step": 721
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5e-06,
+      "loss": 0.6506,
+      "step": 722
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5e-06,
+      "loss": 0.6418,
+      "step": 723
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5e-06,
+      "loss": 0.6041,
+      "step": 724
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5e-06,
+      "loss": 0.7001,
+      "step": 725
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5e-06,
+      "loss": 0.7328,
+      "step": 726
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5e-06,
+      "loss": 0.7258,
+      "step": 727
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5e-06,
+      "loss": 0.748,
+      "step": 728
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5e-06,
+      "loss": 0.7013,
+      "step": 729
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5e-06,
+      "loss": 0.6213,
+      "step": 730
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5e-06,
+      "loss": 0.6569,
+      "step": 731
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5e-06,
+      "loss": 0.6732,
+      "step": 732
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5e-06,
+      "loss": 0.6383,
+      "step": 733
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5e-06,
+      "loss": 0.6867,
+      "step": 734
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5e-06,
+      "loss": 0.6481,
+      "step": 735
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5e-06,
+      "loss": 0.6812,
+      "step": 736
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5e-06,
+      "loss": 0.6404,
+      "step": 737
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5e-06,
+      "loss": 0.7035,
+      "step": 738
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5e-06,
+      "loss": 0.6782,
+      "step": 739
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5e-06,
+      "loss": 0.7029,
+      "step": 740
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5e-06,
+      "loss": 0.6924,
+      "step": 741
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5e-06,
+      "loss": 0.6231,
+      "step": 742
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5e-06,
+      "loss": 0.6826,
+      "step": 743
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5e-06,
+      "loss": 0.6974,
+      "step": 744
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5e-06,
+      "loss": 0.6854,
+      "step": 745
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5e-06,
+      "loss": 0.6832,
+      "step": 746
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5e-06,
+      "loss": 0.6478,
+      "step": 747
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5e-06,
+      "loss": 0.6969,
+      "step": 748
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5e-06,
+      "loss": 0.5926,
+      "step": 749
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5e-06,
+      "loss": 0.6663,
+      "step": 750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5e-06,
+      "loss": 0.634,
+      "step": 751
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5e-06,
+      "loss": 0.6282,
+      "step": 752
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5e-06,
+      "loss": 0.6698,
+      "step": 753
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5e-06,
+      "loss": 0.6108,
+      "step": 754
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5e-06,
+      "loss": 0.7208,
+      "step": 755
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5e-06,
+      "loss": 0.6549,
+      "step": 756
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5e-06,
+      "loss": 0.6543,
+      "step": 757
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5e-06,
+      "loss": 0.7182,
+      "step": 758
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5e-06,
+      "loss": 0.599,
+      "step": 759
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5e-06,
+      "loss": 0.6647,
+      "step": 760
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5e-06,
+      "loss": 0.6433,
+      "step": 761
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5e-06,
+      "loss": 0.6173,
+      "step": 762
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5e-06,
+      "loss": 0.712,
+      "step": 763
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5e-06,
+      "loss": 0.6192,
+      "step": 764
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5e-06,
+      "loss": 0.6745,
+      "step": 765
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5e-06,
+      "loss": 0.6414,
+      "step": 766
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-06,
+      "loss": 0.6044,
+      "step": 767
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-06,
+      "loss": 0.6593,
+      "step": 768
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-06,
+      "loss": 0.6206,
+      "step": 769
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-06,
+      "loss": 0.7393,
+      "step": 770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-06,
+      "loss": 0.6372,
+      "step": 771
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5e-06,
+      "loss": 0.6863,
+      "step": 772
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5e-06,
+      "loss": 0.6698,
+      "step": 773
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5e-06,
+      "loss": 0.7513,
+      "step": 774
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5e-06,
+      "loss": 0.674,
+      "step": 775
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5e-06,
+      "loss": 0.6808,
+      "step": 776
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5e-06,
+      "loss": 0.6522,
+      "step": 777
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5e-06,
+      "loss": 0.6485,
+      "step": 778
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5e-06,
+      "loss": 0.6837,
+      "step": 779
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5e-06,
+      "loss": 0.6814,
+      "step": 780
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-06,
+      "loss": 0.6531,
+      "step": 781
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-06,
+      "loss": 0.6139,
+      "step": 782
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-06,
+      "loss": 0.6811,
+      "step": 783
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-06,
+      "loss": 0.6896,
+      "step": 784
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-06,
+      "loss": 0.7136,
+      "step": 785
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5e-06,
+      "loss": 0.6082,
+      "step": 786
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5e-06,
+      "loss": 0.7313,
+      "step": 787
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5e-06,
+      "loss": 0.6398,
+      "step": 788
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5e-06,
+      "loss": 0.6908,
+      "step": 789
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5e-06,
+      "loss": 0.6298,
+      "step": 790
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5e-06,
+      "loss": 0.7552,
+      "step": 791
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5e-06,
+      "loss": 0.669,
+      "step": 792
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5e-06,
+      "loss": 0.6953,
+      "step": 793
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5e-06,
+      "loss": 0.7443,
+      "step": 794
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5e-06,
+      "loss": 0.6503,
+      "step": 795
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5e-06,
+      "loss": 0.6544,
+      "step": 796
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5e-06,
+      "loss": 0.6595,
+      "step": 797
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5e-06,
+      "loss": 0.754,
+      "step": 798
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5e-06,
+      "loss": 0.6459,
+      "step": 799
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5e-06,
+      "loss": 0.6171,
+      "step": 800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5e-06,
+      "loss": 0.665,
+      "step": 801
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5e-06,
+      "loss": 0.6774,
+      "step": 802
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5e-06,
+      "loss": 0.637,
+      "step": 803
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5e-06,
+      "loss": 0.6323,
+      "step": 804
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5e-06,
+      "loss": 0.627,
+      "step": 805
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5e-06,
+      "loss": 0.6702,
+      "step": 806
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5e-06,
+      "loss": 0.5686,
+      "step": 807
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5e-06,
+      "loss": 0.6745,
+      "step": 808
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5e-06,
+      "loss": 0.655,
+      "step": 809
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5e-06,
+      "loss": 0.7382,
+      "step": 810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5e-06,
+      "loss": 0.6136,
+      "step": 811
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5e-06,
+      "loss": 0.7075,
+      "step": 812
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5e-06,
+      "loss": 0.663,
+      "step": 813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5e-06,
+      "loss": 0.6568,
+      "step": 814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5e-06,
+      "loss": 0.7007,
+      "step": 815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5e-06,
+      "loss": 0.6555,
+      "step": 816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5e-06,
+      "loss": 0.6258,
+      "step": 817
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5e-06,
+      "loss": 0.6512,
+      "step": 818
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5e-06,
+      "loss": 0.6314,
+      "step": 819
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5e-06,
+      "loss": 0.7334,
+      "step": 820
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5e-06,
+      "loss": 0.6789,
+      "step": 821
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5e-06,
+      "loss": 0.6971,
+      "step": 822
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5e-06,
+      "loss": 0.6555,
+      "step": 823
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5e-06,
+      "loss": 0.5877,
+      "step": 824
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5e-06,
+      "loss": 0.6174,
+      "step": 825
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5e-06,
+      "loss": 0.5753,
+      "step": 826
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5e-06,
+      "loss": 0.6425,
+      "step": 827
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5e-06,
+      "loss": 0.5827,
+      "step": 828
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5e-06,
+      "loss": 0.7313,
+      "step": 829
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5e-06,
+      "loss": 0.6363,
+      "step": 830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5e-06,
+      "loss": 0.695,
+      "step": 831
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5e-06,
+      "loss": 0.6793,
+      "step": 832
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5e-06,
+      "loss": 0.6595,
+      "step": 833
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5e-06,
+      "loss": 0.689,
+      "step": 834
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5e-06,
+      "loss": 0.6273,
+      "step": 835
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5e-06,
+      "loss": 0.5842,
+      "step": 836
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5e-06,
+      "loss": 0.6253,
+      "step": 837
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5e-06,
+      "loss": 0.6028,
+      "step": 838
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5e-06,
+      "loss": 0.6788,
+      "step": 839
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5e-06,
+      "loss": 0.6564,
+      "step": 840
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5e-06,
+      "loss": 0.6375,
+      "step": 841
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5e-06,
+      "loss": 0.6534,
+      "step": 842
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5e-06,
+      "loss": 0.6574,
+      "step": 843
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5e-06,
+      "loss": 0.6827,
+      "step": 844
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5e-06,
+      "loss": 0.6365,
+      "step": 845
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5e-06,
+      "loss": 0.7116,
+      "step": 846
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5e-06,
+      "loss": 0.7119,
+      "step": 847
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5e-06,
+      "loss": 0.6449,
+      "step": 848
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5e-06,
+      "loss": 0.7404,
+      "step": 849
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5e-06,
+      "loss": 0.637,
+      "step": 850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5e-06,
+      "loss": 0.7162,
+      "step": 851
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5e-06,
+      "loss": 0.6568,
+      "step": 852
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5e-06,
+      "loss": 0.6537,
+      "step": 853
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5e-06,
+      "loss": 0.6479,
+      "step": 854
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5e-06,
+      "loss": 0.6823,
+      "step": 855
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5e-06,
+      "loss": 0.6025,
+      "step": 856
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5e-06,
+      "loss": 0.6211,
+      "step": 857
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5e-06,
+      "loss": 0.6495,
+      "step": 858
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5e-06,
+      "loss": 0.6787,
+      "step": 859
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5e-06,
+      "loss": 0.6624,
+      "step": 860
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5e-06,
+      "loss": 0.6696,
+      "step": 861
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5e-06,
+      "loss": 0.6729,
+      "step": 862
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5e-06,
+      "loss": 0.5713,
+      "step": 863
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5e-06,
+      "loss": 0.6875,
+      "step": 864
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5e-06,
+      "loss": 0.6179,
+      "step": 865
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5e-06,
+      "loss": 0.627,
+      "step": 866
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5e-06,
+      "loss": 0.6349,
+      "step": 867
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5e-06,
+      "loss": 0.6884,
+      "step": 868
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5e-06,
+      "loss": 0.6596,
+      "step": 869
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5e-06,
+      "loss": 0.5994,
+      "step": 870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5e-06,
+      "loss": 0.6682,
+      "step": 871
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5e-06,
+      "loss": 0.6339,
+      "step": 872
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5e-06,
+      "loss": 0.5892,
+      "step": 873
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5e-06,
+      "loss": 0.6953,
+      "step": 874
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5e-06,
+      "loss": 0.6633,
+      "step": 875
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5e-06,
+      "loss": 0.6882,
+      "step": 876
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5e-06,
+      "loss": 0.6541,
+      "step": 877
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5e-06,
+      "loss": 0.6291,
+      "step": 878
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5e-06,
+      "loss": 0.62,
+      "step": 879
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5e-06,
+      "loss": 0.7414,
+      "step": 880
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5e-06,
