diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,47539 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 20.0,
+  "eval_steps": 500,
+  "global_step": 7920,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1008403361344538e-07,
+      "loss": 1.5876,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2016806722689076e-07,
+      "loss": 1.5182,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.302521008403362e-07,
+      "loss": 1.5689,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.403361344537815e-07,
+      "loss": 1.5553,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0504201680672271e-06,
+      "loss": 1.6128,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2605042016806724e-06,
+      "loss": 1.6315,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 1.5931,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.680672268907563e-06,
+      "loss": 1.5025,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8907563025210083e-06,
+      "loss": 1.5827,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1008403361344543e-06,
+      "loss": 1.6094,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3109243697478996e-06,
+      "loss": 1.5292,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.521008403361345e-06,
+      "loss": 1.5733,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.73109243697479e-06,
+      "loss": 1.5625,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.73109243697479e-06,
+      "loss": 1.5231,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 1.5277,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.1512605042016808e-06,
+      "loss": 1.6315,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.361344537815126e-06,
+      "loss": 1.5733,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 1.6953,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.7815126050420167e-06,
+      "loss": 1.5537,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.991596638655462e-06,
+      "loss": 1.5977,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.2016806722689085e-06,
+      "loss": 1.5781,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 1.6193,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.621848739495799e-06,
+      "loss": 1.5887,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.831932773109244e-06,
+      "loss": 1.5289,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.04201680672269e-06,
+      "loss": 1.532,
+      "step": 25
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.252100840336135e-06,
+      "loss": 1.5904,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.46218487394958e-06,
+      "loss": 1.5379,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.672268907563025e-06,
+      "loss": 1.547,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 1.7111,
+      "step": 29
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.092436974789916e-06,
+      "loss": 1.7052,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.3025210084033615e-06,
+      "loss": 1.584,
+      "step": 31
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.512605042016807e-06,
+      "loss": 1.4648,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.722689075630252e-06,
+      "loss": 1.593,
+      "step": 33
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.932773109243698e-06,
+      "loss": 1.4343,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 1.651,
+      "step": 35
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.3529411764705884e-06,
+      "loss": 1.4766,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.563025210084033e-06,
+      "loss": 1.5839,
+      "step": 37
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.773109243697479e-06,
+      "loss": 1.562,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.983193277310924e-06,
+      "loss": 1.4845,
+      "step": 39
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.19327731092437e-06,
+      "loss": 1.5984,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.403361344537817e-06,
+      "loss": 1.5673,
+      "step": 41
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.61344537815126e-06,
+      "loss": 1.3826,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 1.4052,
+      "step": 43
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.033613445378152e-06,
+      "loss": 1.5043,
+      "step": 44
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.243697478991598e-06,
+      "loss": 1.4258,
+      "step": 45
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.453781512605041e-06,
+      "loss": 1.4779,
+      "step": 46
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.663865546218488e-06,
+      "loss": 1.4838,
+      "step": 47
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.873949579831933e-06,
+      "loss": 1.3363,
+      "step": 48
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.008403361344538e-05,
+      "loss": 1.4055,
+      "step": 49
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0294117647058824e-05,
+      "loss": 1.3249,
+      "step": 50
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.050420168067227e-05,
+      "loss": 1.328,
+      "step": 51
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 1.5204,
+      "step": 52
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.092436974789916e-05,
+      "loss": 1.3939,
+      "step": 53
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1134453781512606e-05,
+      "loss": 1.2778,
+      "step": 54
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.134453781512605e-05,
+      "loss": 1.3495,
+      "step": 55
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1554621848739497e-05,
+      "loss": 1.3314,
+      "step": 56
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 1.4842,
+      "step": 57
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.1974789915966388e-05,
+      "loss": 1.3994,
+      "step": 58
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.2184873949579832e-05,
+      "loss": 1.2979,
+      "step": 59
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.2394957983193278e-05,
+      "loss": 1.311,
+      "step": 60
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.2605042016806723e-05,
+      "loss": 1.3971,
+      "step": 61
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.2815126050420168e-05,
+      "loss": 1.3584,
+      "step": 62
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3025210084033614e-05,
+      "loss": 1.2716,
+      "step": 63
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.323529411764706e-05,
+      "loss": 1.2459,
+      "step": 64
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3445378151260504e-05,
+      "loss": 1.4626,
+      "step": 65
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3655462184873949e-05,
+      "loss": 1.4032,
+      "step": 66
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3865546218487396e-05,
+      "loss": 1.3199,
+      "step": 67
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.407563025210084e-05,
+      "loss": 1.2999,
+      "step": 68
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 1.4247,
+      "step": 69
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4495798319327734e-05,
+      "loss": 1.2854,
+      "step": 70
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 1.271,
+      "step": 71
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4915966386554622e-05,
+      "loss": 1.3709,
+      "step": 72
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5126050420168067e-05,
+      "loss": 1.5038,
+      "step": 73
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.5336134453781513e-05,
+      "loss": 1.1235,
+      "step": 74
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.5546218487394958e-05,
+      "loss": 1.364,
+      "step": 75
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.5756302521008403e-05,
+      "loss": 1.2995,
+      "step": 76
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.5966386554621848e-05,
+      "loss": 1.1678,
+      "step": 77
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6176470588235296e-05,
+      "loss": 1.1212,
+      "step": 78
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.638655462184874e-05,
+      "loss": 1.3429,
+      "step": 79
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6596638655462186e-05,
+      "loss": 1.3052,
+      "step": 80
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6806722689075634e-05,
+      "loss": 1.3764,
+      "step": 81
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7016806722689076e-05,
+      "loss": 1.3205,
+      "step": 82
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.722689075630252e-05,
+      "loss": 1.3059,
+      "step": 83
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7436974789915965e-05,
+      "loss": 1.32,
+      "step": 84
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 1.258,
+      "step": 85
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 1.281,
+      "step": 86
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8067226890756303e-05,
+      "loss": 1.3335,
+      "step": 87
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8277310924369748e-05,
+      "loss": 1.2937,
+      "step": 88
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8487394957983196e-05,
+      "loss": 1.0739,
+      "step": 89
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.869747899159664e-05,
+      "loss": 1.3757,
+      "step": 90
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8907563025210083e-05,
+      "loss": 1.3711,
+      "step": 91
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9117647058823528e-05,
+      "loss": 1.1688,
+      "step": 92
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9327731092436976e-05,
+      "loss": 1.2692,
+      "step": 93
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.953781512605042e-05,
+      "loss": 1.354,
+      "step": 94
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9747899159663866e-05,
+      "loss": 1.3778,
+      "step": 95
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9957983193277314e-05,
+      "loss": 1.1939,
+      "step": 96
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.016806722689076e-05,
+      "loss": 1.2572,
+      "step": 97
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.0378151260504204e-05,
+      "loss": 1.4585,
+      "step": 98
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.058823529411765e-05,
+      "loss": 1.3385,
+      "step": 99
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.0798319327731094e-05,
+      "loss": 1.2751,
+      "step": 100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.100840336134454e-05,
+      "loss": 1.2456,
+      "step": 101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.1218487394957983e-05,
+      "loss": 1.0534,
+      "step": 102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 1.0753,
+      "step": 103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.1638655462184876e-05,
+      "loss": 1.344,
+      "step": 104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.184873949579832e-05,
+      "loss": 1.2536,
+      "step": 105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2058823529411766e-05,
+      "loss": 1.2373,
+      "step": 106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.226890756302521e-05,
+      "loss": 1.2278,
+      "step": 107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2478991596638656e-05,
+      "loss": 1.1401,
+      "step": 108
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.26890756302521e-05,
+      "loss": 1.3442,
+      "step": 109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.2899159663865546e-05,
+      "loss": 1.1057,
+      "step": 110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.3109243697478994e-05,
+      "loss": 1.1379,
+      "step": 111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.331932773109244e-05,
+      "loss": 1.1742,
+      "step": 112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 1.2117,
+      "step": 113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.373949579831933e-05,
+      "loss": 1.2152,
+      "step": 114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.3949579831932777e-05,
+      "loss": 1.3178,
+      "step": 115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.415966386554622e-05,
+      "loss": 1.2022,
+      "step": 116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.4369747899159663e-05,
+      "loss": 1.1978,
+      "step": 117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.4579831932773108e-05,
+      "loss": 1.2136,
+      "step": 118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.4789915966386556e-05,
+      "loss": 1.0799,
+      "step": 119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5e-05,
+      "loss": 1.1831,
+      "step": 120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5210084033613446e-05,
+      "loss": 1.2642,
+      "step": 121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.542016806722689e-05,
+      "loss": 1.2713,
+      "step": 122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5630252100840336e-05,
+      "loss": 1.151,
+      "step": 123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5840336134453784e-05,
+      "loss": 1.3836,
+      "step": 124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.605042016806723e-05,
+      "loss": 1.0946,
+      "step": 125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.6260504201680674e-05,
+      "loss": 1.3823,
+      "step": 126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 1.3451,
+      "step": 127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.6680672268907564e-05,
+      "loss": 1.24,
+      "step": 128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.689075630252101e-05,
+      "loss": 1.3355,
+      "step": 129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7100840336134453e-05,
+      "loss": 1.0377,
+      "step": 130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7310924369747898e-05,
+      "loss": 1.3199,
+      "step": 131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.7521008403361346e-05,
+      "loss": 1.1339,
+      "step": 132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.773109243697479e-05,
+      "loss": 1.173,
+      "step": 133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.7941176470588236e-05,
+      "loss": 1.3031,
+      "step": 134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.815126050420168e-05,
+      "loss": 1.1791,
+      "step": 135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8361344537815126e-05,
+      "loss": 1.0673,
+      "step": 136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 1.2011,
+      "step": 137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8781512605042016e-05,
+      "loss": 1.3557,
+      "step": 138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8991596638655467e-05,
+      "loss": 1.217,
+      "step": 139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9201680672268912e-05,
+      "loss": 1.232,
+      "step": 140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 1.2552,
+      "step": 141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.96218487394958e-05,
+      "loss": 1.2206,
+      "step": 142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9831932773109244e-05,
+      "loss": 1.3544,
+      "step": 143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.004201680672269e-05,
+      "loss": 1.2189,
+      "step": 144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.0252100840336133e-05,
+      "loss": 1.2766,
+      "step": 145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.0462184873949578e-05,
+      "loss": 1.2199,
+      "step": 146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.0672268907563026e-05,
+      "loss": 1.1241,
+      "step": 147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.0882352941176475e-05,
+      "loss": 1.3088,
+      "step": 148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1092436974789916e-05,
+      "loss": 1.1277,
+      "step": 149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1302521008403364e-05,
+      "loss": 1.2857,
+      "step": 150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1512605042016806e-05,
+      "loss": 1.1811,
+      "step": 151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1722689075630254e-05,
+      "loss": 1.3894,
+      "step": 152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.1932773109243696e-05,
+      "loss": 1.1439,
+      "step": 153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 1.2718,
+      "step": 154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.235294117647059e-05,
+      "loss": 1.3236,
+      "step": 155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.2563025210084034e-05,
+      "loss": 1.2592,
+      "step": 156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.277310924369748e-05,
+      "loss": 1.2368,
+      "step": 157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2983193277310923e-05,
+      "loss": 1.2567,
+      "step": 158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.319327731092437e-05,
+      "loss": 1.3338,
+      "step": 159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.340336134453781e-05,
+      "loss": 1.2339,
+      "step": 160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.361344537815127e-05,
+      "loss": 1.2439,
+      "step": 161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.382352941176471e-05,
+      "loss": 1.1949,
+      "step": 162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.403361344537815e-05,
+      "loss": 1.2615,
+      "step": 163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.42436974789916e-05,
+      "loss": 1.2662,
+      "step": 164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.445378151260504e-05,
+      "loss": 1.4243,
+      "step": 165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.466386554621849e-05,
+      "loss": 1.1379,
+      "step": 166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.487394957983193e-05,
+      "loss": 1.0432,
+      "step": 167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.508403361344538e-05,
+      "loss": 1.2545,
+      "step": 168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 1.0497,
+      "step": 169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5504201680672275e-05,
+      "loss": 1.2193,
+      "step": 170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 1.1459,
+      "step": 171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.592436974789916e-05,
+      "loss": 1.17,
+      "step": 172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.613445378151261e-05,
+      "loss": 1.1217,
+      "step": 173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.634453781512605e-05,
+      "loss": 1.2967,
+      "step": 174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6554621848739496e-05,
+      "loss": 1.1275,
+      "step": 175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6764705882352945e-05,
+      "loss": 1.2771,
+      "step": 176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.697478991596639e-05,
+      "loss": 1.14,
+      "step": 177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.7184873949579834e-05,
+      "loss": 1.2857,
+      "step": 178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.739495798319328e-05,
+      "loss": 1.3195,
+      "step": 179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.7605042016806724e-05,
+      "loss": 1.217,
+      "step": 180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.7815126050420166e-05,
+      "loss": 1.1541,
+      "step": 181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8025210084033614e-05,
+      "loss": 1.172,
+      "step": 182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8235294117647055e-05,
+      "loss": 1.3176,
+      "step": 183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.844537815126051e-05,
+      "loss": 1.2697,
+      "step": 184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.865546218487395e-05,
+      "loss": 1.0679,
+      "step": 185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.88655462184874e-05,
+      "loss": 1.2168,
+      "step": 186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.907563025210084e-05,
+      "loss": 1.0521,