+      "loss": 0.6507,
+      "step": 881
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5e-06,
+      "loss": 0.6564,
+      "step": 882
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5e-06,
+      "loss": 0.6911,
+      "step": 883
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5e-06,
+      "loss": 0.6462,
+      "step": 884
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5e-06,
+      "loss": 0.6604,
+      "step": 885
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5e-06,
+      "loss": 0.7406,
+      "step": 886
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5e-06,
+      "loss": 0.7544,
+      "step": 887
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5e-06,
+      "loss": 0.6412,
+      "step": 888
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5e-06,
+      "loss": 0.6535,
+      "step": 889
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5e-06,
+      "loss": 0.6588,
+      "step": 890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5e-06,
+      "loss": 0.6733,
+      "step": 891
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5e-06,
+      "loss": 0.6719,
+      "step": 892
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5e-06,
+      "loss": 0.7041,
+      "step": 893
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5e-06,
+      "loss": 0.6974,
+      "step": 894
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5e-06,
+      "loss": 0.6368,
+      "step": 895
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5e-06,
+      "loss": 0.6292,
+      "step": 896
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5e-06,
+      "loss": 0.6835,
+      "step": 897
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5e-06,
+      "loss": 0.6837,
+      "step": 898
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5e-06,
+      "loss": 0.6805,
+      "step": 899
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5e-06,
+      "loss": 0.6634,
+      "step": 900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5e-06,
+      "loss": 0.6356,
+      "step": 901
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5e-06,
+      "loss": 0.5687,
+      "step": 902
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5e-06,
+      "loss": 0.6755,
+      "step": 903
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5e-06,
+      "loss": 0.6948,
+      "step": 904
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5e-06,
+      "loss": 0.6719,
+      "step": 905
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5e-06,
+      "loss": 0.623,
+      "step": 906
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5e-06,
+      "loss": 0.7182,
+      "step": 907
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5e-06,
+      "loss": 0.6367,
+      "step": 908
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5e-06,
+      "loss": 0.5907,
+      "step": 909
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5e-06,
+      "loss": 0.6743,
+      "step": 910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5e-06,
+      "loss": 0.6393,
+      "step": 911
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5e-06,
+      "loss": 0.6271,
+      "step": 912
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5e-06,
+      "loss": 0.596,
+      "step": 913
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5e-06,
+      "loss": 0.661,
+      "step": 914
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5e-06,
+      "loss": 0.6998,
+      "step": 915
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5e-06,
+      "loss": 0.6651,
+      "step": 916
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5e-06,
+      "loss": 0.6324,
+      "step": 917
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5e-06,
+      "loss": 0.6627,
+      "step": 918
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5e-06,
+      "loss": 0.6647,
+      "step": 919
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5e-06,
+      "loss": 0.6784,
+      "step": 920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5e-06,
+      "loss": 0.6775,
+      "step": 921
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5e-06,
+      "loss": 0.6388,
+      "step": 922
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5e-06,
+      "loss": 0.652,
+      "step": 923
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5e-06,
+      "loss": 0.6082,
+      "step": 924
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5e-06,
+      "loss": 0.6633,
+      "step": 925
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5e-06,
+      "loss": 0.6191,
+      "step": 926
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5e-06,
+      "loss": 0.6025,
+      "step": 927
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5e-06,
+      "loss": 0.6318,
+      "step": 928
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5e-06,
+      "loss": 0.5633,
+      "step": 929
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5e-06,
+      "loss": 0.6696,
+      "step": 930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5e-06,
+      "loss": 0.6188,
+      "step": 931
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5e-06,
+      "loss": 0.6501,
+      "step": 932
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5e-06,
+      "loss": 0.6752,
+      "step": 933
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5e-06,
+      "loss": 0.6481,
+      "step": 934
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5e-06,
+      "loss": 0.5967,
+      "step": 935
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6819,
+      "step": 936
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6676,
+      "step": 937
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5e-06,
+      "loss": 0.628,
+      "step": 938
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5e-06,
+      "loss": 0.5186,
+      "step": 939
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5e-06,
+      "loss": 0.4675,
+      "step": 940
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5e-06,
+      "loss": 0.4772,
+      "step": 941
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5e-06,
+      "loss": 0.4556,
+      "step": 942
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5e-06,
+      "loss": 0.5055,
+      "step": 943
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5e-06,
+      "loss": 0.4549,
+      "step": 944
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5e-06,
+      "loss": 0.5028,
+      "step": 945
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5e-06,
+      "loss": 0.4828,
+      "step": 946
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5e-06,
+      "loss": 0.4572,
+      "step": 947
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5e-06,
+      "loss": 0.4994,
+      "step": 948
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5e-06,
+      "loss": 0.4992,
+      "step": 949
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5e-06,
+      "loss": 0.4543,
+      "step": 950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5e-06,
+      "loss": 0.4747,
+      "step": 951
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5e-06,
+      "loss": 0.4959,
+      "step": 952
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5e-06,
+      "loss": 0.4663,
+      "step": 953
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5e-06,
+      "loss": 0.5015,
+      "step": 954
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5e-06,
+      "loss": 0.4362,
+      "step": 955
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5e-06,
+      "loss": 0.4341,
+      "step": 956
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5e-06,
+      "loss": 0.4915,
+      "step": 957
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5e-06,
+      "loss": 0.4976,
+      "step": 958
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5e-06,
+      "loss": 0.409,
+      "step": 959
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5e-06,
+      "loss": 0.4256,
+      "step": 960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5e-06,
+      "loss": 0.4777,
+      "step": 961
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5e-06,
+      "loss": 0.5021,
+      "step": 962
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5e-06,
+      "loss": 0.504,
+      "step": 963
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5e-06,
+      "loss": 0.428,
+      "step": 964
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5e-06,
+      "loss": 0.4087,
+      "step": 965
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5e-06,
+      "loss": 0.4566,
+      "step": 966
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5e-06,
+      "loss": 0.4736,
+      "step": 967
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5e-06,
+      "loss": 0.457,
+      "step": 968
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5e-06,
+      "loss": 0.4424,
+      "step": 969
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5e-06,
+      "loss": 0.419,
+      "step": 970
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5e-06,
+      "loss": 0.4657,
+      "step": 971
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5e-06,
+      "loss": 0.4365,
+      "step": 972
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5e-06,
+      "loss": 0.4528,
+      "step": 973
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5e-06,
+      "loss": 0.5007,
+      "step": 974
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5e-06,
+      "loss": 0.4573,
+      "step": 975
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5e-06,
+      "loss": 0.4779,
+      "step": 976
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5e-06,
+      "loss": 0.5054,
+      "step": 977
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5e-06,
+      "loss": 0.4605,
+      "step": 978
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5e-06,
+      "loss": 0.4708,
+      "step": 979
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5e-06,
+      "loss": 0.4036,
+      "step": 980
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5e-06,
+      "loss": 0.4118,
+      "step": 981
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5e-06,
+      "loss": 0.4611,
+      "step": 982
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5e-06,
+      "loss": 0.4326,
+      "step": 983
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5e-06,
+      "loss": 0.5211,
+      "step": 984
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5e-06,
+      "loss": 0.4591,
+      "step": 985
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5e-06,
+      "loss": 0.4512,
+      "step": 986
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5e-06,
+      "loss": 0.4619,
+      "step": 987
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5e-06,
+      "loss": 0.4822,
+      "step": 988
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5e-06,
+      "loss": 0.4614,
+      "step": 989
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5e-06,
+      "loss": 0.4838,
+      "step": 990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5e-06,
+      "loss": 0.4364,
+      "step": 991
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5e-06,
+      "loss": 0.4469,
+      "step": 992
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5e-06,
+      "loss": 0.4861,
+      "step": 993
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5e-06,
+      "loss": 0.5073,
+      "step": 994
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5e-06,
+      "loss": 0.4891,
+      "step": 995
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5e-06,
+      "loss": 0.4373,
+      "step": 996
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5e-06,
+      "loss": 0.48,
+      "step": 997
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5e-06,
+      "loss": 0.4696,
+      "step": 998
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5e-06,
+      "loss": 0.4889,
+      "step": 999
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5e-06,
+      "loss": 0.4653,
+      "step": 1000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5e-06,
+      "loss": 0.4551,
+      "step": 1001
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5e-06,
+      "loss": 0.4436,
+      "step": 1002
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5e-06,
+      "loss": 0.4544,
+      "step": 1003
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5e-06,
+      "loss": 0.4508,
+      "step": 1004
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5e-06,
+      "loss": 0.4137,
+      "step": 1005
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5e-06,
+      "loss": 0.472,
+      "step": 1006
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5e-06,
+      "loss": 0.4519,
+      "step": 1007
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5e-06,
+      "loss": 0.4376,
+      "step": 1008
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5e-06,
+      "loss": 0.4642,
+      "step": 1009
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5e-06,
+      "loss": 0.4591,
+      "step": 1010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5e-06,
+      "loss": 0.475,
+      "step": 1011
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5e-06,
+      "loss": 0.4433,
+      "step": 1012
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5e-06,
+      "loss": 0.4952,
+      "step": 1013
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5e-06,
+      "loss": 0.4165,
+      "step": 1014
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5e-06,
+      "loss": 0.4649,
+      "step": 1015
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5e-06,
+      "loss": 0.4572,
+      "step": 1016
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5e-06,
+      "loss": 0.4608,
+      "step": 1017
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5e-06,
+      "loss": 0.4751,
+      "step": 1018
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5e-06,
+      "loss": 0.444,
+      "step": 1019
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5e-06,
+      "loss": 0.4434,
+      "step": 1020
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5e-06,
+      "loss": 0.4713,
+      "step": 1021
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5e-06,
+      "loss": 0.4788,
+      "step": 1022
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5e-06,
+      "loss": 0.4312,
+      "step": 1023
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5e-06,
+      "loss": 0.4247,
+      "step": 1024
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5e-06,
+      "loss": 0.4696,
+      "step": 1025
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5e-06,
+      "loss": 0.4558,
+      "step": 1026
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5e-06,
+      "loss": 0.4522,
+      "step": 1027
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5e-06,
+      "loss": 0.4271,
+      "step": 1028
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5e-06,
+      "loss": 0.4902,
+      "step": 1029
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5e-06,
+      "loss": 0.4871,
+      "step": 1030
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5e-06,
+      "loss": 0.4563,
+      "step": 1031
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5e-06,
+      "loss": 0.4299,
+      "step": 1032
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5e-06,
+      "loss": 0.4129,
+      "step": 1033
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5e-06,
+      "loss": 0.4955,
+      "step": 1034
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5e-06,
+      "loss": 0.5171,
+      "step": 1035
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5e-06,
+      "loss": 0.4318,
+      "step": 1036
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5e-06,
+      "loss": 0.476,
+      "step": 1037
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5e-06,
+      "loss": 0.4585,
+      "step": 1038
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5e-06,
+      "loss": 0.4495,
+      "step": 1039
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5e-06,
+      "loss": 0.533,
+      "step": 1040
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5e-06,
+      "loss": 0.4953,
+      "step": 1041
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5e-06,
+      "loss": 0.4988,
+      "step": 1042
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5e-06,
+      "loss": 0.4808,
+      "step": 1043
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5e-06,
+      "loss": 0.4716,
+      "step": 1044
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5e-06,
+      "loss": 0.4835,
+      "step": 1045
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5e-06,
+      "loss": 0.4627,
+      "step": 1046
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5e-06,
+      "loss": 0.4187,
+      "step": 1047
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5e-06,
+      "loss": 0.4042,
+      "step": 1048
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5e-06,
+      "loss": 0.4663,
+      "step": 1049
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5e-06,
+      "loss": 0.4584,
+      "step": 1050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5e-06,
+      "loss": 0.4843,
+      "step": 1051
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5e-06,
+      "loss": 0.4578,
+      "step": 1052
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5e-06,
+      "loss": 0.4873,
+      "step": 1053
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5e-06,
+      "loss": 0.5044,
+      "step": 1054
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5e-06,
+      "loss": 0.4847,
+      "step": 1055
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5e-06,
+      "loss": 0.4691,
+      "step": 1056
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5e-06,
+      "loss": 0.419,
+      "step": 1057
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5e-06,
+      "loss": 0.4458,
+      "step": 1058
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5e-06,
+      "loss": 0.4864,
+      "step": 1059
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5e-06,
+      "loss": 0.4579,
+      "step": 1060
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5e-06,
+      "loss": 0.4138,
+      "step": 1061
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5e-06,
+      "loss": 0.4812,
+      "step": 1062
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5e-06,
+      "loss": 0.4751,
+      "step": 1063
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5e-06,
+      "loss": 0.4436,
+      "step": 1064
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5e-06,
+      "loss": 0.5126,
+      "step": 1065
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5e-06,
+      "loss": 0.4814,
+      "step": 1066
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5e-06,
+      "loss": 0.4779,
+      "step": 1067
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5e-06,
+      "loss": 0.4711,
+      "step": 1068
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5e-06,
+      "loss": 0.4196,
+      "step": 1069
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5e-06,
+      "loss": 0.4634,
+      "step": 1070
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5e-06,
+      "loss": 0.4641,
+      "step": 1071
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5e-06,
+      "loss": 0.4076,
+      "step": 1072
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5e-06,
+      "loss": 0.4378,
+      "step": 1073
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5e-06,
+      "loss": 0.4605,
+      "step": 1074
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5e-06,
+      "loss": 0.4636,
+      "step": 1075
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5e-06,
+      "loss": 0.4529,
+      "step": 1076
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5e-06,
+      "loss": 0.453,
+      "step": 1077
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5e-06,
+      "loss": 0.5067,
+      "step": 1078
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5e-06,
+      "loss": 0.4677,
+      "step": 1079
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5e-06,
+      "loss": 0.4288,
+      "step": 1080
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5e-06,
+      "loss": 0.4404,
+      "step": 1081
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5e-06,
+      "loss": 0.539,
+      "step": 1082
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5e-06,
+      "loss": 0.4423,
+      "step": 1083
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5e-06,
+      "loss": 0.4589,
+      "step": 1084
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5e-06,
+      "loss": 0.4591,
+      "step": 1085
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5e-06,
+      "loss": 0.4649,
+      "step": 1086
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5e-06,
+      "loss": 0.5232,
+      "step": 1087
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5e-06,
+      "loss": 0.4626,
+      "step": 1088
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5e-06,
+      "loss": 0.4823,
+      "step": 1089
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5e-06,
+      "loss": 0.4942,
+      "step": 1090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5e-06,
+      "loss": 0.4546,
+      "step": 1091
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5e-06,
+      "loss": 0.497,
+      "step": 1092
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5e-06,
+      "loss": 0.4692,
+      "step": 1093
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5e-06,
+      "loss": 0.4565,
+      "step": 1094
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5e-06,
+      "loss": 0.4338,
+      "step": 1095
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5e-06,
+      "loss": 0.5049,
+      "step": 1096
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5e-06,
+      "loss": 0.4168,
+      "step": 1097
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5e-06,
+      "loss": 0.5077,
+      "step": 1098
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5e-06,
+      "loss": 0.4265,
+      "step": 1099
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5e-06,
+      "loss": 0.4508,
+      "step": 1100
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5e-06,
+      "loss": 0.4571,
+      "step": 1101
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5e-06,
+      "loss": 0.4577,
+      "step": 1102
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5e-06,
+      "loss": 0.4374,
+      "step": 1103
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5e-06,
+      "loss": 0.4955,
+      "step": 1104
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5e-06,
+      "loss": 0.4792,
+      "step": 1105
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5e-06,
+      "loss": 0.4664,
+      "step": 1106
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5e-06,
+      "loss": 0.49,
+      "step": 1107
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5e-06,
+      "loss": 0.4195,
+      "step": 1108
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5e-06,
+      "loss": 0.4954,
+      "step": 1109
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5e-06,
+      "loss": 0.4671,
+      "step": 1110
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5e-06,
+      "loss": 0.4928,
+      "step": 1111
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5e-06,
+      "loss": 0.4172,
+      "step": 1112
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5e-06,
+      "loss": 0.437,
+      "step": 1113
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5e-06,
+      "loss": 0.4683,
+      "step": 1114
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5e-06,
+      "loss": 0.4365,
+      "step": 1115
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5e-06,
+      "loss": 0.4032,
+      "step": 1116
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5e-06,
+      "loss": 0.4653,
+      "step": 1117
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5e-06,
+      "loss": 0.5175,
+      "step": 1118
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5e-06,
+      "loss": 0.52,
+      "step": 1119
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5e-06,
+      "loss": 0.4449,
+      "step": 1120
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5e-06,
+      "loss": 0.4385,
+      "step": 1121
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5e-06,
+      "loss": 0.4825,
+      "step": 1122
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5e-06,
+      "loss": 0.5465,
+      "step": 1123
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5e-06,
+      "loss": 0.4844,
+      "step": 1124
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5e-06,
+      "loss": 0.4391,
+      "step": 1125
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5e-06,
+      "loss": 0.429,
+      "step": 1126
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5e-06,
+      "loss": 0.4597,
+      "step": 1127
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5e-06,
+      "loss": 0.4498,
+      "step": 1128
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5e-06,
+      "loss": 0.498,
+      "step": 1129
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5e-06,
+      "loss": 0.4765,
+      "step": 1130
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5e-06,
+      "loss": 0.4508,
+      "step": 1131
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5e-06,
+      "loss": 0.4713,
+      "step": 1132
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5e-06,
+      "loss": 0.4575,
+      "step": 1133
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5e-06,
+      "loss": 0.439,
+      "step": 1134
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5e-06,
+      "loss": 0.5138,
+      "step": 1135
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5e-06,
+      "loss": 0.4405,
+      "step": 1136
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5e-06,
+      "loss": 0.4421,
+      "step": 1137
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5e-06,
+      "loss": 0.5195,
+      "step": 1138
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5e-06,
+      "loss": 0.4836,
+      "step": 1139
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5e-06,
+      "loss": 0.5066,
+      "step": 1140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5e-06,
+      "loss": 0.4882,
+      "step": 1141
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5e-06,
+      "loss": 0.4413,
+      "step": 1142
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5e-06,
+      "loss": 0.4911,
+      "step": 1143
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5e-06,
+      "loss": 0.4449,
+      "step": 1144
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5e-06,
+      "loss": 0.475,
+      "step": 1145
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5e-06,
+      "loss": 0.5143,
+      "step": 1146
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5e-06,
+      "loss": 0.4631,
+      "step": 1147
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5e-06,
+      "loss": 0.5141,
+      "step": 1148
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5e-06,
+      "loss": 0.5305,
+      "step": 1149
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5e-06,
+      "loss": 0.4522,
+      "step": 1150
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5e-06,
+      "loss": 0.4989,
+      "step": 1151
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5e-06,
+      "loss": 0.4853,
+      "step": 1152
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5e-06,
+      "loss": 0.5036,
+      "step": 1153
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5e-06,
+      "loss": 0.4898,
+      "step": 1154
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5e-06,
+      "loss": 0.4514,
+      "step": 1155
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5e-06,
+      "loss": 0.4421,
+      "step": 1156
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5e-06,
+      "loss": 0.5016,
+      "step": 1157
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5e-06,
+      "loss": 0.4635,
+      "step": 1158
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5e-06,
+      "loss": 0.4643,
+      "step": 1159
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5e-06,
+      "loss": 0.4418,
+      "step": 1160
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5e-06,
+      "loss": 0.4699,
+      "step": 1161
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5e-06,
+      "loss": 0.4369,
+      "step": 1162
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5e-06,
+      "loss": 0.4658,
+      "step": 1163
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5e-06,
+      "loss": 0.5148,
+      "step": 1164
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5e-06,
+      "loss": 0.4171,
+      "step": 1165
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5e-06,
+      "loss": 0.4613,
+      "step": 1166
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5e-06,
+      "loss": 0.4579,
+      "step": 1167
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5e-06,
+      "loss": 0.4645,
+      "step": 1168
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5e-06,
+      "loss": 0.4434,
+      "step": 1169
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5e-06,
+      "loss": 0.4594,
+      "step": 1170
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5e-06,
+      "loss": 0.481,
+      "step": 1171
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5e-06,
+      "loss": 0.4971,
+      "step": 1172
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5e-06,
+      "loss": 0.5189,
+      "step": 1173
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5e-06,
+      "loss": 0.4535,
+      "step": 1174
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5e-06,
+      "loss": 0.4907,
+      "step": 1175
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5e-06,
+      "loss": 0.4596,
+      "step": 1176
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5e-06,
+      "loss": 0.4928,
+      "step": 1177
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5e-06,
+      "loss": 0.4594,
+      "step": 1178
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5e-06,
+      "loss": 0.4182,
+      "step": 1179
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5e-06,
+      "loss": 0.488,
+      "step": 1180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5e-06,
+      "loss": 0.431,
+      "step": 1181
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5e-06,
+      "loss": 0.5113,
+      "step": 1182
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5e-06,
+      "loss": 0.4867,
+      "step": 1183
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5e-06,
+      "loss": 0.4409,
+      "step": 1184
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5e-06,
+      "loss": 0.4756,
+      "step": 1185
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5e-06,
+      "loss": 0.4216,
+      "step": 1186
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5e-06,
+      "loss": 0.5288,
+      "step": 1187
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5e-06,
+      "loss": 0.5078,
+      "step": 1188
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5e-06,
+      "loss": 0.4657,
+      "step": 1189
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5e-06,
+      "loss": 0.3979,
+      "step": 1190
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5e-06,
+      "loss": 0.4614,
+      "step": 1191
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5e-06,
+      "loss": 0.4566,
+      "step": 1192
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5e-06,
+      "loss": 0.4549,
+      "step": 1193
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5e-06,
+      "loss": 0.4708,
+      "step": 1194
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5e-06,
+      "loss": 0.531,
+      "step": 1195
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5e-06,
+      "loss": 0.4483,
+      "step": 1196
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5e-06,
+      "loss": 0.4784,
+      "step": 1197