+      "step": 187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 1.4091,
+      "step": 188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.949579831932773e-05,
+      "loss": 1.2033,
+      "step": 189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.970588235294117e-05,
+      "loss": 1.233,
+      "step": 190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.991596638655463e-05,
+      "loss": 1.3121,
+      "step": 191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.012605042016807e-05,
+      "loss": 1.1555,
+      "step": 192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.033613445378152e-05,
+      "loss": 1.281,
+      "step": 193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.054621848739496e-05,
+      "loss": 1.151,
+      "step": 194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.075630252100841e-05,
+      "loss": 1.2137,
+      "step": 195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.096638655462185e-05,
+      "loss": 1.2445,
+      "step": 196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 1.3027,
+      "step": 197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.138655462184874e-05,
+      "loss": 1.2125,
+      "step": 198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.159663865546219e-05,
+      "loss": 1.2618,
+      "step": 199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1806722689075635e-05,
+      "loss": 1.0976,
+      "step": 200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.201680672268908e-05,
+      "loss": 1.1819,
+      "step": 201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.2226890756302525e-05,
+      "loss": 1.3087,
+      "step": 202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.2436974789915967e-05,
+      "loss": 1.2341,
+      "step": 203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.2647058823529415e-05,
+      "loss": 1.2167,
+      "step": 204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 1.1564,
+      "step": 205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.3067226890756305e-05,
+      "loss": 1.2352,
+      "step": 206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.327731092436975e-05,
+      "loss": 1.1795,
+      "step": 207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.3487394957983194e-05,
+      "loss": 1.1662,
+      "step": 208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.369747899159664e-05,
+      "loss": 1.2107,
+      "step": 209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.3907563025210084e-05,
+      "loss": 1.3208,
+      "step": 210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 1.3201,
+      "step": 211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4327731092436974e-05,
+      "loss": 1.0391,
+      "step": 212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.453781512605042e-05,
+      "loss": 1.0112,
+      "step": 213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.474789915966387e-05,
+      "loss": 1.1751,
+      "step": 214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.495798319327731e-05,
+      "loss": 1.2475,
+      "step": 215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.516806722689076e-05,
+      "loss": 1.2615,
+      "step": 216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.53781512605042e-05,
+      "loss": 1.194,
+      "step": 217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.558823529411765e-05,
+      "loss": 1.2252,
+      "step": 218
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.579831932773109e-05,
+      "loss": 1.3252,
+      "step": 219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.600840336134454e-05,
+      "loss": 1.0741,
+      "step": 220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.621848739495799e-05,
+      "loss": 1.1994,
+      "step": 221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 1.3506,
+      "step": 222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.663865546218488e-05,
+      "loss": 1.2557,
+      "step": 223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.684873949579832e-05,
+      "loss": 1.2406,
+      "step": 224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 1.3019,
+      "step": 225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.726890756302521e-05,
+      "loss": 1.2124,
+      "step": 226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.747899159663866e-05,
+      "loss": 1.2196,
+      "step": 227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7689075630252105e-05,
+      "loss": 1.1524,
+      "step": 228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7899159663865554e-05,
+      "loss": 1.1456,
+      "step": 229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8109243697478995e-05,
+      "loss": 1.22,
+      "step": 230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.831932773109244e-05,
+      "loss": 1.2457,
+      "step": 231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8529411764705885e-05,
+      "loss": 1.0593,
+      "step": 232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8529411764705885e-05,
+      "loss": 1.3126,
+      "step": 233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8739495798319326e-05,
+      "loss": 1.2346,
+      "step": 234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8949579831932775e-05,
+      "loss": 1.339,
+      "step": 235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9159663865546216e-05,
+      "loss": 1.1249,
+      "step": 236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.936974789915967e-05,
+      "loss": 1.2785,
+      "step": 237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.957983193277311e-05,
+      "loss": 1.1619,
+      "step": 238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.978991596638656e-05,
+      "loss": 1.0591,
+      "step": 239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5e-05,
+      "loss": 1.19,
+      "step": 240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.9999997909447174e-05,
+      "loss": 1.3782,
+      "step": 241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.9999991637789034e-05,
+      "loss": 1.146,
+      "step": 242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.9999981185026646e-05,
+      "loss": 1.2539,
+      "step": 243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.999996655116174e-05,
+      "loss": 1.2716,
+      "step": 244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.999994773619677e-05,
+      "loss": 1.2958,
+      "step": 245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.999992474013489e-05,
+      "loss": 1.277,
+      "step": 246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9999897562979934e-05,
+      "loss": 1.1097,
+      "step": 247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.999986620473646e-05,
+      "loss": 1.1541,
+      "step": 248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.9999830665409694e-05,
+      "loss": 1.3017,
+      "step": 249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.9999790945005604e-05,
+      "loss": 1.3413,
+      "step": 250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.9999747043530816e-05,
+      "loss": 1.1783,
+      "step": 251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9999698960992675e-05,
+      "loss": 1.278,
+      "step": 252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9999646697399225e-05,
+      "loss": 1.295,
+      "step": 253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9999590252759206e-05,
+      "loss": 1.0626,
+      "step": 254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.999952962708205e-05,
+      "loss": 1.0146,
+      "step": 255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9999464820377914e-05,
+      "loss": 1.2714,
+      "step": 256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9999395832657627e-05,
+      "loss": 1.2314,
+      "step": 257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.999932266393272e-05,
+      "loss": 1.128,
+      "step": 258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.999924531421545e-05,
+      "loss": 1.2181,
+      "step": 259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.999916378351872e-05,
+      "loss": 1.1783,
+      "step": 260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.9999078071856204e-05,
+      "loss": 1.0623,
+      "step": 261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.9998988179242215e-05,
+      "loss": 1.1929,
+      "step": 262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.9998894105691785e-05,
+      "loss": 1.2603,
+      "step": 263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.999879585122065e-05,
+      "loss": 1.1657,
+      "step": 264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.999869341584524e-05,
+      "loss": 1.1949,
+      "step": 265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.999858679958271e-05,
+      "loss": 1.3015,
+      "step": 266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.999847600245085e-05,
+      "loss": 1.126,
+      "step": 267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.999836102446823e-05,
+      "loss": 1.2356,
+      "step": 268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.999824186565406e-05,
+      "loss": 1.13,
+      "step": 269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.999811852602827e-05,
+      "loss": 1.2068,
+      "step": 270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.999799100561149e-05,
+      "loss": 1.2476,
+      "step": 271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.999785930442504e-05,
+      "loss": 1.1065,
+      "step": 272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.999772342249096e-05,
+      "loss": 1.1645,
+      "step": 273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.999758335983196e-05,
+      "loss": 1.0554,
+      "step": 274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.999743911647148e-05,
+      "loss": 1.1183,
+      "step": 275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.999729069243363e-05,
+      "loss": 1.159,
+      "step": 276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.999713808774324e-05,
+      "loss": 1.1213,
+      "step": 277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9996981302425827e-05,
+      "loss": 1.1783,
+      "step": 278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.999682033650762e-05,
+      "loss": 1.2467,
+      "step": 279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.999665519001554e-05,
+      "loss": 1.1973,
+      "step": 280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9996485862977205e-05,
+      "loss": 1.0827,
+      "step": 281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.999631235542093e-05,
+      "loss": 1.2515,
+      "step": 282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.999613466737573e-05,
+      "loss": 1.1232,
+      "step": 283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9995952798871334e-05,
+      "loss": 1.2791,
+      "step": 284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9995766749938144e-05,
+      "loss": 1.235,
+      "step": 285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.999557652060729e-05,
+      "loss": 1.1967,
+      "step": 286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9995382110910585e-05,
+      "loss": 1.1661,
+      "step": 287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999518352088053e-05,
+      "loss": 1.2766,
+      "step": 288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9994980750550355e-05,
+      "loss": 1.1923,
+      "step": 289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999477379995395e-05,
+      "loss": 1.1607,
+      "step": 290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9994562669125955e-05,
+      "loss": 1.0825,
+      "step": 291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9994347358101656e-05,
+      "loss": 1.2252,
+      "step": 292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9994127866917076e-05,
+      "loss": 1.2264,
+      "step": 293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999390419560892e-05,
+      "loss": 1.1464,
+      "step": 294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999367634421459e-05,
+      "loss": 1.0376,
+      "step": 295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9993444312772196e-05,
+      "loss": 0.9929,
+      "step": 296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.999320810132055e-05,
+      "loss": 1.3676,
+      "step": 297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.999296770989916e-05,
+      "loss": 1.1376,
+      "step": 298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.999272313854821e-05,
+      "loss": 1.1817,
+      "step": 299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.999247438730863e-05,
+      "loss": 1.2443,
+      "step": 300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9992221456222e-05,
+      "loss": 1.1172,
+      "step": 301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.999196434533063e-05,
+      "loss": 1.24,
+      "step": 302
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.999170305467752e-05,
+      "loss": 1.1454,
+      "step": 303
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9991437584306365e-05,
+      "loss": 0.9806,
+      "step": 304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.999116793426157e-05,
+      "loss": 1.3199,
+      "step": 305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.999089410458824e-05,
+      "loss": 1.2885,
+      "step": 306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9990616095332155e-05,
+      "loss": 1.2345,
+      "step": 307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.999033390653981e-05,
+      "loss": 1.4096,
+      "step": 308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.999004753825842e-05,
+      "loss": 1.2623,
+      "step": 309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.998975699053585e-05,
+      "loss": 1.1542,
+      "step": 310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.998946226342072e-05,
+      "loss": 1.193,
+      "step": 311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.99891633569623e-05,
+      "loss": 1.1933,
+      "step": 312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9988860271210594e-05,
+      "loss": 1.2694,
+      "step": 313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.998855300621629e-05,
+      "loss": 1.33,
+      "step": 314
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9988241562030764e-05,
+      "loss": 1.0904,
+      "step": 315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9987925938706115e-05,
+      "loss": 1.1461,
+      "step": 316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.998760613629513e-05,
+      "loss": 1.1735,
+      "step": 317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.998728215485128e-05,
+      "loss": 1.2437,
+      "step": 318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.998695399442877e-05,
+      "loss": 1.1993,
+      "step": 319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9986621655082465e-05,
+      "loss": 1.2393,
+      "step": 320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.998628513686796e-05,
+      "loss": 1.2163,
+      "step": 321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.998594443984153e-05,
+      "loss": 1.2464,
+      "step": 322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9985599564060146e-05,
+      "loss": 1.1305,
+      "step": 323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9985250509581506e-05,
+      "loss": 1.0459,
+      "step": 324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.998489727646397e-05,
+      "loss": 0.9089,
+      "step": 325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.998453986476662e-05,
+      "loss": 1.2582,
+      "step": 326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.998417827454923e-05,
+      "loss": 1.1716,
+      "step": 327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.998381250587228e-05,
+      "loss": 1.4073,
+      "step": 328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9983442558796934e-05,
+      "loss": 1.3321,
+      "step": 329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.998306843338507e-05,
+      "loss": 1.278,
+      "step": 330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9982690129699255e-05,
+      "loss": 1.2617,
+      "step": 331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9982307647802765e-05,
+      "loss": 1.2956,
+      "step": 332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.998192098775955e-05,
+      "loss": 1.1725,
+      "step": 333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.998153014963429e-05,
+      "loss": 1.1861,
+      "step": 334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.998113513349236e-05,
+      "loss": 0.9948,
+      "step": 335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9980735939399806e-05,
+      "loss": 1.1099,
+      "step": 336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.99803325674234e-05,
+      "loss": 1.2673,
+      "step": 337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9979925017630605e-05,
+      "loss": 1.1724,
+      "step": 338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.997951329008956e-05,
+      "loss": 1.2176,
+      "step": 339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.997909738486916e-05,
+      "loss": 1.1009,
+      "step": 340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9978677302038946e-05,
+      "loss": 1.3729,
+      "step": 341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.997825304166916e-05,
+      "loss": 1.1886,
+      "step": 342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.997782460383078e-05,
+      "loss": 1.1904,
+      "step": 343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9977391988595445e-05,
+      "loss": 1.1808,
+      "step": 344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.997695519603552e-05,
+      "loss": 1.2683,
+      "step": 345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.997651422622404e-05,
+      "loss": 1.2131,
+      "step": 346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.997606907923477e-05,
+      "loss": 1.2026,
+      "step": 347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9975619755142146e-05,
+      "loss": 1.2162,
+      "step": 348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.997516625402132e-05,
+      "loss": 1.1816,
+      "step": 349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.997470857594814e-05,
+      "loss": 1.2854,
+      "step": 350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9974246720999153e-05,
+      "loss": 1.1657,
+      "step": 351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9973780689251585e-05,
+      "loss": 1.2289,
+      "step": 352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9973310480783394e-05,
+      "loss": 1.2468,
+      "step": 353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9972836095673214e-05,
+      "loss": 1.1699,
+      "step": 354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9972357534000394e-05,
+      "loss": 1.1558,
+      "step": 355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.997187479584494e-05,
+      "loss": 1.2368,
+      "step": 356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.997138788128762e-05,
+      "loss": 1.2147,
+      "step": 357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.997089679040985e-05,
+      "loss": 1.2255,
+      "step": 358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.997040152329378e-05,
+      "loss": 1.3019,
+      "step": 359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.996990208002221e-05,
+      "loss": 1.1046,
+      "step": 360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.996939846067869e-05,
+      "loss": 1.1403,
+      "step": 361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9968890665347444e-05,
+      "loss": 1.1529,
+      "step": 362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.996837869411341e-05,
+      "loss": 1.1768,
+      "step": 363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.996786254706218e-05,
+      "loss": 1.2436,
+      "step": 364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9967342224280115e-05,
+      "loss": 1.2515,
+      "step": 365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9966817725854195e-05,