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5e-06,
+      "loss": 0.4981,
+      "step": 1198
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5e-06,
+      "loss": 0.4577,
+      "step": 1199
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5e-06,
+      "loss": 0.4991,
+      "step": 1200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5e-06,
+      "loss": 0.4577,
+      "step": 1201
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5e-06,
+      "loss": 0.4605,
+      "step": 1202
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5e-06,
+      "loss": 0.4788,
+      "step": 1203
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5e-06,
+      "loss": 0.4865,
+      "step": 1204
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5e-06,
+      "loss": 0.4526,
+      "step": 1205
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5e-06,
+      "loss": 0.4846,
+      "step": 1206
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5e-06,
+      "loss": 0.4418,
+      "step": 1207
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5e-06,
+      "loss": 0.4387,
+      "step": 1208
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5e-06,
+      "loss": 0.48,
+      "step": 1209
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5e-06,
+      "loss": 0.4957,
+      "step": 1210
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5e-06,
+      "loss": 0.4683,
+      "step": 1211
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5e-06,
+      "loss": 0.4622,
+      "step": 1212
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5e-06,
+      "loss": 0.5335,
+      "step": 1213
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5e-06,
+      "loss": 0.462,
+      "step": 1214
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5e-06,
+      "loss": 0.4833,
+      "step": 1215
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5e-06,
+      "loss": 0.5121,
+      "step": 1216
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5e-06,
+      "loss": 0.4854,
+      "step": 1217
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5e-06,
+      "loss": 0.4911,
+      "step": 1218
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5e-06,
+      "loss": 0.5076,
+      "step": 1219
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5e-06,
+      "loss": 0.5053,
+      "step": 1220
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5e-06,
+      "loss": 0.4672,
+      "step": 1221
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5e-06,
+      "loss": 0.4487,
+      "step": 1222
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5e-06,
+      "loss": 0.5166,
+      "step": 1223
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5e-06,
+      "loss": 0.4578,
+      "step": 1224
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5e-06,
+      "loss": 0.4626,
+      "step": 1225
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5e-06,
+      "loss": 0.5091,
+      "step": 1226
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5e-06,
+      "loss": 0.5026,
+      "step": 1227
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5e-06,
+      "loss": 0.4546,
+      "step": 1228
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5e-06,
+      "loss": 0.4343,
+      "step": 1229
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5e-06,
+      "loss": 0.476,
+      "step": 1230
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5e-06,
+      "loss": 0.4368,
+      "step": 1231
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5e-06,
+      "loss": 0.5612,
+      "step": 1232
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5e-06,
+      "loss": 0.5228,
+      "step": 1233
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5e-06,
+      "loss": 0.4407,
+      "step": 1234
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5e-06,
+      "loss": 0.4783,
+      "step": 1235
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5e-06,
+      "loss": 0.464,
+      "step": 1236
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5e-06,
+      "loss": 0.5213,
+      "step": 1237
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5e-06,
+      "loss": 0.4831,
+      "step": 1238
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5e-06,
+      "loss": 0.5182,
+      "step": 1239
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5e-06,
+      "loss": 0.5142,
+      "step": 1240
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5e-06,
+      "loss": 0.4527,
+      "step": 1241
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5e-06,
+      "loss": 0.4997,
+      "step": 1242
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5e-06,
+      "loss": 0.4493,
+      "step": 1243
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5e-06,
+      "loss": 0.4374,
+      "step": 1244
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5e-06,
+      "loss": 0.4542,
+      "step": 1245
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5e-06,
+      "loss": 0.4864,
+      "step": 1246
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5e-06,
+      "loss": 0.5218,
+      "step": 1247
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5e-06,
+      "loss": 0.4939,
+      "step": 1248
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5e-06,
+      "loss": 0.4372,
+      "step": 1249
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5e-06,
+      "loss": 0.5079,
+      "step": 1250
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5e-06,
+      "loss": 0.4827,
+      "step": 1251
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5e-06,
+      "loss": 0.4771,
+      "step": 1252
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5e-06,
+      "loss": 0.4563,
+      "step": 1253
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5e-06,
+      "loss": 0.471,
+      "step": 1254
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5e-06,
+      "loss": 0.5081,
+      "step": 1255
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5e-06,
+      "loss": 0.4403,
+      "step": 1256
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5e-06,
+      "loss": 0.4816,
+      "step": 1257
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5e-06,
+      "loss": 0.4771,
+      "step": 1258
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5e-06,
+      "loss": 0.4815,
+      "step": 1259
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5e-06,
+      "loss": 0.4467,
+      "step": 1260
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5e-06,
+      "loss": 0.4871,
+      "step": 1261
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5e-06,
+      "loss": 0.4699,
+      "step": 1262
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5e-06,
+      "loss": 0.525,
+      "step": 1263
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5e-06,
+      "loss": 0.493,
+      "step": 1264
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5e-06,
+      "loss": 0.5069,
+      "step": 1265
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5e-06,
+      "loss": 0.4603,
+      "step": 1266
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5e-06,
+      "loss": 0.4585,
+      "step": 1267
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5e-06,
+      "loss": 0.4865,
+      "step": 1268
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5e-06,
+      "loss": 0.469,
+      "step": 1269
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5e-06,
+      "loss": 0.446,
+      "step": 1270
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5e-06,
+      "loss": 0.5073,
+      "step": 1271
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5e-06,
+      "loss": 0.4289,
+      "step": 1272
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5e-06,
+      "loss": 0.525,
+      "step": 1273
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5e-06,
+      "loss": 0.4531,
+      "step": 1274
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5e-06,
+      "loss": 0.4689,
+      "step": 1275
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5e-06,
+      "loss": 0.4295,
+      "step": 1276
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5e-06,
+      "loss": 0.4737,
+      "step": 1277
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5e-06,
+      "loss": 0.5005,
+      "step": 1278
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5e-06,
+      "loss": 0.4525,
+      "step": 1279
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5e-06,
+      "loss": 0.4664,
+      "step": 1280
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5e-06,
+      "loss": 0.4184,
+      "step": 1281
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5e-06,
+      "loss": 0.4857,
+      "step": 1282
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5e-06,
+      "loss": 0.4312,
+      "step": 1283
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5e-06,
+      "loss": 0.4746,
+      "step": 1284
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5e-06,
+      "loss": 0.4687,
+      "step": 1285
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5e-06,
+      "loss": 0.4783,
+      "step": 1286
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5e-06,
+      "loss": 0.5589,
+      "step": 1287
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5e-06,
+      "loss": 0.494,
+      "step": 1288
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5e-06,
+      "loss": 0.433,
+      "step": 1289
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5e-06,
+      "loss": 0.4839,
+      "step": 1290
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5e-06,
+      "loss": 0.4658,
+      "step": 1291
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5e-06,
+      "loss": 0.445,
+      "step": 1292
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5e-06,
+      "loss": 0.4452,
+      "step": 1293
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5e-06,
+      "loss": 0.469,
+      "step": 1294
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5e-06,
+      "loss": 0.5178,
+      "step": 1295
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5e-06,
+      "loss": 0.4713,
+      "step": 1296
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5e-06,
+      "loss": 0.4483,
+      "step": 1297
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5e-06,
+      "loss": 0.5444,
+      "step": 1298
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5e-06,
+      "loss": 0.5405,
+      "step": 1299
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5e-06,
+      "loss": 0.4863,
+      "step": 1300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5e-06,
+      "loss": 0.4032,
+      "step": 1301
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5e-06,
+      "loss": 0.4608,
+      "step": 1302
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5e-06,
+      "loss": 0.5379,
+      "step": 1303
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5e-06,
+      "loss": 0.4557,
+      "step": 1304
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5e-06,
+      "loss": 0.5074,
+      "step": 1305
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5e-06,
+      "loss": 0.5336,
+      "step": 1306
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5e-06,
+      "loss": 0.4824,
+      "step": 1307
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5e-06,
+      "loss": 0.4458,
+      "step": 1308
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5e-06,
+      "loss": 0.5125,
+      "step": 1309
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5e-06,
+      "loss": 0.5257,
+      "step": 1310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5e-06,
+      "loss": 0.4806,
+      "step": 1311
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5e-06,
+      "loss": 0.4719,
+      "step": 1312
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5e-06,
+      "loss": 0.4676,
+      "step": 1313
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5e-06,
+      "loss": 0.4335,
+      "step": 1314
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5e-06,
+      "loss": 0.4807,
+      "step": 1315
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5e-06,
+      "loss": 0.4658,
+      "step": 1316
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5e-06,
+      "loss": 0.5139,
+      "step": 1317
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5e-06,
+      "loss": 0.4753,
+      "step": 1318
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5e-06,
+      "loss": 0.4801,
+      "step": 1319
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5e-06,
+      "loss": 0.5199,
+      "step": 1320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5e-06,
+      "loss": 0.5093,
+      "step": 1321
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5e-06,
+      "loss": 0.5164,
+      "step": 1322
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5e-06,
+      "loss": 0.471,
+      "step": 1323
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5e-06,
+      "loss": 0.4557,
+      "step": 1324
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5e-06,
+      "loss": 0.5154,
+      "step": 1325
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5e-06,
+      "loss": 0.4884,
+      "step": 1326
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5e-06,
+      "loss": 0.5154,
+      "step": 1327
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5e-06,
+      "loss": 0.4603,
+      "step": 1328
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5e-06,
+      "loss": 0.4563,
+      "step": 1329
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5e-06,
+      "loss": 0.4688,
+      "step": 1330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5e-06,
+      "loss": 0.431,
+      "step": 1331
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5e-06,
+      "loss": 0.5006,
+      "step": 1332
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5e-06,
+      "loss": 0.4178,
+      "step": 1333
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5e-06,
+      "loss": 0.4873,
+      "step": 1334
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5e-06,
+      "loss": 0.5417,
+      "step": 1335
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5e-06,
+      "loss": 0.486,
+      "step": 1336
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5e-06,
+      "loss": 0.5242,
+      "step": 1337
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5e-06,
+      "loss": 0.4569,
+      "step": 1338
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5e-06,
+      "loss": 0.4998,
+      "step": 1339
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5e-06,
+      "loss": 0.4955,
+      "step": 1340
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5e-06,
+      "loss": 0.4405,
+      "step": 1341
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5e-06,
+      "loss": 0.4536,
+      "step": 1342
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5e-06,
+      "loss": 0.4911,
+      "step": 1343
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5e-06,
+      "loss": 0.452,
+      "step": 1344
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5e-06,
+      "loss": 0.4854,
+      "step": 1345
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5e-06,
+      "loss": 0.4809,
+      "step": 1346
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5e-06,
+      "loss": 0.4704,
+      "step": 1347
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5e-06,
+      "loss": 0.5076,
+      "step": 1348
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5e-06,
+      "loss": 0.4823,
+      "step": 1349
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5e-06,
+      "loss": 0.4734,
+      "step": 1350
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5e-06,
+      "loss": 0.5126,
+      "step": 1351
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5e-06,
+      "loss": 0.4851,
+      "step": 1352
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5e-06,
+      "loss": 0.5091,
+      "step": 1353
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5e-06,
+      "loss": 0.444,
+      "step": 1354
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5e-06,
+      "loss": 0.4529,
+      "step": 1355
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5e-06,
+      "loss": 0.4443,
+      "step": 1356
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5e-06,
+      "loss": 0.4718,
+      "step": 1357
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5e-06,
+      "loss": 0.4932,
+      "step": 1358
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5e-06,
+      "loss": 0.4883,
+      "step": 1359
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5e-06,
+      "loss": 0.4333,
+      "step": 1360
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5e-06,
+      "loss": 0.4873,
+      "step": 1361
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5e-06,
+      "loss": 0.5205,
+      "step": 1362
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5e-06,
+      "loss": 0.4692,
+      "step": 1363
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5e-06,
+      "loss": 0.4913,
+      "step": 1364
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5e-06,
+      "loss": 0.4703,
+      "step": 1365
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5e-06,
+      "loss": 0.468,
+      "step": 1366
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5e-06,
+      "loss": 0.4371,
+      "step": 1367
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5e-06,
+      "loss": 0.4939,
+      "step": 1368
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5e-06,
+      "loss": 0.5303,
+      "step": 1369
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5e-06,
+      "loss": 0.4693,
+      "step": 1370
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5e-06,
+      "loss": 0.4983,
+      "step": 1371
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5e-06,
+      "loss": 0.538,
+      "step": 1372
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5e-06,
+      "loss": 0.502,
+      "step": 1373
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5e-06,
+      "loss": 0.4485,
+      "step": 1374
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5e-06,
+      "loss": 0.492,
+      "step": 1375
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5e-06,
+      "loss": 0.5347,
+      "step": 1376
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5e-06,
+      "loss": 0.5042,
+      "step": 1377
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5e-06,
+      "loss": 0.465,
+      "step": 1378
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5e-06,
+      "loss": 0.5282,
+      "step": 1379
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5e-06,
+      "loss": 0.5263,
+      "step": 1380
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5e-06,
+      "loss": 0.4385,
+      "step": 1381
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5e-06,
+      "loss": 0.5507,
+      "step": 1382
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5e-06,
+      "loss": 0.4796,
+      "step": 1383
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5e-06,
+      "loss": 0.4903,
+      "step": 1384
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5e-06,
+      "loss": 0.4237,
+      "step": 1385
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5e-06,
+      "loss": 0.4514,
+      "step": 1386
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5e-06,
+      "loss": 0.5599,
+      "step": 1387
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5e-06,
+      "loss": 0.4761,
+      "step": 1388
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5e-06,
+      "loss": 0.432,
+      "step": 1389
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5e-06,
+      "loss": 0.4586,
+      "step": 1390
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5e-06,
+      "loss": 0.4985,
+      "step": 1391
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5e-06,
+      "loss": 0.4388,
+      "step": 1392
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5e-06,
+      "loss": 0.4854,
+      "step": 1393
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5e-06,
+      "loss": 0.5341,
+      "step": 1394
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5e-06,
+      "loss": 0.4839,
+      "step": 1395
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5e-06,
+      "loss": 0.5139,
+      "step": 1396
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5e-06,
+      "loss": 0.4496,
+      "step": 1397
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5e-06,
+      "loss": 0.4269,
+      "step": 1398
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5e-06,
+      "loss": 0.5157,
+      "step": 1399
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5e-06,
+      "loss": 0.453,
+      "step": 1400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5e-06,
+      "loss": 0.4525,
+      "step": 1401
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5e-06,
+      "loss": 0.5278,
+      "step": 1402
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5e-06,
+      "loss": 0.4676,
+      "step": 1403
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5e-06,
+      "loss": 0.4959,
+      "step": 1404
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5e-06,
+      "loss": 0.4697,
+      "step": 1405
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5e-06,
+      "loss": 0.4733,
+      "step": 1406
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5e-06,
+      "loss": 0.3208,
+      "step": 1407
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5e-06,
+      "loss": 0.2758,
+      "step": 1408
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5e-06,
+      "loss": 0.3046,
+      "step": 1409
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5e-06,
+      "loss": 0.3174,
+      "step": 1410
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5e-06,
+      "loss": 0.2731,
+      "step": 1411
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5e-06,
+      "loss": 0.2572,
+      "step": 1412
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5e-06,
+      "loss": 0.2664,
+      "step": 1413
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5e-06,
+      "loss": 0.2294,
+      "step": 1414
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5e-06,
+      "loss": 0.273,
+      "step": 1415
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5e-06,
+      "loss": 0.2588,
+      "step": 1416
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5e-06,
+      "loss": 0.2524,
+      "step": 1417
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5e-06,
+      "loss": 0.2362,
+      "step": 1418
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5e-06,
+      "loss": 0.2868,
+      "step": 1419
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5e-06,
+      "loss": 0.2445,
+      "step": 1420
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5e-06,
+      "loss": 0.2515,
+      "step": 1421
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5e-06,
+      "loss": 0.2632,
+      "step": 1422
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5e-06,
+      "loss": 0.2551,
+      "step": 1423
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5e-06,
+      "loss": 0.2611,
+      "step": 1424
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5e-06,
+      "loss": 0.2911,
+      "step": 1425
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5e-06,
+      "loss": 0.2599,
+      "step": 1426
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5e-06,
+      "loss": 0.319,
+      "step": 1427
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5e-06,
+      "loss": 0.2409,
+      "step": 1428
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5e-06,
+      "loss": 0.2501,
+      "step": 1429
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5e-06,
+      "loss": 0.2425,
+      "step": 1430
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5e-06,
+      "loss": 0.2582,
+      "step": 1431
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5e-06,
+      "loss": 0.2236,
+      "step": 1432
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5e-06,
+      "loss": 0.2153,
+      "step": 1433
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5e-06,
+      "loss": 0.2538,
+      "step": 1434
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5e-06,
+      "loss": 0.2448,
+      "step": 1435
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5e-06,
+      "loss": 0.2395,
+      "step": 1436
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5e-06,
+      "loss": 0.2379,
+      "step": 1437
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5e-06,
+      "loss": 0.2188,
+      "step": 1438
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5e-06,
+      "loss": 0.2425,
+      "step": 1439
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5e-06,
+      "loss": 0.2593,
+      "step": 1440
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5e-06,
+      "loss": 0.2396,
+      "step": 1441
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5e-06,
+      "loss": 0.2174,
+      "step": 1442
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5e-06,
+      "loss": 0.2453,
+      "step": 1443
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5e-06,
+      "loss": 0.2398,
+      "step": 1444
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5e-06,
+      "loss": 0.2031,
+      "step": 1445
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5e-06,
+      "loss": 0.2599,
+      "step": 1446
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5e-06,
+      "loss": 0.2187,
+      "step": 1447
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5e-06,
+      "loss": 0.2292,
+      "step": 1448
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5e-06,
+      "loss": 0.2584,
+      "step": 1449
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5e-06,
+      "loss": 0.2971,
+      "step": 1450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5e-06,
+      "loss": 0.2384,
+      "step": 1451
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5e-06,
+      "loss": 0.3018,
+      "step": 1452
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5e-06,
+      "loss": 0.2751,
+      "step": 1453
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5e-06,
+      "loss": 0.2459,
+      "step": 1454
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5e-06,
+      "loss": 0.2435,
+      "step": 1455
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5e-06,
+      "loss": 0.2679,
+      "step": 1456
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5e-06,
+      "loss": 0.2671,
+      "step": 1457
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5e-06,
+      "loss": 0.2208,
+      "step": 1458
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5e-06,
+      "loss": 0.2141,
+      "step": 1459
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5e-06,
+      "loss": 0.2352,
+      "step": 1460
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5e-06,
+      "loss": 0.2378,
+      "step": 1461
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5e-06,
+      "loss": 0.2373,
+      "step": 1462
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5e-06,
+      "loss": 0.264,
+      "step": 1463
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5e-06,
+      "loss": 0.2281,
+      "step": 1464
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5e-06,
+      "loss": 0.2242,
+      "step": 1465
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5e-06,
+      "loss": 0.2459,
+      "step": 1466
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5e-06,
+      "loss": 0.2543,
+      "step": 1467
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5e-06,
+      "loss": 0.23,
+      "step": 1468
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5e-06,
+      "loss": 0.2983,
+      "step": 1469
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5e-06,
+      "loss": 0.277,
+      "step": 1470
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5e-06,
+      "loss": 0.2712,
+      "step": 1471
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5e-06,
+      "loss": 0.211,
+      "step": 1472
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5e-06,
+      "loss": 0.2149,
+      "step": 1473
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5e-06,
+      "loss": 0.3059,
+      "step": 1474
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5e-06,
+      "loss": 0.2333,
+      "step": 1475
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5e-06,
+      "loss": 0.2445,
+      "step": 1476
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5e-06,
+      "loss": 0.2647,
+      "step": 1477
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5e-06,
+      "loss": 0.2314,
+      "step": 1478
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5e-06,
+      "loss": 0.2756,
+      "step": 1479
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5e-06,
+      "loss": 0.2219,
+      "step": 1480
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5e-06,
+      "loss": 0.2328,
+      "step": 1481
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5e-06,
+      "loss": 0.2757,
+      "step": 1482
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5e-06,
+      "loss": 0.2298,
+      "step": 1483
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5e-06,
+      "loss": 0.2873,
+      "step": 1484
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5e-06,
+      "loss": 0.2437,
+      "step": 1485
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5e-06,
+      "loss": 0.2439,
+      "step": 1486
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5e-06,
+      "loss": 0.2571,
+      "step": 1487
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5e-06,
+      "loss": 0.2446,
+      "step": 1488
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5e-06,
+      "loss": 0.2326,
+      "step": 1489
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5e-06,
+      "loss": 0.283,
+      "step": 1490
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5e-06,
+      "loss": 0.2443,
+      "step": 1491
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5e-06,
+      "loss": 0.2388,
+      "step": 1492
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5e-06,
+      "loss": 0.247,
+      "step": 1493
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5e-06,
+      "loss": 0.2597,
+      "step": 1494
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5e-06,
+      "loss": 0.2509,
+      "step": 1495
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5e-06,
+      "loss": 0.215,
+      "step": 1496
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5e-06,
+      "loss": 0.2704,
+      "step": 1497
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5e-06,
+      "loss": 0.2586,
+      "step": 1498
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5e-06,
+      "loss": 0.2431,
+      "step": 1499
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5e-06,
+      "loss": 0.2439,
+      "step": 1500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5e-06,
+      "loss": 0.2489,
+      "step": 1501
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5e-06,
+      "loss": 0.2418,
+      "step": 1502
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5e-06,
+      "loss": 0.2305,
+      "step": 1503
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5e-06,
+      "loss": 0.2895,
+      "step": 1504
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5e-06,
+      "loss": 0.212,
+      "step": 1505
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5e-06,
+      "loss": 0.2026,
+      "step": 1506
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5e-06,
+      "loss": 0.2013,
+      "step": 1507
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5e-06,
+      "loss": 0.2474,
+      "step": 1508
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5e-06,
+      "loss": 0.2082,
+      "step": 1509
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5e-06,
+      "loss": 0.2658,
+      "step": 1510
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5e-06,
+      "loss": 0.2466,
+      "step": 1511
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5e-06,
+      "loss": 0.3044,
+      "step": 1512
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5e-06,