+      "loss": 1.083,
+      "step": 366
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.996628905187218e-05,
+      "loss": 1.1629,
+      "step": 367
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9965756202422465e-05,
+      "loss": 1.1602,
+      "step": 368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.996521917759417e-05,
+      "loss": 1.2093,
+      "step": 369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9964677977477115e-05,
+      "loss": 1.1995,
+      "step": 370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.99641326021618e-05,
+      "loss": 1.1195,
+      "step": 371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9963583051739446e-05,
+      "loss": 1.1906,
+      "step": 372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9963029326301955e-05,
+      "loss": 1.1793,
+      "step": 373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.996247142594195e-05,
+      "loss": 1.0121,
+      "step": 374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.996190935075272e-05,
+      "loss": 1.2108,
+      "step": 375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.996134310082827e-05,
+      "loss": 1.2171,
+      "step": 376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9960772676263314e-05,
+      "loss": 1.1004,
+      "step": 377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.996019807715324e-05,
+      "loss": 1.1869,
+      "step": 378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.995961930359415e-05,
+      "loss": 1.1952,
+      "step": 379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.995903635568284e-05,
+      "loss": 1.1479,
+      "step": 380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.995844923351681e-05,
+      "loss": 1.1212,
+      "step": 381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9957857937194245e-05,
+      "loss": 1.2436,
+      "step": 382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.995726246681404e-05,
+      "loss": 1.1467,
+      "step": 383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.995666282247579e-05,
+      "loss": 1.3118,
+      "step": 384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.995605900427977e-05,
+      "loss": 1.1185,
+      "step": 385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.995545101232697e-05,
+      "loss": 1.2543,
+      "step": 386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.995483884671908e-05,
+      "loss": 1.0646,
+      "step": 387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.995422250755847e-05,
+      "loss": 1.0976,
+      "step": 388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.995360199494823e-05,
+      "loss": 1.1474,
+      "step": 389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.995297730899212e-05,
+      "loss": 1.1998,
+      "step": 390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.995234844979464e-05,
+      "loss": 1.2406,
+      "step": 391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.995171541746094e-05,
+      "loss": 1.3599,
+      "step": 392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.995107821209691e-05,
+      "loss": 1.01,
+      "step": 393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.99504368338091e-05,
+      "loss": 1.247,
+      "step": 394
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.994979128270479e-05,
+      "loss": 1.156,
+      "step": 395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.994914155889194e-05,
+      "loss": 1.2575,
+      "step": 396
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.994848766247923e-05,
+      "loss": 1.1445,
+      "step": 397
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.994782959357598e-05,
+      "loss": 1.09,
+      "step": 398
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.9947167352292283e-05,
+      "loss": 1.1936,
+      "step": 399
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.994650093873888e-05,
+      "loss": 1.0999,
+      "step": 400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.994583035302723e-05,
+      "loss": 1.173,
+      "step": 401
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.9945155595269494e-05,
+      "loss": 1.2298,
+      "step": 402
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.9944476665578504e-05,
+      "loss": 1.1961,
+      "step": 403
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.994379356406782e-05,
+      "loss": 1.105,
+      "step": 404
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.994310629085167e-05,
+      "loss": 1.1473,
+      "step": 405
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.994241484604502e-05,
+      "loss": 1.2055,
+      "step": 406
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.994171922976348e-05,
+      "loss": 1.1926,
+      "step": 407
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.9941019442123416e-05,
+      "loss": 1.1225,
+      "step": 408
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.994031548324185e-05,
+      "loss": 1.1808,
+      "step": 409
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.993960735323653e-05,
+      "loss": 1.1413,
+      "step": 410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.993889505222586e-05,
+      "loss": 1.1223,
+      "step": 411
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.9938178580328986e-05,
+      "loss": 1.206,
+      "step": 412
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.9937457937665733e-05,
+      "loss": 1.2238,
+      "step": 413
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.993673312435662e-05,
+      "loss": 1.3092,
+      "step": 414
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.9936004140522876e-05,
+      "loss": 1.1419,
+      "step": 415
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.9935270986286405e-05,
+      "loss": 1.2037,
+      "step": 416
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.993453366176983e-05,
+      "loss": 1.1804,
+      "step": 417
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.993379216709647e-05,
+      "loss": 1.183,
+      "step": 418
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.9933046502390335e-05,
+      "loss": 1.1569,
+      "step": 419
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.993229666777612e-05,
+      "loss": 1.1375,
+      "step": 420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.9931542663379246e-05,
+      "loss": 1.0957,
+      "step": 421
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.993078448932581e-05,
+      "loss": 1.2084,
+      "step": 422
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.993002214574261e-05,
+      "loss": 1.0958,
+      "step": 423
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.992925563275714e-05,
+      "loss": 1.0741,
+      "step": 424
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.99284849504976e-05,
+      "loss": 1.3485,
+      "step": 425
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.992771009909288e-05,
+      "loss": 1.3158,
+      "step": 426
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.9926931078672586e-05,
+      "loss": 1.2855,
+      "step": 427
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.992614788936698e-05,
+      "loss": 1.13,
+      "step": 428
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.992536053130706e-05,
+      "loss": 1.2445,
+      "step": 429
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.9924569004624497e-05,
+      "loss": 1.0846,
+      "step": 430
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.992377330945168e-05,
+      "loss": 1.3384,
+      "step": 431
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.992297344592168e-05,
+      "loss": 1.138,
+      "step": 432
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.9922169414168276e-05,
+      "loss": 1.1689,
+      "step": 433
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.992136121432592e-05,
+      "loss": 1.2103,
+      "step": 434
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.992054884652979e-05,
+      "loss": 1.1235,
+      "step": 435
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.991973231091576e-05,
+      "loss": 1.2724,
+      "step": 436
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.991891160762037e-05,
+      "loss": 1.2344,
+      "step": 437
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.9918086736780903e-05,
+      "loss": 1.066,
+      "step": 438
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.991725769853528e-05,
+      "loss": 1.0976,
+      "step": 439
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.991642449302218e-05,
+      "loss": 1.2212,
+      "step": 440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.991558712038095e-05,
+      "loss": 1.0925,
+      "step": 441
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.991474558075163e-05,
+      "loss": 1.1371,
+      "step": 442
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.991389987427496e-05,
+      "loss": 1.2096,
+      "step": 443
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.991305000109238e-05,
+      "loss": 1.0509,
+      "step": 444
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.9912195961346026e-05,
+      "loss": 1.1367,
+      "step": 445
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.9911337755178735e-05,
+      "loss": 1.0513,
+      "step": 446
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.9910475382734034e-05,
+      "loss": 0.9934,
+      "step": 447
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.9909608844156155e-05,
+      "loss": 1.3165,
+      "step": 448
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.9908738139590016e-05,
+      "loss": 1.1908,
+      "step": 449
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9907863269181246e-05,
+      "loss": 1.0347,
+      "step": 450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.990698423307615e-05,
+      "loss": 1.0915,
+      "step": 451
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.990610103142175e-05,
+      "loss": 1.1217,
+      "step": 452
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9905213664365745e-05,
+      "loss": 1.249,
+      "step": 453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.990432213205656e-05,
+      "loss": 1.2384,
+      "step": 454
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.9903426434643287e-05,
+      "loss": 1.0959,
+      "step": 455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.9902526572275723e-05,
+      "loss": 1.1325,
+      "step": 456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.990162254510438e-05,
+      "loss": 1.2229,
+      "step": 457
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.9900714353280434e-05,
+      "loss": 1.172,
+      "step": 458
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.9899801996955775e-05,
+      "loss": 1.0692,
+      "step": 459
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.989888547628301e-05,
+      "loss": 1.0313,
+      "step": 460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.9897964791415406e-05,
+      "loss": 1.313,
+      "step": 461
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.989703994250694e-05,
+      "loss": 1.2509,
+      "step": 462
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.989611092971229e-05,
+      "loss": 1.1865,
+      "step": 463
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.989517775318684e-05,
+      "loss": 1.168,
+      "step": 464
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.989424041308664e-05,
+      "loss": 1.3127,
+      "step": 465
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.989329890956846e-05,
+      "loss": 1.1705,
+      "step": 466
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.989235324278978e-05,
+      "loss": 1.162,
+      "step": 467
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.989140341290873e-05,
+      "loss": 1.2591,
+      "step": 468
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.9890449420084174e-05,
+      "loss": 1.397,
+      "step": 469
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.988949126447567e-05,
+      "loss": 0.9481,
+      "step": 470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.988852894624345e-05,
+      "loss": 1.2532,
+      "step": 471
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.9887562465548474e-05,
+      "loss": 1.2063,
+      "step": 472
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.988659182255236e-05,
+      "loss": 1.0449,
+      "step": 473
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.988561701741745e-05,
+      "loss": 1.0282,
+      "step": 474
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.988463805030679e-05,
+      "loss": 1.2576,
+      "step": 475
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.988365492138408e-05,
+      "loss": 1.2225,
+      "step": 476
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.988266763081376e-05,
+      "loss": 1.2761,
+      "step": 477
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.9881676178760945e-05,
+      "loss": 1.2446,
+      "step": 478
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.988068056539145e-05,
+      "loss": 1.2159,
+      "step": 479
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.987968079087178e-05,
+      "loss": 1.233,
+      "step": 480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.987867685536915e-05,
+      "loss": 1.1958,
+      "step": 481
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.9877668759051456e-05,
+      "loss": 1.1898,
+      "step": 482
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.987665650208731e-05,
+      "loss": 1.2647,
+      "step": 483
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.987564008464598e-05,
+      "loss": 1.1884,
+      "step": 484
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.987461950689747e-05,
+      "loss": 0.9946,
+      "step": 485
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.987359476901248e-05,
+      "loss": 1.3072,
+      "step": 486
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.9872565871162366e-05,
+      "loss": 1.2988,
+      "step": 487
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.987153281351922e-05,
+      "loss": 1.084,
+      "step": 488
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.9870495596255815e-05,
+      "loss": 1.1946,
+      "step": 489
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.986945421954562e-05,
+      "loss": 1.2764,
+      "step": 490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.986840868356279e-05,
+      "loss": 1.2836,
+      "step": 491
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.986735898848219e-05,
+      "loss": 1.0863,
+      "step": 492
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.986630513447938e-05,
+      "loss": 1.1874,
+      "step": 493
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.9865247121730615e-05,
+      "loss": 1.3916,
+      "step": 494
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.986418495041283e-05,
+      "loss": 1.2488,
+      "step": 495
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.9863118620703666e-05,
+      "loss": 1.212,
+      "step": 496
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.986204813278147e-05,
+      "loss": 1.1705,
+      "step": 497
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.986097348682526e-05,
+      "loss": 1.0022,
+      "step": 498
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.9859894683014785e-05,
+      "loss": 1.0085,
+      "step": 499
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.9858811721530455e-05,
+      "loss": 1.262,
+      "step": 500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.985772460255339e-05,
+      "loss": 1.1786,
+      "step": 501
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.985663332626541e-05,
+      "loss": 1.1715,
+      "step": 502
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9855537892849026e-05,
+      "loss": 1.1792,
+      "step": 503
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.985443830248743e-05,
+      "loss": 1.0383,
+      "step": 504
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.985333455536454e-05,
+      "loss": 1.2687,
+      "step": 505
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9852226651664935e-05,
+      "loss": 1.0465,
+      "step": 506
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.985111459157391e-05,
+      "loss": 1.0624,
+      "step": 507
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.984999837527746e-05,
+      "loss": 1.1071,
+      "step": 508
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.984887800296225e-05,
+      "loss": 1.1253,
+      "step": 509
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9847753474815674e-05,
+      "loss": 1.1606,
+      "step": 510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.984662479102579e-05,
+      "loss": 1.2434,
+      "step": 511
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.984549195178136e-05,
+      "loss": 1.1374,
+      "step": 512
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.984435495727187e-05,
+      "loss": 1.136,
+      "step": 513
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9843213807687447e-05,
+      "loss": 1.0858,
+      "step": 514
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9842068503218954e-05,
+      "loss": 1.0224,
+      "step": 515
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.984091904405793e-05,
+      "loss": 1.0993,
+      "step": 516
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.9839765430396625e-05,
+      "loss": 1.1984,
+      "step": 517
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.9838607662427964e-05,
+      "loss": 1.2088,
+      "step": 518
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.983744574034559e-05,
+      "loss": 1.0501,
+      "step": 519
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.9836279664343816e-05,
+      "loss": 1.3173,
+      "step": 520
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.9835109434617665e-05,
+      "loss": 1.0444,
+      "step": 521
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.983393505136286e-05,
+      "loss": 1.3217,
+      "step": 522
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.983275651477579e-05,
+      "loss": 1.2865,
+      "step": 523
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.983157382505358e-05,
+      "loss": 1.1688,
+      "step": 524
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.9830386982394015e-05,
+      "loss": 1.2672,
+      "step": 525
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.9829195986995594e-05,
+      "loss": 0.9723,
+      "step": 526
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.98280008390575e-05,
+      "loss": 1.2553,
+      "step": 527
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.982680153877961e-05,
+      "loss": 1.068,
+      "step": 528
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.982559808636251e-05,
+      "loss": 1.1099,
+      "step": 529
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.982439048200746e-05,
+      "loss": 1.2435,
+      "step": 530
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.982317872591644e-05,
+      "loss": 1.1212,
+      "step": 531
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.98219628182921e-05,
+      "loss": 1.0007,
+      "step": 532
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9820742759337794e-05,
+      "loss": 1.1405,
+      "step": 533
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.981951854925757e-05,
+      "loss": 1.2947,
+      "step": 534
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.981829018825616e-05,
+      "loss": 1.1558,
+      "step": 535
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9817057676539015e-05,
+      "loss": 1.1706,
+      "step": 536
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9815821014312264e-05,
+      "loss": 1.1981,
+      "step": 537
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9814580201782724e-05,
+      "loss": 1.1724,
+      "step": 538
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.981333523915792e-05,
+      "loss": 1.3096,
+      "step": 539
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.981208612664606e-05,
+      "loss": 1.1536,
+      "step": 540
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.981083286445606e-05,
+      "loss": 1.2152,
+      "step": 541
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.9809575452797505e-05,
+      "loss": 1.1563,
+      "step": 542
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.980831389188071e-05,
+      "loss": 1.0553,
+      "step": 543
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.980704818191665e-05,
+      "loss": 1.2502,
+      "step": 544
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.9805778323117004e-05,
+      "loss": 1.0682,
+      "step": 545
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.980450431569416e-05,
+      "loss": 1.2311,
+      "step": 546
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.980322615986118e-05,
+      "loss": 1.1377,
+      "step": 547
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.980194385583184e-05,
+      "loss": 1.3399,
+      "step": 548
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.980065740382058e-05,
+      "loss": 1.0707,
+      "step": 549
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.9799366804042566e-05,
+      "loss": 1.2149,
+      "step": 550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.9798072056713643e-05,
+      "loss": 1.2656,
+      "step": 551
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.9796773162050335e-05,
+      "loss": 1.2011,
+      "step": 552
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.9795470120269896e-05,
+      "loss": 1.1722,
+      "step": 553
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.979416293159023e-05,
+      "loss": 1.2014,
+      "step": 554
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.979285159622997e-05,
+      "loss": 1.2816,
+      "step": 555
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.979153611440844e-05,
+      "loss": 1.1916,
+      "step": 556
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9790216486345624e-05,
+      "loss": 1.1783,
+      "step": 557
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.978889271226224e-05,
+      "loss": 1.1454,
+      "step": 558
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.978756479237966e-05,
+      "loss": 1.2203,
+      "step": 559
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9786232726920005e-05,
+      "loss": 1.2149,
+      "step": 560
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.978489651610602e-05,
+      "loss": 1.3803,
+      "step": 561
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.97835561601612e-05,
+      "loss": 1.079,
+      "step": 562
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.97822116593097e-05,
+      "loss": 1.0008,
+      "step": 563
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.978086301377639e-05,
+      "loss": 1.201,
+      "step": 564
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.977951022378682e-05,
+      "loss": 1.0017,
+      "step": 565
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.9778153289567234e-05,
+      "loss": 1.1603,
+      "step": 566
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.977679221134457e-05,
+      "loss": 1.0954,
+      "step": 567
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.977542698934647e-05,
+      "loss": 1.1289,
+      "step": 568
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.977405762380125e-05,
+      "loss": 1.0652,
+      "step": 569
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.977268411493793e-05,
+      "loss": 1.2368,
+      "step": 570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.9771306462986215e-05,
+      "loss": 1.0881,
+      "step": 571
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.9769924668176524e-05,
+      "loss": 1.2216,
+      "step": 572
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.9768538730739945e-05,
+      "loss": 1.0808,
+      "step": 573
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.976714865090827e-05,
+      "loss": 1.2347,
+      "step": 574
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.976575442891398e-05,
+      "loss": 1.2651,
+      "step": 575
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.976435606499026e-05,
+      "loss": 1.1457,
+      "step": 576
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.976295355937096e-05,
+      "loss": 1.1046,
+      "step": 577
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.976154691229066e-05,
+      "loss": 1.1218,
+      "step": 578
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.97601361239846e-05,
+      "loss": 1.2622,
+      "step": 579
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.975872119468874e-05,
+      "loss": 1.2036,
+      "step": 580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.97573021246397e-05,
+      "loss": 1.0256,
+      "step": 581
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.975587891407483e-05,
+      "loss": 1.1636,
+      "step": 582
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.975445156323214e-05,
+      "loss": 1.0066,
+      "step": 583
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.9753020072350353e-05,
+      "loss": 1.3734,
+      "step": 584
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9751584441668874e-05,
+      "loss": 1.1601,
+      "step": 585
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.975014467142781e-05,
+      "loss": 1.2013,
+      "step": 586
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9748700761867945e-05,
+      "loss": 1.2677,
+      "step": 587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.974725271323078e-05,
+      "loss": 1.1123,
+      "step": 588
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.974580052575847e-05,
+      "loss": 1.2322,
+      "step": 589
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.97443441996939e-05,
+      "loss": 1.099,
+      "step": 590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.974288373528063e-05,
+      "loss": 1.1579,
+      "step": 591
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.974141913276291e-05,
+      "loss": 1.1819,
+      "step": 592
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.973995039238571e-05,
+      "loss": 1.2456,
+      "step": 593
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.973847751439462e-05,
+      "loss": 1.1653,
+      "step": 594
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9737000499036015e-05,
+      "loss": 1.2057,
+      "step": 595
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.97355193465569e-05,
+      "loss": 1.0338,
+      "step": 596
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9734034057204995e-05,
+      "loss": 1.1191,
+      "step": 597
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.97325446312287e-05,
+      "loss": 1.2552,
+      "step": 598
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.97310510688771e-05,
+      "loss": 1.1865,
+      "step": 599
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.972955337040001e-05,
+      "loss": 1.1677,
+      "step": 600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.9728051536047906e-05,
+      "loss": 1.1167,
+      "step": 601
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.9726545566071945e-05,
+      "loss": 1.1622,
+      "step": 602
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.972503546072401e-05,
+      "loss": 1.1254,
+      "step": 603
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9723521220256644e-05,
+      "loss": 1.1174,
+      "step": 604
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.97220028449231e-05,
+      "loss": 1.1586,
+      "step": 605
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9720480334977324e-05,
+      "loss": 1.2658,
+      "step": 606
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.971895369067394e-05,
+      "loss": 1.2766,
+      "step": 607
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.971742291226827e-05,
+      "loss": 0.9814,
+      "step": 608
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.971588800001633e-05,
+      "loss": 0.9614,
+      "step": 609
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.971434895417482e-05,
+      "loss": 1.1165,
+      "step": 610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.971280577500115e-05,
+      "loss": 1.1917,
+      "step": 611
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.971125846275339e-05,
+      "loss": 1.2108,
+      "step": 612
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.970970701769033e-05,
+      "loss": 1.1487,
+      "step": 613
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.970815144007144e-05,
+      "loss": 1.1687,
+      "step": 614
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.970659173015688e-05,
+      "loss": 1.2519,
+      "step": 615
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.97050278882075e-05,
+      "loss": 1.0176,
+      "step": 616
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.9703459914484855e-05,
+      "loss": 1.146,
+      "step": 617
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.970188780925116e-05,
+      "loss": 1.2864,
+      "step": 618
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.970031157276936e-05,
+      "loss": 1.2156,
+      "step": 619
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.969873120530306e-05,
+      "loss": 1.1895,
+      "step": 620
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.969714670711656e-05,
+      "loss": 1.2414,
+      "step": 621
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.9695558078474876e-05,
+      "loss": 1.1616,
+      "step": 622
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.9693965319643685e-05,
+      "loss": 1.1671,
+      "step": 623
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.969236843088937e-05,
+      "loss": 1.107,
+      "step": 624
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.9690767412479e-05,
+      "loss": 1.0805,
+      "step": 625
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.968916226468034e-05,
+      "loss": 1.1713,
+      "step": 626
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.968755298776185e-05,
+      "loss": 1.1827,
+      "step": 627
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.968593958199265e-05,
+      "loss": 0.988,
+      "step": 628
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.968432204764259e-05,
+      "loss": 1.2782,
+      "step": 629
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.9682700384982185e-05,
+      "loss": 1.1851,
+      "step": 630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.968107459428265e-05,
+      "loss": 1.2875,
+      "step": 631
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.96794446758159e-05,
+      "loss": 1.0611,
+      "step": 632
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.9677810629854514e-05,
+      "loss": 1.2246,
+      "step": 633
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.9676172456671776e-05,
+      "loss": 1.1152,
+      "step": 634
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.967453015654168e-05,
+      "loss": 1.0072,
+      "step": 635
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.967288372973887e-05,
+      "loss": 1.1482,
+      "step": 636
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.967123317653872e-05,
+      "loss": 1.3152,
+      "step": 637
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.966957849721726e-05,
+      "loss": 1.0824,
+      "step": 638
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.966791969205123e-05,
+      "loss": 1.2001,
+      "step": 639
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.966625676131807e-05,
+      "loss": 1.2081,
+      "step": 640
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.9664589705295863e-05,
+      "loss": 1.2483,
+      "step": 641
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.9662918524263444e-05,
+      "loss": 1.2097,
+      "step": 642
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.9661243218500296e-05,
+      "loss": 1.0388,
+      "step": 643
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.965956378828661e-05,
+      "loss": 1.108,
+      "step": 644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.965788023390326e-05,
+      "loss": 1.259,
+      "step": 645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.9656192555631806e-05,
+      "loss": 1.2804,
+      "step": 646
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.9654500753754505e-05,
+      "loss": 1.1249,
+      "step": 647
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.96528048285543e-05,
+      "loss": 1.2232,
+      "step": 648
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.9651104780314826e-05,
+      "loss": 1.2237,
+      "step": 649
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.964940060932041e-05,
+      "loss": 1.0149,
+      "step": 650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.964769231585605e-05,
+      "loss": 0.9649,
+      "step": 651
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9645979900207475e-05,
+      "loss": 1.2087,
+      "step": 652
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.964426336266105e-05,
+      "loss": 1.1764,
+      "step": 653
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.964254270350387e-05,
+      "loss": 1.064,
+      "step": 654
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.964081792302371e-05,
+      "loss": 1.1668,
+      "step": 655
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.963908902150901e-05,
+      "loss": 1.1251,
+      "step": 656
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.963735599924893e-05,
+      "loss": 1.0068,
+      "step": 657
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9635618856533315e-05,
+      "loss": 1.1454,
+      "step": 658
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.963387759365269e-05,
+      "loss": 1.2251,
+      "step": 659
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.963213221089825e-05,
+      "loss": 1.1114,
+      "step": 660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.963038270856193e-05,
+      "loss": 1.1427,
+      "step": 661
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9628629086936304e-05,
+      "loss": 1.2456,
+      "step": 662
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.962687134631467e-05,
+      "loss": 1.0733,
+      "step": 663
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.9625109486990986e-05,
+      "loss": 1.1886,
+      "step": 664
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.9623343509259926e-05,
+      "loss": 1.0791,
+      "step": 665
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.9621573413416824e-05,
+      "loss": 1.1453,
+      "step": 666
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.9619799199757724e-05,
+      "loss": 1.1931,
+      "step": 667
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.9618020868579364e-05,
+      "loss": 1.0641,
+      "step": 668
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.961623842017915e-05,
+      "loss": 1.0945,
+      "step": 669
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.9614451854855185e-05,
+      "loss": 0.9968,
+      "step": 670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.9612661172906265e-05,
+      "loss": 1.0686,
+      "step": 671
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.961086637463187e-05,
+      "loss": 1.0965,
+      "step": 672
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.960906746033217e-05,
+      "loss": 1.0711,
+      "step": 673
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.9607264430308035e-05,
+      "loss": 1.1276,
+      "step": 674
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.960545728486098e-05,
+      "loss": 1.1983,
+      "step": 675
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.960364602429327e-05,
+      "loss": 1.1502,
+      "step": 676
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.960183064890782e-05,
+      "loss": 1.0231,
+      "step": 677
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.960001115900823e-05,
+      "loss": 1.2042,
+      "step": 678
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.959818755489881e-05,
+      "loss": 1.0803,
+      "step": 679
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.959635983688455e-05,
+      "loss": 1.2228,
+      "step": 680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.959452800527111e-05,
+      "loss": 1.1883,
+      "step": 681
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.959269206036487e-05,
+      "loss": 1.1448,
+      "step": 682
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.959085200247288e-05,
+      "loss": 1.1184,
+      "step": 683
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.958900783190286e-05,
+      "loss": 1.2268,
+      "step": 684
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.958715954896326e-05,
+      "loss": 1.1502,
+      "step": 685
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.958530715396318e-05,
+      "loss": 1.1044,
+      "step": 686
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.958345064721243e-05,
+      "loss": 1.0381,
+      "step": 687
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.9581590029021496e-05,
+      "loss": 1.1815,
+      "step": 688
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.957972529970155e-05,
+      "loss": 1.1688,
+      "step": 689
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.957785645956447e-05,
+      "loss": 1.0963,
+      "step": 690
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.957598350892281e-05,
+      "loss": 0.988,
+      "step": 691
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.9574106448089794e-05,
+      "loss": 0.9373,
+      "step": 692
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.9572225277379364e-05,
+      "loss": 1.3083,
+      "step": 693
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.957033999710613e-05,
+      "loss": 1.0873,
+      "step": 694
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.9568450607585395e-05,
+      "loss": 1.146,
+      "step": 695
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.956655710913314e-05,
+      "loss": 1.1901,
+      "step": 696
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.9564659502066045e-05,
+      "loss": 1.0595,
+      "step": 697
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.956275778670149e-05,
+      "loss": 1.1862,
+      "step": 698
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.9560851963357514e-05,
+      "loss": 1.0936,
+      "step": 699
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.955894203235284e-05,
+      "loss": 0.9087,
+      "step": 700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.955702799400692e-05,
+      "loss": 1.2826,
+      "step": 701
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.955510984863985e-05,
+      "loss": 1.2408,
+      "step": 702
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.9553187596572436e-05,
+      "loss": 1.1899,
+      "step": 703
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.955126123812616e-05,
+      "loss": 1.3626,
+      "step": 704
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.954933077362319e-05,
+      "loss": 1.2219,
+      "step": 705
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.954739620338639e-05,
+      "loss": 1.0973,
+      "step": 706
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.95454575277393e-05,
+      "loss": 1.1371,
+      "step": 707
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.9543514747006165e-05,
+      "loss": 1.1481,
+      "step": 708
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.9541567861511895e-05,
+      "loss": 1.2199,
+      "step": 709
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.9539616871582096e-05,
+      "loss": 1.2915,
+      "step": 710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.9537661777543054e-05,
+      "loss": 1.0548,
+      "step": 711
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.953570257972176e-05,
+      "loss": 1.0912,
+      "step": 712
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.953373927844587e-05,
+      "loss": 1.1085,
+      "step": 713
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.953177187404374e-05,
+      "loss": 1.1855,
+      "step": 714
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.95298003668444e-05,
+      "loss": 1.1405,
+      "step": 715
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.9527824757177574e-05,