+      "loss": 0.2928,
+      "step": 1513
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5e-06,
+      "loss": 0.2644,
+      "step": 1514
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5e-06,
+      "loss": 0.2145,
+      "step": 1515
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5e-06,
+      "loss": 0.2557,
+      "step": 1516
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5e-06,
+      "loss": 0.2243,
+      "step": 1517
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5e-06,
+      "loss": 0.2061,
+      "step": 1518
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5e-06,
+      "loss": 0.2778,
+      "step": 1519
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5e-06,
+      "loss": 0.2693,
+      "step": 1520
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5e-06,
+      "loss": 0.2054,
+      "step": 1521
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5e-06,
+      "loss": 0.2201,
+      "step": 1522
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5e-06,
+      "loss": 0.2481,
+      "step": 1523
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5e-06,
+      "loss": 0.244,
+      "step": 1524
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5e-06,
+      "loss": 0.2342,
+      "step": 1525
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5e-06,
+      "loss": 0.2688,
+      "step": 1526
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5e-06,
+      "loss": 0.2696,
+      "step": 1527
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5e-06,
+      "loss": 0.2539,
+      "step": 1528
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5e-06,
+      "loss": 0.3076,
+      "step": 1529
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5e-06,
+      "loss": 0.2626,
+      "step": 1530
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5e-06,
+      "loss": 0.2349,
+      "step": 1531
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5e-06,
+      "loss": 0.2892,
+      "step": 1532
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5e-06,
+      "loss": 0.2446,
+      "step": 1533
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5e-06,
+      "loss": 0.2377,
+      "step": 1534
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5e-06,
+      "loss": 0.2278,
+      "step": 1535
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5e-06,
+      "loss": 0.2471,
+      "step": 1536
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5e-06,
+      "loss": 0.2497,
+      "step": 1537
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5e-06,
+      "loss": 0.2361,
+      "step": 1538
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5e-06,
+      "loss": 0.2889,
+      "step": 1539
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5e-06,
+      "loss": 0.2589,
+      "step": 1540
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5e-06,
+      "loss": 0.2753,
+      "step": 1541
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5e-06,
+      "loss": 0.2334,
+      "step": 1542
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5e-06,
+      "loss": 0.3063,
+      "step": 1543
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5e-06,
+      "loss": 0.2809,
+      "step": 1544
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5e-06,
+      "loss": 0.2566,
+      "step": 1545
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5e-06,
+      "loss": 0.2449,
+      "step": 1546
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5e-06,
+      "loss": 0.2586,
+      "step": 1547
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5e-06,
+      "loss": 0.2463,
+      "step": 1548
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5e-06,
+      "loss": 0.2454,
+      "step": 1549
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5e-06,
+      "loss": 0.2534,
+      "step": 1550
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5e-06,
+      "loss": 0.2368,
+      "step": 1551
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5e-06,
+      "loss": 0.2168,
+      "step": 1552
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5e-06,
+      "loss": 0.2808,
+      "step": 1553
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "step": 1554
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5e-06,
+      "loss": 0.2759,
+      "step": 1555
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5e-06,
+      "loss": 0.2518,
+      "step": 1556
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5e-06,
+      "loss": 0.2335,
+      "step": 1557
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5e-06,
+      "loss": 0.2442,
+      "step": 1558
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5e-06,
+      "loss": 0.2034,
+      "step": 1559
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5e-06,
+      "loss": 0.2438,
+      "step": 1560
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5e-06,
+      "loss": 0.2359,
+      "step": 1561
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5e-06,
+      "loss": 0.2628,
+      "step": 1562
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5e-06,
+      "loss": 0.2577,
+      "step": 1563
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5e-06,
+      "loss": 0.2444,
+      "step": 1564
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5e-06,
+      "loss": 0.2449,
+      "step": 1565
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5e-06,
+      "loss": 0.228,
+      "step": 1566
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5e-06,
+      "loss": 0.2684,
+      "step": 1567
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5e-06,
+      "loss": 0.245,
+      "step": 1568
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5e-06,
+      "loss": 0.2394,
+      "step": 1569
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5e-06,
+      "loss": 0.2661,
+      "step": 1570
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5e-06,
+      "loss": 0.2791,
+      "step": 1571
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5e-06,
+      "loss": 0.2529,
+      "step": 1572
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5e-06,
+      "loss": 0.2618,
+      "step": 1573
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5e-06,
+      "loss": 0.2714,
+      "step": 1574
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5e-06,
+      "loss": 0.221,
+      "step": 1575
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5e-06,
+      "loss": 0.2582,
+      "step": 1576
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5e-06,
+      "loss": 0.2339,
+      "step": 1577
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5e-06,
+      "loss": 0.2526,
+      "step": 1578
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5e-06,
+      "loss": 0.2596,
+      "step": 1579
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5e-06,
+      "loss": 0.2507,
+      "step": 1580
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5e-06,
+      "loss": 0.2693,
+      "step": 1581
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5e-06,
+      "loss": 0.2679,
+      "step": 1582
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5e-06,
+      "loss": 0.2716,
+      "step": 1583
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5e-06,
+      "loss": 0.2189,
+      "step": 1584
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5e-06,
+      "loss": 0.2613,
+      "step": 1585
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5e-06,
+      "loss": 0.2774,
+      "step": 1586
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5e-06,
+      "loss": 0.277,
+      "step": 1587
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5e-06,
+      "loss": 0.2511,
+      "step": 1588
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5e-06,
+      "loss": 0.2568,
+      "step": 1589
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5e-06,
+      "loss": 0.309,
+      "step": 1590
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5e-06,
+      "loss": 0.2765,
+      "step": 1591
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5e-06,
+      "loss": 0.2659,
+      "step": 1592
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5e-06,
+      "loss": 0.2606,
+      "step": 1593
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5e-06,
+      "loss": 0.2371,
+      "step": 1594
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5e-06,
+      "loss": 0.243,
+      "step": 1595
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5e-06,
+      "loss": 0.3037,
+      "step": 1596
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5e-06,
+      "loss": 0.2459,
+      "step": 1597
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5e-06,
+      "loss": 0.2296,
+      "step": 1598
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5e-06,
+      "loss": 0.2337,
+      "step": 1599
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5e-06,
+      "loss": 0.2739,
+      "step": 1600
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5e-06,
+      "loss": 0.2558,
+      "step": 1601
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5e-06,
+      "loss": 0.294,
+      "step": 1602
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5e-06,
+      "loss": 0.2395,
+      "step": 1603
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5e-06,
+      "loss": 0.3244,
+      "step": 1604
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5e-06,
+      "loss": 0.2711,
+      "step": 1605
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5e-06,
+      "loss": 0.225,
+      "step": 1606
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5e-06,
+      "loss": 0.2512,
+      "step": 1607
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5e-06,
+      "loss": 0.2528,
+      "step": 1608
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5e-06,
+      "loss": 0.2589,
+      "step": 1609
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5e-06,
+      "loss": 0.284,
+      "step": 1610
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5e-06,
+      "loss": 0.2293,
+      "step": 1611
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5e-06,
+      "loss": 0.2561,
+      "step": 1612
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5e-06,
+      "loss": 0.2448,
+      "step": 1613
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5e-06,
+      "loss": 0.2725,
+      "step": 1614
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5e-06,
+      "loss": 0.2422,
+      "step": 1615
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5e-06,
+      "loss": 0.2618,
+      "step": 1616
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5e-06,
+      "loss": 0.2235,
+      "step": 1617
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5e-06,
+      "loss": 0.2303,
+      "step": 1618
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5e-06,
+      "loss": 0.2395,
+      "step": 1619
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5e-06,
+      "loss": 0.3189,
+      "step": 1620
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5e-06,
+      "loss": 0.2452,
+      "step": 1621
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5e-06,
+      "loss": 0.2561,
+      "step": 1622
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5e-06,
+      "loss": 0.2802,
+      "step": 1623
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5e-06,
+      "loss": 0.2581,
+      "step": 1624
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5e-06,
+      "loss": 0.2429,
+      "step": 1625
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5e-06,
+      "loss": 0.2789,
+      "step": 1626
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5e-06,
+      "loss": 0.248,
+      "step": 1627
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5e-06,
+      "loss": 0.2266,
+      "step": 1628
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5e-06,
+      "loss": 0.2766,
+      "step": 1629
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5e-06,
+      "loss": 0.2712,
+      "step": 1630
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5e-06,
+      "loss": 0.2602,
+      "step": 1631
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5e-06,
+      "loss": 0.2671,
+      "step": 1632
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5e-06,
+      "loss": 0.2384,
+      "step": 1633
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5e-06,
+      "loss": 0.2394,
+      "step": 1634
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5e-06,
+      "loss": 0.288,
+      "step": 1635
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5e-06,
+      "loss": 0.2824,
+      "step": 1636
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5e-06,
+      "loss": 0.2619,
+      "step": 1637
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5e-06,
+      "loss": 0.2332,
+      "step": 1638
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5e-06,
+      "loss": 0.256,
+      "step": 1639
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5e-06,
+      "loss": 0.2877,
+      "step": 1640
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5e-06,
+      "loss": 0.2862,
+      "step": 1641
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5e-06,
+      "loss": 0.2413,
+      "step": 1642
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5e-06,
+      "loss": 0.2676,
+      "step": 1643
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5e-06,
+      "loss": 0.2341,
+      "step": 1644
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5e-06,
+      "loss": 0.2792,
+      "step": 1645
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5e-06,
+      "loss": 0.2715,
+      "step": 1646
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5e-06,
+      "loss": 0.2867,
+      "step": 1647
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5e-06,
+      "loss": 0.2355,
+      "step": 1648
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5e-06,
+      "loss": 0.2506,
+      "step": 1649
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5e-06,
+      "loss": 0.2626,
+      "step": 1650
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5e-06,
+      "loss": 0.2522,
+      "step": 1651
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5e-06,
+      "loss": 0.254,
+      "step": 1652
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5e-06,
+      "loss": 0.2608,
+      "step": 1653
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5e-06,
+      "loss": 0.2808,
+      "step": 1654
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5e-06,
+      "loss": 0.2415,
+      "step": 1655
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5e-06,
+      "loss": 0.2528,
+      "step": 1656
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5e-06,
+      "loss": 0.2862,
+      "step": 1657
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5e-06,
+      "loss": 0.2624,
+      "step": 1658
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5e-06,
+      "loss": 0.2356,
+      "step": 1659
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5e-06,
+      "loss": 0.2773,
+      "step": 1660
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5e-06,
+      "loss": 0.254,
+      "step": 1661
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5e-06,
+      "loss": 0.2611,
+      "step": 1662
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5e-06,
+      "loss": 0.245,
+      "step": 1663
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5e-06,
+      "loss": 0.2612,
+      "step": 1664
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5e-06,
+      "loss": 0.2662,
+      "step": 1665
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5e-06,
+      "loss": 0.2264,
+      "step": 1666
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5e-06,
+      "loss": 0.253,
+      "step": 1667
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5e-06,
+      "loss": 0.2752,
+      "step": 1668
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5e-06,
+      "loss": 0.2695,
+      "step": 1669
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5e-06,
+      "loss": 0.2392,
+      "step": 1670
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5e-06,
+      "loss": 0.2739,
+      "step": 1671
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5e-06,
+      "loss": 0.2576,
+      "step": 1672
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5e-06,
+      "loss": 0.2525,
+      "step": 1673
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5e-06,
+      "loss": 0.2705,
+      "step": 1674
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5e-06,
+      "loss": 0.2487,
+      "step": 1675
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5e-06,
+      "loss": 0.2287,
+      "step": 1676
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5e-06,
+      "loss": 0.2696,
+      "step": 1677
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5e-06,
+      "loss": 0.2849,
+      "step": 1678
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5e-06,
+      "loss": 0.2848,
+      "step": 1679
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5e-06,
+      "loss": 0.264,
+      "step": 1680
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5e-06,
+      "loss": 0.2419,
+      "step": 1681
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5e-06,
+      "loss": 0.2615,
+      "step": 1682
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5e-06,
+      "loss": 0.2408,
+      "step": 1683
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5e-06,
+      "loss": 0.2436,
+      "step": 1684
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5e-06,
+      "loss": 0.245,
+      "step": 1685
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5e-06,
+      "loss": 0.2439,
+      "step": 1686
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5e-06,
+      "loss": 0.2353,
+      "step": 1687
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5e-06,
+      "loss": 0.2801,
+      "step": 1688
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5e-06,
+      "loss": 0.2644,
+      "step": 1689
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5e-06,
+      "loss": 0.2944,
+      "step": 1690
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5e-06,
+      "loss": 0.2323,
+      "step": 1691
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5e-06,
+      "loss": 0.2366,
+      "step": 1692
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5e-06,
+      "loss": 0.2233,
+      "step": 1693
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5e-06,
+      "loss": 0.2327,
+      "step": 1694
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5e-06,
+      "loss": 0.2819,
+      "step": 1695
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5e-06,
+      "loss": 0.2722,
+      "step": 1696
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5e-06,
+      "loss": 0.2434,
+      "step": 1697
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5e-06,
+      "loss": 0.2707,
+      "step": 1698
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5e-06,
+      "loss": 0.2257,
+      "step": 1699
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5e-06,
+      "loss": 0.2326,
+      "step": 1700
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5e-06,
+      "loss": 0.2621,
+      "step": 1701
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5e-06,
+      "loss": 0.2652,
+      "step": 1702
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5e-06,
+      "loss": 0.25,
+      "step": 1703
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5e-06,
+      "loss": 0.303,
+      "step": 1704
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5e-06,
+      "loss": 0.2746,
+      "step": 1705
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5e-06,
+      "loss": 0.2427,
+      "step": 1706
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5e-06,
+      "loss": 0.2685,
+      "step": 1707
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5e-06,
+      "loss": 0.2693,
+      "step": 1708
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5e-06,
+      "loss": 0.2786,
+      "step": 1709
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5e-06,
+      "loss": 0.3014,
+      "step": 1710
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5e-06,
+      "loss": 0.2448,
+      "step": 1711
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5e-06,
+      "loss": 0.2943,
+      "step": 1712
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5e-06,
+      "loss": 0.2728,
+      "step": 1713
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5e-06,
+      "loss": 0.2504,
+      "step": 1714
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5e-06,
+      "loss": 0.242,
+      "step": 1715
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5e-06,
+      "loss": 0.2476,
+      "step": 1716
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5e-06,
+      "loss": 0.2811,
+      "step": 1717
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5e-06,
+      "loss": 0.2255,
+      "step": 1718
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5e-06,
+      "loss": 0.2518,
+      "step": 1719
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5e-06,
+      "loss": 0.2378,
+      "step": 1720
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5e-06,
+      "loss": 0.2826,
+      "step": 1721
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5e-06,
+      "loss": 0.2637,
+      "step": 1722
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5e-06,
+      "loss": 0.2663,
+      "step": 1723
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5e-06,
+      "loss": 0.254,
+      "step": 1724
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5e-06,
+      "loss": 0.2825,
+      "step": 1725
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5e-06,
+      "loss": 0.2475,
+      "step": 1726
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5e-06,
+      "loss": 0.253,
+      "step": 1727
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5e-06,
+      "loss": 0.281,
+      "step": 1728
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5e-06,
+      "loss": 0.2245,
+      "step": 1729
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5e-06,
+      "loss": 0.2082,
+      "step": 1730
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5e-06,
+      "loss": 0.2679,
+      "step": 1731
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5e-06,
+      "loss": 0.2736,
+      "step": 1732
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5e-06,
+      "loss": 0.2703,
+      "step": 1733
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5e-06,
+      "loss": 0.3013,
+      "step": 1734
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5e-06,
+      "loss": 0.265,
+      "step": 1735
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5e-06,
+      "loss": 0.2312,
+      "step": 1736
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5e-06,
+      "loss": 0.2717,
+      "step": 1737
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5e-06,
+      "loss": 0.291,
+      "step": 1738
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5e-06,
+      "loss": 0.2612,
+      "step": 1739
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5e-06,
+      "loss": 0.2477,
+      "step": 1740
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5e-06,
+      "loss": 0.2687,
+      "step": 1741
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5e-06,
+      "loss": 0.2724,
+      "step": 1742
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5e-06,
+      "loss": 0.2802,
+      "step": 1743
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5e-06,
+      "loss": 0.2557,
+      "step": 1744
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5e-06,
+      "loss": 0.2078,
+      "step": 1745
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5e-06,
+      "loss": 0.2598,
+      "step": 1746
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5e-06,
+      "loss": 0.2389,
+      "step": 1747
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5e-06,
+      "loss": 0.2645,
+      "step": 1748
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5e-06,
+      "loss": 0.2546,
+      "step": 1749
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5e-06,
+      "loss": 0.2524,
+      "step": 1750
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5e-06,
+      "loss": 0.2134,
+      "step": 1751
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5e-06,
+      "loss": 0.2616,
+      "step": 1752
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5e-06,
+      "loss": 0.2329,
+      "step": 1753
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5e-06,
+      "loss": 0.2017,
+      "step": 1754
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5e-06,
+      "loss": 0.2507,
+      "step": 1755
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5e-06,
+      "loss": 0.2278,
+      "step": 1756
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5e-06,
+      "loss": 0.2335,
+      "step": 1757
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5e-06,
+      "loss": 0.2724,
+      "step": 1758
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5e-06,
+      "loss": 0.2929,
+      "step": 1759
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5e-06,
+      "loss": 0.2233,
+      "step": 1760
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5e-06,
+      "loss": 0.2133,
+      "step": 1761
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5e-06,
+      "loss": 0.276,
+      "step": 1762
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5e-06,
+      "loss": 0.2692,
+      "step": 1763
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5e-06,
+      "loss": 0.2709,
+      "step": 1764
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5e-06,
+      "loss": 0.2565,
+      "step": 1765
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5e-06,
+      "loss": 0.2608,
+      "step": 1766
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5e-06,
+      "loss": 0.2145,
+      "step": 1767
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5e-06,
+      "loss": 0.2726,
+      "step": 1768
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5e-06,
+      "loss": 0.2449,
+      "step": 1769
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5e-06,
+      "loss": 0.2544,
+      "step": 1770
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5e-06,
+      "loss": 0.2474,
+      "step": 1771
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5e-06,
+      "loss": 0.2766,
+      "step": 1772
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5e-06,
+      "loss": 0.2765,
+      "step": 1773
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5e-06,
+      "loss": 0.2538,
+      "step": 1774
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5e-06,
+      "loss": 0.2348,
+      "step": 1775
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5e-06,
+      "loss": 0.2353,
+      "step": 1776
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5e-06,
+      "loss": 0.2656,
+      "step": 1777
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5e-06,
+      "loss": 0.2598,
+      "step": 1778
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5e-06,
+      "loss": 0.2732,
+      "step": 1779
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5e-06,
+      "loss": 0.2567,
+      "step": 1780
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5e-06,
+      "loss": 0.2814,
+      "step": 1781
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5e-06,
+      "loss": 0.2606,
+      "step": 1782
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5e-06,
+      "loss": 0.2688,
+      "step": 1783
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5e-06,
+      "loss": 0.3062,
+      "step": 1784
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5e-06,
+      "loss": 0.299,
+      "step": 1785
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5e-06,
+      "loss": 0.2479,
+      "step": 1786
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5e-06,
+      "loss": 0.2565,
+      "step": 1787
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5e-06,
+      "loss": 0.2556,
+      "step": 1788
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5e-06,
+      "loss": 0.3031,
+      "step": 1789
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5e-06,
+      "loss": 0.2433,
+      "step": 1790
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5e-06,
+      "loss": 0.2748,
+      "step": 1791
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5e-06,
+      "loss": 0.2799,
+      "step": 1792
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5e-06,
+      "loss": 0.2565,
+      "step": 1793
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5e-06,
+      "loss": 0.2397,
+      "step": 1794
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5e-06,
+      "loss": 0.2796,
+      "step": 1795
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5e-06,
+      "loss": 0.2435,
+      "step": 1796
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5e-06,
+      "loss": 0.2656,
+      "step": 1797
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5e-06,
+      "loss": 0.294,
+      "step": 1798
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5e-06,
+      "loss": 0.2336,
+      "step": 1799
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5e-06,
+      "loss": 0.2388,
+      "step": 1800
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5e-06,
+      "loss": 0.2838,
+      "step": 1801
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5e-06,
+      "loss": 0.283,
+      "step": 1802
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5e-06,
+      "loss": 0.2586,
+      "step": 1803
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5e-06,
+      "loss": 0.23,
+      "step": 1804
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5e-06,
+      "loss": 0.2642,
+      "step": 1805
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5e-06,
+      "loss": 0.2332,
+      "step": 1806
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5e-06,
+      "loss": 0.3018,
+      "step": 1807
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5e-06,
+      "loss": 0.2399,
+      "step": 1808
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5e-06,
+      "loss": 0.2561,
+      "step": 1809
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5e-06,
+      "loss": 0.2684,
+      "step": 1810
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5e-06,
+      "loss": 0.2682,
+      "step": 1811
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5e-06,
+      "loss": 0.2911,
+      "step": 1812
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5e-06,
+      "loss": 0.2588,
+      "step": 1813
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5e-06,
+      "loss": 0.2679,
+      "step": 1814
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5e-06,
+      "loss": 0.2728,
+      "step": 1815
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5e-06,
+      "loss": 0.258,
+      "step": 1816
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5e-06,
+      "loss": 0.2595,
+      "step": 1817
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5e-06,
+      "loss": 0.2161,
+      "step": 1818
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5e-06,
+      "loss": 0.3226,
+      "step": 1819
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5e-06,
+      "loss": 0.2533,
+      "step": 1820
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5e-06,
+      "loss": 0.2934,
+      "step": 1821
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5e-06,
+      "loss": 0.2795,
+      "step": 1822
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5e-06,
+      "loss": 0.2791,
+      "step": 1823
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5e-06,
+      "loss": 0.2911,
+      "step": 1824
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5e-06,
+      "loss": 0.2791,
+      "step": 1825
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5e-06,
+      "loss": 0.2889,
+      "step": 1826
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5e-06,
+      "loss": 0.2597,
+      "step": 1827
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5e-06,
+      "loss": 0.2675,
+      "step": 1828