+      "loss": 1.1921,
+      "step": 716
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.9525845045373674e-05,
+      "loss": 1.1516,
+      "step": 717
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.95238612317638e-05,
+      "loss": 1.1958,
+      "step": 718
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.952187331667974e-05,
+      "loss": 1.0845,
+      "step": 719
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.951988130045393e-05,
+      "loss": 1.0041,
+      "step": 720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.951788518341955e-05,
+      "loss": 0.8614,
+      "step": 721
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.9515884965910444e-05,
+      "loss": 1.204,
+      "step": 722
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.9513880648261114e-05,
+      "loss": 1.1207,
+      "step": 723
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.9511872230806774e-05,
+      "loss": 1.3557,
+      "step": 724
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.9509859713883345e-05,
+      "loss": 1.2821,
+      "step": 725
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.950784309782739e-05,
+      "loss": 1.2351,
+      "step": 726
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.950582238297617e-05,
+      "loss": 1.212,
+      "step": 727
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.950379756966765e-05,
+      "loss": 1.2526,
+      "step": 728
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.9501768658240464e-05,
+      "loss": 1.1068,
+      "step": 729
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.949973564903393e-05,
+      "loss": 1.1337,
+      "step": 730
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9497698542388075e-05,
+      "loss": 0.93,
+      "step": 731
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.949565733864357e-05,
+      "loss": 1.0607,
+      "step": 732
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.949361203814182e-05,
+      "loss": 1.2152,
+      "step": 733
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.949156264122487e-05,
+      "loss": 1.1335,
+      "step": 734
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.948950914823548e-05,
+      "loss": 1.1639,
+      "step": 735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9487451559517076e-05,
+      "loss": 1.0562,
+      "step": 736
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.948538987541378e-05,
+      "loss": 1.3108,
+      "step": 737
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.94833240962704e-05,
+      "loss": 1.1452,
+      "step": 738
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.948125422243243e-05,
+      "loss": 1.1316,
+      "step": 739
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.947918025424603e-05,
+      "loss": 1.1363,
+      "step": 740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.947710219205808e-05,
+      "loss": 1.218,
+      "step": 741
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9475020036216104e-05,
+      "loss": 1.1673,
+      "step": 742
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.9472933787068336e-05,
+      "loss": 1.15,
+      "step": 743
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.94708434449637e-05,
+      "loss": 1.178,
+      "step": 744
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.946874901025178e-05,
+      "loss": 1.1139,
+      "step": 745
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.946665048328287e-05,
+      "loss": 1.2404,
+      "step": 746
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.9464547864407926e-05,
+      "loss": 1.1135,
+      "step": 747
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.9462441153978604e-05,
+      "loss": 1.1654,
+      "step": 748
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.946033035234724e-05,
+      "loss": 1.1962,
+      "step": 749
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.945821545986685e-05,
+      "loss": 1.1273,
+      "step": 750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.945609647689114e-05,
+      "loss": 1.1076,
+      "step": 751
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.94539734037745e-05,
+      "loss": 1.1823,
+      "step": 752
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.945184624087199e-05,
+      "loss": 1.165,
+      "step": 753
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.944971498853938e-05,
+      "loss": 1.1729,
+      "step": 754
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.944757964713311e-05,
+      "loss": 1.2564,
+      "step": 755
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.9445440217010284e-05,
+      "loss": 1.0459,
+      "step": 756
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.9443296698528736e-05,
+      "loss": 1.082,
+      "step": 757
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.944114909204693e-05,
+      "loss": 1.1119,
+      "step": 758
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.943899739792407e-05,
+      "loss": 1.1313,
+      "step": 759
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.943684161651999e-05,
+      "loss": 1.2035,
+      "step": 760
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.9434681748195245e-05,
+      "loss": 1.2016,
+      "step": 761
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.943251779331105e-05,
+      "loss": 1.0327,
+      "step": 762
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.9430349752229334e-05,
+      "loss": 1.1174,
+      "step": 763
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.942817762531267e-05,
+      "loss": 1.1078,
+      "step": 764
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.942600141292434e-05,
+      "loss": 1.1612,
+      "step": 765
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.9423821115428294e-05,
+      "loss": 1.1462,
+      "step": 766
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.942163673318919e-05,
+      "loss": 1.0707,
+      "step": 767
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.941944826657234e-05,
+      "loss": 1.1321,
+      "step": 768
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9417255715943766e-05,
+      "loss": 1.1316,
+      "step": 769
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.941505908167016e-05,
+      "loss": 0.955,
+      "step": 770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.941285836411887e-05,
+      "loss": 1.1536,
+      "step": 771
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.941065356365798e-05,
+      "loss": 1.1677,
+      "step": 772
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.940844468065622e-05,
+      "loss": 1.0552,
+      "step": 773
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.940623171548302e-05,
+      "loss": 1.1354,
+      "step": 774
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.9404014668508483e-05,
+      "loss": 1.1489,
+      "step": 775
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.940179354010339e-05,
+      "loss": 1.0927,
+      "step": 776
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.939956833063921e-05,
+      "loss": 1.0671,
+      "step": 777
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.939733904048811e-05,
+      "loss": 1.1875,
+      "step": 778
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.939510567002292e-05,
+      "loss": 1.0972,
+      "step": 779
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.939286821961715e-05,
+      "loss": 1.2588,
+      "step": 780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.9390626689645016e-05,
+      "loss": 1.0602,
+      "step": 781
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.938838108048139e-05,
+      "loss": 1.2027,
+      "step": 782
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9386131392501837e-05,
+      "loss": 1.0082,
+      "step": 783
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.938387762608261e-05,
+      "loss": 1.0491,
+      "step": 784
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.938161978160064e-05,
+      "loss": 1.096,
+      "step": 785
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.937935785943353e-05,
+      "loss": 1.157,
+      "step": 786
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.937709185995958e-05,
+      "loss": 1.1837,
+      "step": 787
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.937482178355776e-05,
+      "loss": 1.3125,
+      "step": 788
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9372547630607726e-05,
+      "loss": 0.9611,
+      "step": 789
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.937026940148983e-05,
+      "loss": 1.1933,
+      "step": 790
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.9367987096585077e-05,
+      "loss": 1.1123,
+      "step": 791
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.936570071627518e-05,
+      "loss": 1.1874,
+      "step": 792
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.936341026094252e-05,
+      "loss": 1.0778,
+      "step": 793
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.936111573097016e-05,
+      "loss": 1.0307,
+      "step": 794
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9358817126741844e-05,
+      "loss": 1.1496,
+      "step": 795
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.935651444864201e-05,
+      "loss": 1.0531,
+      "step": 796
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.935420769705576e-05,
+      "loss": 1.1214,
+      "step": 797
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.935189687236889e-05,
+      "loss": 1.1699,
+      "step": 798
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.934958197496786e-05,
+      "loss": 1.1419,
+      "step": 799
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.934726300523985e-05,
+      "loss": 1.0569,
+      "step": 800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.934493996357266e-05,
+      "loss": 1.0949,
+      "step": 801
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.934261285035483e-05,
+      "loss": 1.157,
+      "step": 802
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.934028166597555e-05,
+      "loss": 1.1424,
+      "step": 803
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.933794641082469e-05,
+      "loss": 1.0711,
+      "step": 804
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.933560708529281e-05,
+      "loss": 1.1355,
+      "step": 805
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.933326368977116e-05,
+      "loss": 1.0939,
+      "step": 806
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.933091622465165e-05,
+      "loss": 1.0835,
+      "step": 807
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.932856469032689e-05,
+      "loss": 1.1496,
+      "step": 808
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9326209087190136e-05,
+      "loss": 1.1712,
+      "step": 809
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.932384941563537e-05,
+      "loss": 1.263,
+      "step": 810
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.932148567605723e-05,
+      "loss": 1.0925,
+      "step": 811
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.931911786885103e-05,
+      "loss": 1.1539,
+      "step": 812
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.931674599441279e-05,
+      "loss": 1.1307,
+      "step": 813
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.9314370053139176e-05,
+      "loss": 1.1383,
+      "step": 814
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.9311990045427553e-05,
+      "loss": 1.1044,
+      "step": 815
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.9309605971675966e-05,
+      "loss": 1.0852,
+      "step": 816
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.930721783228314e-05,
+      "loss": 1.048,
+      "step": 817
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.9304825627648464e-05,
+      "loss": 1.1599,
+      "step": 818
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.930242935817204e-05,
+      "loss": 1.0368,
+      "step": 819
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.9300029024254624e-05,
+      "loss": 1.0306,
+      "step": 820
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.929762462629766e-05,
+      "loss": 1.3,
+      "step": 821
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.929521616470325e-05,
+      "loss": 1.2732,
+      "step": 822
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.929280363987422e-05,
+      "loss": 1.2388,
+      "step": 823
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.929038705221404e-05,
+      "loss": 1.0834,
+      "step": 824
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.9287966402126866e-05,
+      "loss": 1.1864,
+      "step": 825
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.928554169001755e-05,
+      "loss": 1.0441,
+      "step": 826
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.92831129162916e-05,
+      "loss": 1.2931,
+      "step": 827
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.928068008135523e-05,
+      "loss": 1.091,
+      "step": 828
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.927824318561529e-05,
+      "loss": 1.1243,
+      "step": 829
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.9275802229479364e-05,
+      "loss": 1.162,
+      "step": 830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.9273357213355664e-05,
+      "loss": 1.0815,
+      "step": 831
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.9270908137653125e-05,
+      "loss": 1.2216,
+      "step": 832
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.926845500278134e-05,
+      "loss": 1.1935,
+      "step": 833
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.926599780915057e-05,
+      "loss": 1.0181,
+      "step": 834
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.926353655717177e-05,
+      "loss": 1.0479,
+      "step": 835
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.926107124725658e-05,
+      "loss": 1.1773,
+      "step": 836
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.925860187981729e-05,
+      "loss": 1.0416,
+      "step": 837
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.92561284552669e-05,
+      "loss": 1.0793,
+      "step": 838
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.925365097401908e-05,
+      "loss": 1.1624,
+      "step": 839
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.925116943648816e-05,
+      "loss": 1.0042,
+      "step": 840
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.924868384308918e-05,
+      "loss": 1.0816,
+      "step": 841
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.924619419423783e-05,
+      "loss": 0.9979,
+      "step": 842
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.924370049035049e-05,
+      "loss": 0.9449,
+      "step": 843
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.9241202731844216e-05,
+      "loss": 1.271,
+      "step": 844
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.923870091913675e-05,
+      "loss": 1.1519,
+      "step": 845
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.923619505264651e-05,
+      "loss": 0.9836,
+      "step": 846
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.9233685132792575e-05,
+      "loss": 1.0272,
+      "step": 847
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.9231171159994716e-05,
+      "loss": 1.0762,
+      "step": 848
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.922865313467338e-05,
+      "loss": 1.2062,
+      "step": 849
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.9226131057249714e-05,
+      "loss": 1.1942,
+      "step": 850
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.922360492814549e-05,
+      "loss": 1.0529,
+      "step": 851
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.922107474778321e-05,
+      "loss": 1.0889,
+      "step": 852
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.921854051658602e-05,
+      "loss": 1.1773,
+      "step": 853
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.921600223497775e-05,
+      "loss": 1.1164,
+      "step": 854
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.921345990338294e-05,
+      "loss": 1.0152,
+      "step": 855
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.9210913522226755e-05,
+      "loss": 0.9851,
+      "step": 856
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.920836309193507e-05,
+      "loss": 1.2684,
+      "step": 857
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.920580861293443e-05,
+      "loss": 1.1949,
+      "step": 858
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.920325008565207e-05,
+      "loss": 1.1368,
+      "step": 859
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.920068751051586e-05,
+      "loss": 1.1158,
+      "step": 860
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.91981208879544e-05,
+      "loss": 1.2742,
+      "step": 861
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.919555021839693e-05,
+      "loss": 1.1223,
+      "step": 862
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.919297550227339e-05,
+      "loss": 1.1158,
+      "step": 863
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.919039674001437e-05,
+      "loss": 1.2152,
+      "step": 864
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.9187813932051175e-05,
+      "loss": 1.3577,
+      "step": 865
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.918522707881574e-05,
+      "loss": 0.892,
+      "step": 866
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.9182636180740735e-05,
+      "loss": 1.1983,
+      "step": 867
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.918004123825944e-05,
+      "loss": 1.1645,
+      "step": 868
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.917744225180586e-05,
+      "loss": 0.9923,
+      "step": 869
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.917483922181465e-05,
+      "loss": 0.9879,
+      "step": 870
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.917223214872116e-05,
+      "loss": 1.2166,
+      "step": 871
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.916962103296142e-05,
+      "loss": 1.1793,
+      "step": 872
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.9167005874972104e-05,
+      "loss": 1.22,
+      "step": 873
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.916438667519059e-05,
+      "loss": 1.1979,
+      "step": 874
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.916176343405492e-05,
+      "loss": 1.1595,
+      "step": 875
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.9159136152003817e-05,
+      "loss": 1.1903,
+      "step": 876
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.9156504829476685e-05,
+      "loss": 1.1589,
+      "step": 877
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.9153869466913596e-05,
+      "loss": 1.1397,
+      "step": 878
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.915123006475529e-05,
+      "loss": 1.2213,
+      "step": 879
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.91485866234432e-05,
+      "loss": 1.1326,
+      "step": 880
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.914593914341942e-05,
+      "loss": 0.9529,
+      "step": 881
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.914328762512674e-05,
+      "loss": 1.2612,
+      "step": 882
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.91406320690086e-05,
+      "loss": 1.2551,
+      "step": 883
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.913797247550912e-05,
+      "loss": 1.0373,
+      "step": 884
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.9135308845073115e-05,
+      "loss": 1.1486,
+      "step": 885
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.9132641178146055e-05,
+      "loss": 1.2224,
+      "step": 886
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.912996947517409e-05,
+      "loss": 1.2382,
+      "step": 887
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.9127293736604056e-05,
+      "loss": 1.0372,
+      "step": 888
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.912461396288345e-05,
+      "loss": 1.1342,
+      "step": 889
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.912193015446044e-05,
+      "loss": 1.3512,
+      "step": 890
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.911924231178389e-05,
+      "loss": 1.1981,
+      "step": 891
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.911655043530332e-05,
+      "loss": 1.1699,
+      "step": 892
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.911385452546893e-05,
+      "loss": 1.1196,
+      "step": 893
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.91111545827316e-05,
+      "loss": 0.9645,
+      "step": 894
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9108450607542875e-05,
+      "loss": 0.9653,
+      "step": 895
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9105742600354986e-05,
+      "loss": 1.2204,
+      "step": 896
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.910303056162082e-05,
+      "loss": 1.1371,
+      "step": 897
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.910031449179396e-05,
+      "loss": 1.1317,
+      "step": 898
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.909759439132866e-05,
+      "loss": 1.1376,
+      "step": 899
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.909487026067981e-05,
+      "loss": 0.9843,
+      "step": 900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.909214210030304e-05,
+      "loss": 1.2301,
+      "step": 901
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9089409910654604e-05,
+      "loss": 0.9985,
+      "step": 902
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.908667369219144e-05,
+      "loss": 1.0133,
+      "step": 903
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.908393344537118e-05,
+      "loss": 1.0576,
+      "step": 904
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.9081189170652104e-05,
+      "loss": 1.0806,
+      "step": 905
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.907844086849317e-05,
+      "loss": 1.1171,
+      "step": 906
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.9075688539354025e-05,
+      "loss": 1.1968,
+      "step": 907
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.907293218369499e-05,
+      "loss": 1.0889,
+      "step": 908
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.907017180197702e-05,
+      "loss": 1.0869,
+      "step": 909
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.9067407394661805e-05,
+      "loss": 1.0334,
+      "step": 910
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.906463896221165e-05,
+      "loss": 0.9784,
+      "step": 911
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.906186650508958e-05,
+      "loss": 1.0485,
+      "step": 912
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.905909002375926e-05,
+      "loss": 1.1622,
+      "step": 913
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.905630951868504e-05,
+      "loss": 1.1587,
+      "step": 914
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.9053524990331955e-05,
+      "loss": 0.9884,
+      "step": 915
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.905073643916569e-05,
+      "loss": 1.2752,
+      "step": 916
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.9047943865652615e-05,
+      "loss": 1.0065,
+      "step": 917
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.9045147270259785e-05,
+      "loss": 1.2779,
+      "step": 918
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.904234665345489e-05,
+      "loss": 1.2424,
+      "step": 919
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.903954201570634e-05,
+      "loss": 1.1238,
+      "step": 920
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.903673335748318e-05,
+      "loss": 1.2368,
+      "step": 921
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.9033920679255154e-05,
+      "loss": 0.9268,
+      "step": 922
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.903110398149265e-05,
+      "loss": 1.2021,
+      "step": 923
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.902828326466677e-05,
+      "loss": 1.0261,
+      "step": 924
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.902545852924924e-05,
+      "loss": 1.0627,
+      "step": 925
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.902262977571248e-05,
+      "loss": 1.1934,
+      "step": 926
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.901979700452959e-05,
+      "loss": 1.0808,
+      "step": 927
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.901696021617435e-05,
+      "loss": 0.9534,
+      "step": 928
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.9014119411121164e-05,
+      "loss": 1.0927,
+      "step": 929
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.901127458984516e-05,
+      "loss": 1.2552,
+      "step": 930
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.9008425752822126e-05,
+      "loss": 1.1121,
+      "step": 931
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.900557290052849e-05,
+      "loss": 1.1243,
+      "step": 932
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.90027160334414e-05,
+      "loss": 1.1547,
+      "step": 933
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.8999855152038635e-05,
+      "loss": 1.127,
+      "step": 934
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.8996990256798656e-05,
+      "loss": 1.2658,
+      "step": 935
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.899412134820062e-05,
+      "loss": 1.1001,
+      "step": 936
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.8991248426724314e-05,
+      "loss": 1.1673,
+      "step": 937
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.898837149285023e-05,
+      "loss": 1.1022,
+      "step": 938
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.898549054705952e-05,
+      "loss": 1.0041,
+      "step": 939
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.8982605589833996e-05,
+      "loss": 1.2081,
+      "step": 940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.897971662165616e-05,
+      "loss": 1.0209,
+      "step": 941
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.8976823643009176e-05,
+      "loss": 1.1901,
+      "step": 942
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.897392665437686e-05,
+      "loss": 1.0991,
+      "step": 943
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.897102565624374e-05,
+      "loss": 1.2988,
+      "step": 944
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.896812064909498e-05,
+      "loss": 1.0251,
+      "step": 945
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.896521163341642e-05,
+      "loss": 1.164,
+      "step": 946
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.896229860969459e-05,
+      "loss": 1.216,
+      "step": 947
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8959381578416674e-05,
+      "loss": 1.1574,
+      "step": 948
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.895646054007051e-05,
+      "loss": 1.1244,
+      "step": 949
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.895353549514466e-05,
+      "loss": 1.1461,
+      "step": 950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.895060644412828e-05,
+      "loss": 1.2348,
+      "step": 951
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.894767338751126e-05,
+      "loss": 1.1472,
+      "step": 952
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.8944736325784136e-05,
+      "loss": 1.1275,
+      "step": 953
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.894179525943811e-05,
+      "loss": 1.0965,
+      "step": 954
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.893885018896506e-05,
+      "loss": 1.188,
+      "step": 955
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.893590111485753e-05,
+      "loss": 1.1718,
+      "step": 956
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.893294803760874e-05,
+      "loss": 1.336,
+      "step": 957
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.892999095771258e-05,
+      "loss": 1.0233,
+      "step": 958
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.892702987566359e-05,
+      "loss": 0.9586,
+      "step": 959
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.8924064791957004e-05,
+      "loss": 1.1474,
+      "step": 960
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.892109570708871e-05,
+      "loss": 0.9468,
+      "step": 961
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.891812262155527e-05,
+      "loss": 1.112,
+      "step": 962
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.8915145535853925e-05,
+      "loss": 1.0545,
+      "step": 963
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.8912164450482564e-05,
+      "loss": 1.0944,
+      "step": 964
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.8909179365939767e-05,
+      "loss": 1.0121,
+      "step": 965
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.890619028272476e-05,
+      "loss": 1.1881,
+      "step": 966
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.8903197201337456e-05,
+      "loss": 1.0487,
+      "step": 967
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.890020012227844e-05,
+      "loss": 1.1772,
+      "step": 968
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.889719904604893e-05,
+      "loss": 1.0274,
+      "step": 969
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.889419397315087e-05,
+      "loss": 1.1874,
+      "step": 970
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.8891184904086816e-05,
+      "loss": 1.2198,
+      "step": 971
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.888817183936004e-05,
+      "loss": 1.0963,
+      "step": 972
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.888515477947444e-05,
+      "loss": 1.0551,
+      "step": 973
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.8882133724934606e-05,
+      "loss": 1.0842,
+      "step": 974
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.88791086762458e-05,
+      "loss": 1.2128,
+      "step": 975
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.887607963391394e-05,
+      "loss": 1.1506,
+      "step": 976
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.887304659844561e-05,
+      "loss": 0.9805,
+      "step": 977
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.887000957034807e-05,
+      "loss": 1.1289,
+      "step": 978
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.886696855012927e-05,
+      "loss": 0.9593,
+      "step": 979
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.8863923538297754e-05,
+      "loss": 1.3326,
+      "step": 980
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.8860874535362825e-05,
+      "loss": 1.1216,
+      "step": 981
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.8857821541834395e-05,
+      "loss": 1.1667,
+      "step": 982
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.885476455822305e-05,
+      "loss": 1.2297,
+      "step": 983
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.885170358504007e-05,
+      "loss": 1.0766,
+      "step": 984
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.884863862279739e-05,
+      "loss": 1.1822,
+      "step": 985
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.884556967200758e-05,
+      "loss": 1.0397,
+      "step": 986
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.884249673318393e-05,
+      "loss": 1.1113,
+      "step": 987
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.883941980684036e-05,
+      "loss": 1.1325,
+      "step": 988
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.883633889349146e-05,
+      "loss": 1.2022,
+      "step": 989
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.883325399365252e-05,
+      "loss": 1.118,
+      "step": 990
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.883016510783944e-05,
+      "loss": 1.1617,
+      "step": 991
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.882707223656885e-05,
+      "loss": 0.9902,
+      "step": 992
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.882397538035799e-05,
+      "loss": 1.0704,
+      "step": 993
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.88208745397248e-05,
+      "loss": 1.2108,
+      "step": 994
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.881776971518788e-05,
+      "loss": 1.1404,
+      "step": 995
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.88146609072665e-05,
+      "loss": 1.1302,
+      "step": 996
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.881154811648057e-05,
+      "loss": 1.0755,
+      "step": 997
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.8808431343350704e-05,
+      "loss": 1.1074,
+      "step": 998
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.880531058839816e-05,
+      "loss": 1.0761,
+      "step": 999
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.880218585214486e-05,
+      "loss": 1.0748,
+      "step": 1000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.879905713511341e-05,
+      "loss": 1.1122,
+      "step": 1001
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8795924437827065e-05,
+      "loss": 1.2171,
+      "step": 1002
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8792787760809746e-05,
+      "loss": 1.2337,
+      "step": 1003
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8789647104586044e-05,
+      "loss": 0.9352,
+      "step": 1004
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.878650246968122e-05,
+      "loss": 0.9225,
+      "step": 1005
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8783353856621193e-05,
+      "loss": 1.0731,
+      "step": 1006
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8780201265932554e-05,
+      "loss": 1.1429,
+      "step": 1007
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.877704469814256e-05,
+      "loss": 1.1673,
+      "step": 1008
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.8773884153779116e-05,
+      "loss": 1.1082,
+      "step": 1009
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.8770719633370806e-05,
+      "loss": 1.1216,
+      "step": 1010
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.87675511374469e-05,
+      "loss": 1.1886,
+      "step": 1011
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.8764378666537284e-05,
+      "loss": 0.9738,
+      "step": 1012
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.8761202221172536e-05,
+      "loss": 1.098,
+      "step": 1013
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.875802180188391e-05,
+      "loss": 1.2394,
+      "step": 1014
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.875483740920332e-05,
+      "loss": 1.1692,
+      "step": 1015
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.875164904366332e-05,
+      "loss": 1.1416,
+      "step": 1016
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.874845670579715e-05,
+      "loss": 1.196,
+      "step": 1017
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.874526039613872e-05,
+      "loss": 1.1185,
+      "step": 1018
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.874206011522259e-05,
+      "loss": 1.1219,
+      "step": 1019
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8738855863583977e-05,
+      "loss": 1.0658,
+      "step": 1020
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8735647641758785e-05,
+      "loss": 1.0402,
+      "step": 1021
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8732435450283565e-05,
+      "loss": 1.1317,
+      "step": 1022
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.872921928969555e-05,
+      "loss": 1.1211,
+      "step": 1023
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.872599916053261e-05,
+      "loss": 0.9332,
+      "step": 1024
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.872277506333329e-05,
+      "loss": 1.2266,
+      "step": 1025
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.871954699863682e-05,
+      "loss": 1.1362,
+      "step": 1026
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.8716314966983056e-05,
+      "loss": 1.2312,
+      "step": 1027
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.871307896891255e-05,
+      "loss": 1.0029,
+      "step": 1028
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.8709839004966506e-05,
+      "loss": 1.1768,
+      "step": 1029
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.870659507568677e-05,
+      "loss": 1.0698,
+      "step": 1030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.8703347181615885e-05,
+      "loss": 0.9573,
+      "step": 1031
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.8700095323297044e-05,
+      "loss": 1.1081,
+      "step": 1032
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.86968395012741e-05,
+      "loss": 1.2629,
+      "step": 1033
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.869357971609157e-05,
+      "loss": 1.0308,
+      "step": 1034
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.869031596829463e-05,
+      "loss": 1.152,
+      "step": 1035
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.868704825842912e-05,
+      "loss": 1.154,
+      "step": 1036
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.868377658704156e-05,
+      "loss": 1.1987,
+      "step": 1037
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.86805009546791e-05,
+      "loss": 1.1584,
+      "step": 1038
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.867722136188959e-05,
+      "loss": 0.9906,
+      "step": 1039
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.867393780922151e-05,
+      "loss": 1.0705,
+      "step": 1040
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8670650297224015e-05,
+      "loss": 1.2158,
+      "step": 1041
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.866735882644693e-05,
+      "loss": 1.2272,
+      "step": 1042
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8664063397440726e-05,
+      "loss": 1.0846,
+      "step": 1043
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.866076401075654e-05,
+      "loss": 1.1795,
+      "step": 1044
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.86574606669462e-05,
+      "loss": 1.1696,
+      "step": 1045
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.865415336656215e-05,
+      "loss": 0.9745,
+      "step": 1046
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.8650842110157516e-05,
+      "loss": 0.9207,
+      "step": 1047
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.86475268982861e-05,
+      "loss": 1.1601,
+      "step": 1048
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8644207731502346e-05,
+      "loss": 1.1268,
+      "step": 1049
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.864088461036136e-05,
+      "loss": 1.0165,
+      "step": 1050
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.863755753541892e-05,
+      "loss": 1.1122,
+      "step": 1051
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.863422650723145e-05,
+      "loss": 1.0776,
+      "step": 1052
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.863089152635606e-05,
+      "loss": 0.9534,
+      "step": 1053
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.862755259335051e-05,
+      "loss": 1.1021,
+      "step": 1054
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8624209708773195e-05,
+      "loss": 1.1878,
+      "step": 1055
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.862086287318321e-05,
+      "loss": 1.0683,
+      "step": 1056
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8617512087140296e-05,
+      "loss": 1.102,
+      "step": 1057
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.861415735120484e-05,
+      "loss": 1.2008,
+      "step": 1058
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8610798665937906e-05,
+      "loss": 1.0241,
+      "step": 1059
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.860743603190122e-05,
+      "loss": 1.1528,
+      "step": 1060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.8604069449657164e-05,
+      "loss": 1.0386,
+      "step": 1061
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.860069891976877e-05,
+      "loss": 1.1014,
+      "step": 1062
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.859732444279975e-05,
+      "loss": 1.1419,
+      "step": 1063
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.859394601931446e-05,
+      "loss": 1.0215,
+      "step": 1064
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.8590563649877925e-05,
+      "loss": 1.0412,