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5e-06,
+      "loss": 0.2458,
+      "step": 1829
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5e-06,
+      "loss": 0.2839,
+      "step": 1830
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5e-06,
+      "loss": 0.2455,
+      "step": 1831
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5e-06,
+      "loss": 0.2453,
+      "step": 1832
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5e-06,
+      "loss": 0.2853,
+      "step": 1833
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5e-06,
+      "loss": 0.2681,
+      "step": 1834
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5e-06,
+      "loss": 0.2881,
+      "step": 1835
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5e-06,
+      "loss": 0.2774,
+      "step": 1836
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5e-06,
+      "loss": 0.245,
+      "step": 1837
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5e-06,
+      "loss": 0.2731,
+      "step": 1838
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5e-06,
+      "loss": 0.2576,
+      "step": 1839
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5e-06,
+      "loss": 0.2793,
+      "step": 1840
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5e-06,
+      "loss": 0.283,
+      "step": 1841
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5e-06,
+      "loss": 0.2698,
+      "step": 1842
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5e-06,
+      "loss": 0.2846,
+      "step": 1843
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5e-06,
+      "loss": 0.2654,
+      "step": 1844
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5e-06,
+      "loss": 0.2867,
+      "step": 1845
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5e-06,
+      "loss": 0.26,
+      "step": 1846
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5e-06,
+      "loss": 0.2722,
+      "step": 1847
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5e-06,
+      "loss": 0.2297,
+      "step": 1848
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5e-06,
+      "loss": 0.2736,
+      "step": 1849
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5e-06,
+      "loss": 0.2783,
+      "step": 1850
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5e-06,
+      "loss": 0.2873,
+      "step": 1851
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5e-06,
+      "loss": 0.2448,
+      "step": 1852
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5e-06,
+      "loss": 0.2493,
+      "step": 1853
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5e-06,
+      "loss": 0.2752,
+      "step": 1854
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5e-06,
+      "loss": 0.2764,
+      "step": 1855
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5e-06,
+      "loss": 0.2148,
+      "step": 1856
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5e-06,
+      "loss": 0.2655,
+      "step": 1857
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5e-06,
+      "loss": 0.2665,
+      "step": 1858
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5e-06,
+      "loss": 0.2646,
+      "step": 1859
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5e-06,
+      "loss": 0.2646,
+      "step": 1860
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5e-06,
+      "loss": 0.2488,
+      "step": 1861
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5e-06,
+      "loss": 0.2731,
+      "step": 1862
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5e-06,
+      "loss": 0.2638,
+      "step": 1863
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5e-06,
+      "loss": 0.2682,
+      "step": 1864
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5e-06,
+      "loss": 0.2525,
+      "step": 1865
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5e-06,
+      "loss": 0.2486,
+      "step": 1866
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5e-06,
+      "loss": 0.2571,
+      "step": 1867
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5e-06,
+      "loss": 0.2646,
+      "step": 1868
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5e-06,
+      "loss": 0.2425,
+      "step": 1869
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5e-06,
+      "loss": 0.23,
+      "step": 1870
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5e-06,
+      "loss": 0.2646,
+      "step": 1871
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5e-06,
+      "loss": 0.2719,
+      "step": 1872
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5e-06,
+      "loss": 0.3163,
+      "step": 1873
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5e-06,
+      "loss": 0.2584,
+      "step": 1874
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5e-06,
+      "loss": 0.2489,
+      "step": 1875
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "step": 1876
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5e-06,
+      "loss": 0.1175,
+      "step": 1877
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "step": 1878
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "step": 1879
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5e-06,
+      "loss": 0.0856,
+      "step": 1880
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "step": 1881
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5e-06,
+      "loss": 0.0888,
+      "step": 1882
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "step": 1883
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "step": 1884
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5e-06,
+      "loss": 0.1551,
+      "step": 1885
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5e-06,
+      "loss": 0.1833,
+      "step": 1886
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5e-06,
+      "loss": 0.0789,
+      "step": 1887
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5e-06,
+      "loss": 0.1291,
+      "step": 1888
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "step": 1889
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "step": 1890
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "step": 1891
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5e-06,
+      "loss": 0.1261,
+      "step": 1892
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5e-06,
+      "loss": 0.0829,
+      "step": 1893
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5e-06,
+      "loss": 0.1094,
+      "step": 1894
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5e-06,
+      "loss": 0.0916,
+      "step": 1895
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5e-06,
+      "loss": 0.1062,
+      "step": 1896
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "step": 1897
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5e-06,
+      "loss": 0.0766,
+      "step": 1898
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5e-06,
+      "loss": 0.1154,
+      "step": 1899
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5e-06,
+      "loss": 0.089,
+      "step": 1900
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5e-06,
+      "loss": 0.0679,
+      "step": 1901
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "step": 1902
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5e-06,
+      "loss": 0.1072,
+      "step": 1903
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5e-06,
+      "loss": 0.0812,
+      "step": 1904
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "step": 1905
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "step": 1906
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "step": 1907
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "step": 1908
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5e-06,
+      "loss": 0.1197,
+      "step": 1909
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "step": 1910
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5e-06,
+      "loss": 0.0835,
+      "step": 1911
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5e-06,
+      "loss": 0.0744,
+      "step": 1912
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5e-06,
+      "loss": 0.0731,
+      "step": 1913
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "step": 1914
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5e-06,
+      "loss": 0.1155,
+      "step": 1915
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5e-06,
+      "loss": 0.0928,
+      "step": 1916
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "step": 1917
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5e-06,
+      "loss": 0.0757,
+      "step": 1918
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5e-06,
+      "loss": 0.0954,
+      "step": 1919
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5e-06,
+      "loss": 0.0924,
+      "step": 1920
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5e-06,
+      "loss": 0.0774,
+      "step": 1921
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5e-06,
+      "loss": 0.0846,
+      "step": 1922
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "step": 1923
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5e-06,
+      "loss": 0.0812,
+      "step": 1924
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "step": 1925
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5e-06,
+      "loss": 0.1342,
+      "step": 1926
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "step": 1927
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "step": 1928
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "step": 1929
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5e-06,
+      "loss": 0.0645,
+      "step": 1930
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5e-06,
+      "loss": 0.1196,
+      "step": 1931
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5e-06,
+      "loss": 0.1061,
+      "step": 1932
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "step": 1933
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5e-06,
+      "loss": 0.0889,
+      "step": 1934
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5e-06,
+      "loss": 0.1186,
+      "step": 1935
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5e-06,
+      "loss": 0.0887,
+      "step": 1936
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5e-06,
+      "loss": 0.0831,
+      "step": 1937
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5e-06,
+      "loss": 0.0883,
+      "step": 1938
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5e-06,
+      "loss": 0.0756,
+      "step": 1939
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5e-06,
+      "loss": 0.1075,
+      "step": 1940
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "step": 1941
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "step": 1942
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "step": 1943
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5e-06,
+      "loss": 0.1085,
+      "step": 1944
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5e-06,
+      "loss": 0.0875,
+      "step": 1945
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "step": 1946
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5e-06,
+      "loss": 0.1824,
+      "step": 1947
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5e-06,
+      "loss": 0.1037,
+      "step": 1948
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5e-06,
+      "loss": 0.1098,
+      "step": 1949
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5e-06,
+      "loss": 0.0703,
+      "step": 1950
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "step": 1951
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "step": 1952
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5e-06,
+      "loss": 0.0802,
+      "step": 1953
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5e-06,
+      "loss": 0.0995,
+      "step": 1954
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5e-06,
+      "loss": 0.0803,
+      "step": 1955
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5e-06,
+      "loss": 0.0879,
+      "step": 1956
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5e-06,
+      "loss": 0.0885,
+      "step": 1957
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5e-06,
+      "loss": 0.083,
+      "step": 1958
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5e-06,
+      "loss": 0.1154,
+      "step": 1959
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5e-06,
+      "loss": 0.1005,
+      "step": 1960
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "step": 1961
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5e-06,
+      "loss": 0.0963,
+      "step": 1962
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5e-06,
+      "loss": 0.0862,
+      "step": 1963
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "step": 1964
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "step": 1965
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5e-06,
+      "loss": 0.0903,
+      "step": 1966
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "step": 1967
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5e-06,
+      "loss": 0.1086,
+      "step": 1968
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "step": 1969
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5e-06,
+      "loss": 0.1368,
+      "step": 1970
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5e-06,
+      "loss": 0.0754,
+      "step": 1971
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "step": 1972
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5e-06,
+      "loss": 0.0848,
+      "step": 1973
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "step": 1974
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5e-06,
+      "loss": 0.0883,
+      "step": 1975
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "step": 1976
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5e-06,
+      "loss": 0.092,
+      "step": 1977
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5e-06,
+      "loss": 0.1071,
+      "step": 1978
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5e-06,
+      "loss": 0.1141,
+      "step": 1979
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5e-06,
+      "loss": 0.0825,
+      "step": 1980
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "step": 1981
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "step": 1982
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "step": 1983
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5e-06,
+      "loss": 0.0862,
+      "step": 1984
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5e-06,
+      "loss": 0.0694,
+      "step": 1985
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5e-06,
+      "loss": 0.0995,
+      "step": 1986
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "step": 1987
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5e-06,
+      "loss": 0.1029,
+      "step": 1988
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5e-06,
+      "loss": 0.0988,
+      "step": 1989
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5e-06,
+      "loss": 0.0935,
+      "step": 1990
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "step": 1991
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5e-06,
+      "loss": 0.1422,
+      "step": 1992
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5e-06,
+      "loss": 0.1241,
+      "step": 1993
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5e-06,
+      "loss": 0.0975,
+      "step": 1994
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5e-06,
+      "loss": 0.1319,
+      "step": 1995
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5e-06,
+      "loss": 0.113,
+      "step": 1996
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "step": 1997
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5e-06,
+      "loss": 0.1067,
+      "step": 1998
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "step": 1999
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5e-06,
+      "loss": 0.1302,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3744,
+  "num_train_epochs": 8,
+  "save_steps": 500,
+  "total_flos": 4.4775232025237914e+17,
+  "trial_name": null,
+  "trial_params": null
+}