+      "step": 1065
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.858717733505582e-05,
+      "loss": 0.9531,
+      "step": 1066
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.858378707541449e-05,
+      "loss": 1.0338,
+      "step": 1067
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.8580392871520946e-05,
+      "loss": 1.0394,
+      "step": 1068
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.8576994723942834e-05,
+      "loss": 1.0212,
+      "step": 1069
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.857359263324849e-05,
+      "loss": 1.0845,
+      "step": 1070
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.857018660000688e-05,
+      "loss": 1.1604,
+      "step": 1071
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8566776624787655e-05,
+      "loss": 1.1084,
+      "step": 1072
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8563362708161095e-05,
+      "loss": 0.9791,
+      "step": 1073
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8559944850698184e-05,
+      "loss": 1.1704,
+      "step": 1074
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.855652305297052e-05,
+      "loss": 1.0425,
+      "step": 1075
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.855309731555038e-05,
+      "loss": 1.173,
+      "step": 1076
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.8549667639010705e-05,
+      "loss": 1.1475,
+      "step": 1077
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.854623402392509e-05,
+      "loss": 1.092,
+      "step": 1078
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.854279647086778e-05,
+      "loss": 1.077,
+      "step": 1079
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8539354980413686e-05,
+      "loss": 1.1821,
+      "step": 1080
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8535909553138395e-05,
+      "loss": 1.1113,
+      "step": 1081
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.853246018961811e-05,
+      "loss": 1.053,
+      "step": 1082
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.852900689042973e-05,
+      "loss": 0.9827,
+      "step": 1083
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.8525549656150795e-05,
+      "loss": 1.1359,
+      "step": 1084
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.852208848735952e-05,
+      "loss": 1.1196,
+      "step": 1085
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.8518623384634744e-05,
+      "loss": 1.054,
+      "step": 1086
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.8515154348556e-05,
+      "loss": 0.945,
+      "step": 1087
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.8511681379703464e-05,
+      "loss": 0.8866,
+      "step": 1088
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.850820447865797e-05,
+      "loss": 1.2624,
+      "step": 1089
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.850472364600101e-05,
+      "loss": 1.0359,
+      "step": 1090
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8501238882314715e-05,
+      "loss": 1.0978,
+      "step": 1091
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.849775018818191e-05,
+      "loss": 1.1402,
+      "step": 1092
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8494257564186064e-05,
+      "loss": 1.0152,
+      "step": 1093
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.849076101091129e-05,
+      "loss": 1.1453,
+      "step": 1094
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.848726052894236e-05,
+      "loss": 1.0436,
+      "step": 1095
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.848375611886472e-05,
+      "loss": 0.8516,
+      "step": 1096
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.8480247781264456e-05,
+      "loss": 1.2446,
+      "step": 1097
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.8476735516728315e-05,
+      "loss": 1.1928,
+      "step": 1098
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.847321932584371e-05,
+      "loss": 1.149,
+      "step": 1099
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.8469699209198705e-05,
+      "loss": 1.3131,
+      "step": 1100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.846617516738201e-05,
+      "loss": 1.1832,
+      "step": 1101
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.8462647200983e-05,
+      "loss": 1.0457,
+      "step": 1102
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.845911531059172e-05,
+      "loss": 1.0904,
+      "step": 1103
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.845557949679885e-05,
+      "loss": 1.1036,
+      "step": 1104
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.8452039760195736e-05,
+      "loss": 1.1688,
+      "step": 1105
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.844849610137438e-05,
+      "loss": 1.259,
+      "step": 1106
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.844494852092744e-05,
+      "loss": 1.0125,
+      "step": 1107
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.844139701944821e-05,
+      "loss": 1.0476,
+      "step": 1108
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.843784159753068e-05,
+      "loss": 1.0485,
+      "step": 1109
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.843428225576947e-05,
+      "loss": 1.1305,
+      "step": 1110
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.843071899475985e-05,
+      "loss": 1.0886,
+      "step": 1111
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.842715181509776e-05,
+      "loss": 1.1478,
+      "step": 1112
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.8423580717379796e-05,
+      "loss": 1.1012,
+      "step": 1113
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.8420005702203196e-05,
+      "loss": 1.1488,
+      "step": 1114
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.841642677016586e-05,
+      "loss": 1.0415,
+      "step": 1115
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.841284392186635e-05,
+      "loss": 0.9671,
+      "step": 1116
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.8409257157903875e-05,
+      "loss": 0.8175,
+      "step": 1117
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.84056664788783e-05,
+      "loss": 1.1585,
+      "step": 1118
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.84056664788783e-05,
+      "loss": 1.0728,
+      "step": 1119
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.840207188539014e-05,
+      "loss": 1.3042,
+      "step": 1120
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.8398473378040576e-05,
+      "loss": 1.2366,
+      "step": 1121
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.839487095743143e-05,
+      "loss": 1.197,
+      "step": 1122
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.83912646241652e-05,
+      "loss": 1.168,
+      "step": 1123
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.838765437884502e-05,
+      "loss": 1.217,
+      "step": 1124
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.838404022207467e-05,
+      "loss": 1.0453,
+      "step": 1125
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.83804221544586e-05,
+      "loss": 1.09,
+      "step": 1126
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.8376800176601925e-05,
+      "loss": 0.8732,
+      "step": 1127
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.837317428911039e-05,
+      "loss": 1.008,
+      "step": 1128
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.836954449259041e-05,
+      "loss": 1.1732,
+      "step": 1129
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.836591078764904e-05,
+      "loss": 1.0918,
+      "step": 1130
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.8362273174894e-05,
+      "loss": 1.115,
+      "step": 1131
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.8358631654933654e-05,
+      "loss": 1.0046,
+      "step": 1132
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.835498622837704e-05,
+      "loss": 1.252,
+      "step": 1133
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.835133689583381e-05,
+      "loss": 1.0946,
+      "step": 1134
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.834768365791432e-05,
+      "loss": 1.0814,
+      "step": 1135
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.834402651522954e-05,
+      "loss": 1.085,
+      "step": 1136
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.83403654683911e-05,
+      "loss": 1.1751,
+      "step": 1137
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.833670051801131e-05,
+      "loss": 1.1303,
+      "step": 1138
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.8333031664703086e-05,
+      "loss": 1.1067,
+      "step": 1139
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.832935890908004e-05,
+      "loss": 1.1294,
+      "step": 1140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.832568225175641e-05,
+      "loss": 1.0647,
+      "step": 1141
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.8322001693347105e-05,
+      "loss": 1.1915,
+      "step": 1142
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.8318317234467666e-05,
+      "loss": 1.0627,
+      "step": 1143
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.83146288757343e-05,
+      "loss": 1.1119,
+      "step": 1144
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.831093661776387e-05,
+      "loss": 1.1475,
+      "step": 1145
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.8307240461173885e-05,
+      "loss": 1.087,
+      "step": 1146
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.83035404065825e-05,
+      "loss": 1.0561,
+      "step": 1147
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.829983645460853e-05,
+      "loss": 1.1274,
+      "step": 1148
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.829612860587144e-05,
+      "loss": 1.1156,
+      "step": 1149
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.8292416860991355e-05,
+      "loss": 1.1249,
+      "step": 1150
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.828870122058902e-05,
+      "loss": 1.2167,
+      "step": 1151
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.828498168528588e-05,
+      "loss": 0.9957,
+      "step": 1152
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8281258255703986e-05,
+      "loss": 1.0336,
+      "step": 1153
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.827753093246606e-05,
+      "loss": 1.07,
+      "step": 1154
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.82737997161955e-05,
+      "loss": 1.0897,
+      "step": 1155
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.827006460751631e-05,
+      "loss": 1.1623,
+      "step": 1156
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.826632560705316e-05,
+      "loss": 1.1546,
+      "step": 1157
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.82625827154314e-05,
+      "loss": 0.9931,
+      "step": 1158
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.825883593327698e-05,
+      "loss": 1.0768,
+      "step": 1159
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.825508526121655e-05,
+      "loss": 1.0594,
+      "step": 1160
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.825133069987737e-05,
+      "loss": 1.1133,
+      "step": 1161
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.824757224988738e-05,
+      "loss": 1.0863,
+      "step": 1162
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.824380991187517e-05,
+      "loss": 1.0261,
+      "step": 1163
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.824004368646994e-05,
+      "loss": 1.0729,
+      "step": 1164
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.823627357430158e-05,
+      "loss": 1.0857,
+      "step": 1165
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.823249957600064e-05,
+      "loss": 0.9054,
+      "step": 1166
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.822872169219828e-05,
+      "loss": 1.105,
+      "step": 1167
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.8224939923526324e-05,
+      "loss": 1.1302,
+      "step": 1168
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.822115427061726e-05,
+      "loss": 1.0097,
+      "step": 1169
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.821736473410423e-05,
+      "loss": 1.0875,
+      "step": 1170
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.821357131462099e-05,
+      "loss": 1.1034,
+      "step": 1171
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.820977401280198e-05,
+      "loss": 1.0482,
+      "step": 1172
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.820597282928227e-05,
+      "loss": 1.0078,
+      "step": 1173
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.820216776469759e-05,
+      "loss": 1.138,
+      "step": 1174
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.819835881968432e-05,
+      "loss": 1.0485,
+      "step": 1175
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.819454599487947e-05,
+      "loss": 1.2125,
+      "step": 1176
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.819072929092072e-05,
+      "loss": 1.0008,
+      "step": 1177
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8186908708446397e-05,
+      "loss": 1.1589,
+      "step": 1178
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.818308424809547e-05,
+      "loss": 0.9593,
+      "step": 1179
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.8179255910507555e-05,
+      "loss": 1.0073,
+      "step": 1180
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.817542369632292e-05,
+      "loss": 1.0443,
+      "step": 1181
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.817158760618247e-05,
+      "loss": 1.1317,
+      "step": 1182
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.816774764072779e-05,
+      "loss": 1.1257,
+      "step": 1183
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.816390380060108e-05,
+      "loss": 1.2697,
+      "step": 1184
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8160056086445194e-05,
+      "loss": 0.9222,
+      "step": 1185
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8156204498903655e-05,
+      "loss": 1.1415,
+      "step": 1186
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.8152349038620607e-05,
+      "loss": 1.0593,
+      "step": 1187
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.8148489706240864e-05,
+      "loss": 1.1186,
+      "step": 1188
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.8144626502409865e-05,
+      "loss": 1.0086,
+      "step": 1189
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.814075942777372e-05,
+      "loss": 0.9753,
+      "step": 1190
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.813688848297916e-05,
+      "loss": 1.1117,
+      "step": 1191
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.813301366867359e-05,
+      "loss": 1.0023,
+      "step": 1192
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.8129134985505046e-05,
+      "loss": 1.0713,
+      "step": 1193
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.812525243412223e-05,
+      "loss": 1.1231,
+      "step": 1194
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.812136601517444e-05,
+      "loss": 1.0869,
+      "step": 1195
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.81174757293117e-05,
+      "loss": 1.0166,
+      "step": 1196
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.811358157718461e-05,
+      "loss": 1.0485,
+      "step": 1197
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.810968355944445e-05,
+      "loss": 1.1127,
+      "step": 1198
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.810578167674316e-05,
+      "loss": 1.0975,
+      "step": 1199
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.810187592973327e-05,
+      "loss": 1.0269,
+      "step": 1200
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.809796631906802e-05,
+      "loss": 1.0855,
+      "step": 1201
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.809405284540127e-05,
+      "loss": 1.0431,
+      "step": 1202
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.8090135509387527e-05,
+      "loss": 1.04,
+      "step": 1203
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.808621431168192e-05,
+      "loss": 1.0995,
+      "step": 1204
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.808228925294027e-05,
+      "loss": 1.1182,
+      "step": 1205
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.807836033381901e-05,
+      "loss": 1.2184,
+      "step": 1206
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.807442755497524e-05,
+      "loss": 1.0487,
+      "step": 1207
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.807049091706668e-05,
+      "loss": 1.0988,
+      "step": 1208
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.806655042075172e-05,
+      "loss": 1.0896,
+      "step": 1209
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.806260606668938e-05,
+      "loss": 1.0896,
+      "step": 1210
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.805865785553934e-05,
+      "loss": 1.0492,
+      "step": 1211
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.805470578796189e-05,
+      "loss": 1.044,
+      "step": 1212
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.805074986461803e-05,
+      "loss": 1.0027,
+      "step": 1213
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.804679008616933e-05,
+      "loss": 1.1141,
+      "step": 1214
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.804282645327806e-05,
+      "loss": 0.9824,
+      "step": 1215
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.80388589666071e-05,
+      "loss": 0.9886,
+      "step": 1216
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.803488762682e-05,
+      "loss": 1.2537,
+      "step": 1217
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.8030912434580954e-05,
+      "loss": 1.2307,
+      "step": 1218
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.8026933390554765e-05,
+      "loss": 1.2005,
+      "step": 1219
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.8022950495406925e-05,
+      "loss": 1.0312,
+      "step": 1220
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.8018963749803534e-05,
+      "loss": 1.1327,
+      "step": 1221
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.801497315441137e-05,
+      "loss": 1.0034,
+      "step": 1222
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.801097870989782e-05,
+      "loss": 1.2499,
+      "step": 1223
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.800698041693095e-05,
+      "loss": 1.0363,
+      "step": 1224
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.800297827617943e-05,
+      "loss": 1.0661,
+      "step": 1225
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.7998972288312616e-05,
+      "loss": 1.1207,
+      "step": 1226
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.799496245400047e-05,
+      "loss": 1.0311,
+      "step": 1227
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.799094877391363e-05,
+      "loss": 1.1634,
+      "step": 1228
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.7986931248723335e-05,
+      "loss": 1.1517,
+      "step": 1229
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.7982909879101515e-05,
+      "loss": 0.9758,
+      "step": 1230
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.797888466572072e-05,
+      "loss": 0.9969,
+      "step": 1231
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.797485560925413e-05,
+      "loss": 1.1351,
+      "step": 1232
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.79708227103756e-05,
+      "loss": 0.9946,
+      "step": 1233
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.7966785969759594e-05,
+      "loss": 1.0221,
+      "step": 1234
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.7962745388081234e-05,
+      "loss": 1.1209,
+      "step": 1235
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.795870096601629e-05,
+      "loss": 0.9509,
+      "step": 1236
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.795465270424117e-05,
+      "loss": 1.0313,
+      "step": 1237
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.795060060343292e-05,
+      "loss": 0.939,
+      "step": 1238
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.794654466426923e-05,
+      "loss": 0.8975,
+      "step": 1239
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.794248488742843e-05,
+      "loss": 1.2259,
+      "step": 1240
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.793842127358949e-05,
+      "loss": 1.1054,
+      "step": 1241
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.7934353823432046e-05,
+      "loss": 0.9365,
+      "step": 1242
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.793028253763633e-05,
+      "loss": 0.9626,
+      "step": 1243
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.792620741688325e-05,
+      "loss": 1.0341,
+      "step": 1244
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.792212846185436e-05,
+      "loss": 1.1603,
+      "step": 1245
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.791804567323182e-05,
+      "loss": 1.1524,
+      "step": 1246
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.7913959051698477e-05,
+      "loss": 1.0065,
+      "step": 1247
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.790986859793777e-05,
+      "loss": 1.0429,
+      "step": 1248
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.790577431263382e-05,
+      "loss": 1.1293,
+      "step": 1249
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.790167619647137e-05,
+      "loss": 1.0631,
+      "step": 1250
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.7897574250135805e-05,
+      "loss": 0.9682,
+      "step": 1251
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.789346847431314e-05,
+      "loss": 0.941,
+      "step": 1252
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.7889358869690056e-05,
+      "loss": 1.2343,
+      "step": 1253
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.788524543695387e-05,
+      "loss": 1.1385,
+      "step": 1254
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.788112817679251e-05,
+      "loss": 1.0963,
+      "step": 1255
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.7877007089894565e-05,
+      "loss": 1.0661,
+      "step": 1256
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.787288217694928e-05,
+      "loss": 1.2292,
+      "step": 1257
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.7868753438646504e-05,
+      "loss": 1.0744,
+      "step": 1258
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.7864620875676766e-05,
+      "loss": 1.0699,
+      "step": 1259
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.786048448873119e-05,
+      "loss": 1.1656,
+      "step": 1260
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.785634427850157e-05,
+      "loss": 1.3201,
+      "step": 1261
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.7852200245680343e-05,
+      "loss": 0.8484,
+      "step": 1262
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.7848052390960575e-05,
+      "loss": 1.1495,
+      "step": 1263
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.784390071503595e-05,
+      "loss": 1.1208,
+      "step": 1264
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.783974521860083e-05,
+      "loss": 0.938,
+      "step": 1265
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.783558590235019e-05,
+      "loss": 0.9459,
+      "step": 1266
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.783142276697966e-05,
+      "loss": 1.1732,
+      "step": 1267
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.782725581318548e-05,
+      "loss": 1.1314,
+      "step": 1268
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.782308504166458e-05,
+      "loss": 1.1714,
+      "step": 1269
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.781891045311447e-05,
+      "loss": 1.1531,
+      "step": 1270
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.7814732048233345e-05,
+      "loss": 1.107,
+      "step": 1271
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.7810549827720006e-05,
+      "loss": 1.1468,
+      "step": 1272
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.780636379227391e-05,
+      "loss": 1.1241,
+      "step": 1273
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.780217394259515e-05,
+      "loss": 1.0904,
+      "step": 1274
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.779798027938445e-05,
+      "loss": 1.1787,
+      "step": 1275
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.779378280334318e-05,
+      "loss": 1.0773,
+      "step": 1276
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.7789581515173335e-05,
+      "loss": 0.9112,
+      "step": 1277
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.778537641557757e-05,
+      "loss": 1.2179,
+      "step": 1278
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.778116750525915e-05,
+      "loss": 1.2112,
+      "step": 1279
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.7776954784922e-05,
+      "loss": 0.9886,
+      "step": 1280
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.7772738255270666e-05,
+      "loss": 1.0914,
+      "step": 1281
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.7768517917010345e-05,
+      "loss": 1.1683,
+      "step": 1282
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.776429377084687e-05,
+      "loss": 1.1935,
+      "step": 1283
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.776006581748669e-05,
+      "loss": 0.9934,
+      "step": 1284
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.7755834057636907e-05,
+      "loss": 1.0844,
+      "step": 1285
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.775159849200527e-05,
+      "loss": 1.3109,
+      "step": 1286
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.774735912130014e-05,
+      "loss": 1.1474,
+      "step": 1287
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.7743115946230536e-05,
+      "loss": 1.1244,
+      "step": 1288
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.773886896750611e-05,
+      "loss": 1.064,
+      "step": 1289
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.773461818583713e-05,
+      "loss": 0.9165,
+      "step": 1290
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.773036360193452e-05,
+      "loss": 0.9214,
+      "step": 1291
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.7726105216509845e-05,
+      "loss": 1.1717,
+      "step": 1292
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.772184303027528e-05,
+      "loss": 1.0976,
+      "step": 1293
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.7717577043943665e-05,
+      "loss": 1.0897,
+      "step": 1294
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.7713307258228445e-05,
+      "loss": 1.1009,
+      "step": 1295
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.770903367384373e-05,
+      "loss": 0.9324,
+      "step": 1296
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.770475629150425e-05,
+      "loss": 1.1884,
+      "step": 1297
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.770047511192537e-05,
+      "loss": 0.9502,
+      "step": 1298
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.7696190135823094e-05,
+      "loss": 0.9627,
+      "step": 1299
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.769190136391406e-05,
+      "loss": 1.0068,
+      "step": 1300
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.768760879691554e-05,
+      "loss": 1.0354,
+      "step": 1301
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.768331243554544e-05,
+      "loss": 1.0733,
+      "step": 1302
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.76790122805223e-05,
+      "loss": 1.1491,
+      "step": 1303
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.7674708332565296e-05,
+      "loss": 1.0411,
+      "step": 1304
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.767040059239425e-05,
+      "loss": 1.0336,
+      "step": 1305
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.766608906072959e-05,
+      "loss": 0.9779,
+      "step": 1306
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.766177373829241e-05,
+      "loss": 0.9321,
+      "step": 1307
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.765745462580441e-05,
+      "loss": 0.9964,
+      "step": 1308
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.7653131723987945e-05,
+      "loss": 1.1236,
+      "step": 1309
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.764880503356599e-05,
+      "loss": 1.1159,
+      "step": 1310
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.764447455526215e-05,
+      "loss": 0.9426,
+      "step": 1311
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.76401402898007e-05,
+      "loss": 1.2339,
+      "step": 1312
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.7635802237906504e-05,
+      "loss": 0.9662,
+      "step": 1313
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.7631460400305064e-05,
+      "loss": 1.2311,
+      "step": 1314
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.762711477772255e-05,
+      "loss": 1.2009,
+      "step": 1315
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.762276537088573e-05,
+      "loss": 1.0881,
+      "step": 1316
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.761841218052202e-05,
+      "loss": 1.1999,
+      "step": 1317
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.761405520735945e-05,
+      "loss": 0.8791,
+      "step": 1318
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.7609694452126735e-05,
+      "loss": 1.1511,
+      "step": 1319
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.760532991555315e-05,
+      "loss": 0.9848,
+      "step": 1320
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.7600961598368665e-05,
+      "loss": 1.0115,
+      "step": 1321
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.759658950130385e-05,
+      "loss": 1.1487,
+      "step": 1322
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.759221362508989e-05,
+      "loss": 1.0452,
+      "step": 1323
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.7587833970458654e-05,
+      "loss": 0.8973,
+      "step": 1324
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.7583450538142596e-05,
+      "loss": 1.0475,
+      "step": 1325
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.757906332887483e-05,
+      "loss": 1.2135,
+      "step": 1326
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.757467234338908e-05,
+      "loss": 1.0657,
+      "step": 1327
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.757027758241973e-05,
+      "loss": 1.0676,
+      "step": 1328
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.7565879046701765e-05,
+      "loss": 1.1054,
+      "step": 1329
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.7561476736970824e-05,
+      "loss": 1.0785,
+      "step": 1330
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.7557070653963154e-05,
+      "loss": 1.2202,
+      "step": 1331
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.755266079841566e-05,
+      "loss": 1.0444,
+      "step": 1332
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.7548247171065854e-05,
+      "loss": 1.1298,
+      "step": 1333
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.7543829772651904e-05,
+      "loss": 1.054,
+      "step": 1334
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.753940860391258e-05,
+      "loss": 0.9518,
+      "step": 1335
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.7534983665587314e-05,
+      "loss": 1.1667,
+      "step": 1336
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.753055495841613e-05,
+      "loss": 0.9729,
+      "step": 1337
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.752612248313972e-05,
+      "loss": 1.1434,
+      "step": 1338
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.752168624049938e-05,
+      "loss": 1.0557,
+      "step": 1339
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.751724623123705e-05,
+      "loss": 1.2572,
+      "step": 1340
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.75128024560953e-05,
+      "loss": 0.9773,
+      "step": 1341
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.750835491581732e-05,
+      "loss": 1.1188,
+      "step": 1342
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.7503903611146935e-05,
+      "loss": 1.1653,
+      "step": 1343
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.749944854282861e-05,
+      "loss": 1.1089,
+      "step": 1344
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.749498971160742e-05,
+      "loss": 1.0734,
+      "step": 1345
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.749052711822908e-05,
+      "loss": 1.0948,
+      "step": 1346
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.7486060763439926e-05,
+      "loss": 1.1857,
+      "step": 1347
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.748159064798694e-05,
+      "loss": 1.1001,
+      "step": 1348
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.7477116772617725e-05,
+      "loss": 1.0824,
+      "step": 1349
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.747263913808051e-05,
+      "loss": 1.0456,
+      "step": 1350
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.746815774512414e-05,
+      "loss": 1.1458,
+      "step": 1351
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.746367259449812e-05,
+      "loss": 1.1218,
+      "step": 1352
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.745918368695256e-05,
+      "loss": 1.2907,
+      "step": 1353
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7454691023238187e-05,
+      "loss": 0.9682,
+      "step": 1354
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7450194604106404e-05,
+      "loss": 0.9141,
+      "step": 1355
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7445694430309186e-05,
+      "loss": 1.0893,
+      "step": 1356
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7441190502599164e-05,
+      "loss": 0.8839,
+      "step": 1357
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.74366828217296e-05,
+      "loss": 1.0656,
+      "step": 1358
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7432171388454396e-05,
+      "loss": 1.0039,
+      "step": 1359
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7427656203528024e-05,
+      "loss": 1.056,
+      "step": 1360
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7423137267705655e-05,
+      "loss": 0.968,
+      "step": 1361
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.741861458174304e-05,
+      "loss": 1.1377,
+      "step": 1362
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.741408814639657e-05,
+      "loss": 0.9976,
+      "step": 1363
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.740955796242328e-05,
+      "loss": 1.1285,
+      "step": 1364
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.74050240305808e-05,
+      "loss": 0.9693,
+      "step": 1365
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.7400486351627415e-05,
+      "loss": 1.1433,
+      "step": 1366
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.739594492632202e-05,
+      "loss": 1.1723,
+      "step": 1367
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.739139975542415e-05,
+      "loss": 1.0497,
+      "step": 1368
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.7386850839693944e-05,
+      "loss": 1.0049,
+      "step": 1369
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.738229817989219e-05,
+      "loss": 1.0461,
+      "step": 1370
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.7377741776780296e-05,
+      "loss": 1.1654,
+      "step": 1371
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.73731816311203e-05,
+      "loss": 1.0979,
+      "step": 1372
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.736861774367485e-05,
+      "loss": 0.9366,
+      "step": 1373
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.736405011520723e-05,
+      "loss": 1.0927,
+      "step": 1374
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.7359478746481354e-05,
+      "loss": 0.9057,
+      "step": 1375
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.7354903638261746e-05,
+      "loss": 1.2918,
+      "step": 1376
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.735032479131358e-05,
+      "loss": 1.0791,
+      "step": 1377
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.734574220640264e-05,
+      "loss": 1.1342,
+      "step": 1378
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.734115588429534e-05,
+      "loss": 1.1846,
+      "step": 1379
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.7336565825758706e-05,
+      "loss": 1.0362,
+      "step": 1380
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.7331972031560404e-05,
+      "loss": 1.1277,
+      "step": 1381
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.732737450246871e-05,
+      "loss": 0.9899,
+      "step": 1382
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.7322773239252547e-05,
+      "loss": 1.0567,
+      "step": 1383
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.731816824268145e-05,
+      "loss": 1.081,
+      "step": 1384
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.731355951352557e-05,
+      "loss": 1.1533,
+      "step": 1385
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.730894705255569e-05,
+      "loss": 1.0674,
+      "step": 1386
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.730433086054322e-05,
+      "loss": 1.1165,
+      "step": 1387
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.72997109382602e-05,
+      "loss": 0.949,
+      "step": 1388
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.729508728647928e-05,
+      "loss": 1.0272,
+      "step": 1389
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.7290459905973727e-05,
+      "loss": 1.1629,
+      "step": 1390
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.7285828797517465e-05,
+      "loss": 1.0957,
+      "step": 1391
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.7281193961885e-05,
+      "loss": 1.0839,
+      "step": 1392
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.7276555399851495e-05,
+      "loss": 1.0261,
+      "step": 1393
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.727191311219271e-05,
+      "loss": 1.0513,
+      "step": 1394
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.726726709968506e-05,
+      "loss": 1.0259,
+      "step": 1395
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.7262617363105545e-05,
+      "loss": 1.0363,
+      "step": 1396
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.7257963903231814e-05,
+      "loss": 1.0706,
+      "step": 1397
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.725330672084214e-05,
+      "loss": 1.1628,
+      "step": 1398
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.72486458167154e-05,
+      "loss": 1.1895,
+      "step": 1399
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.724398119163109e-05,
+      "loss": 0.8837,
+      "step": 1400
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.7239312846369375e-05,
+      "loss": 0.8794,
+      "step": 1401
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.723464078171098e-05,
+      "loss": 1.0211,
+      "step": 1402
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.722996499843729e-05,
+      "loss": 1.0954,
+      "step": 1403
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.722528549733031e-05,
+      "loss": 1.1208,
+      "step": 1404
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.7220602279172646e-05,
+      "loss": 1.0662,
+      "step": 1405
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.721591534474755e-05,
+      "loss": 1.0731,
+      "step": 1406
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.721122469483888e-05,
+      "loss": 1.1327,
+      "step": 1407
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.720653033023113e-05,
+      "loss": 0.9303,
+      "step": 1408
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.720183225170938e-05,
+      "loss": 1.0451,
+      "step": 1409
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.719713046005938e-05,
+      "loss": 1.1937,
+      "step": 1410
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.719242495606748e-05,
+      "loss": 1.1178,
+      "step": 1411
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.7187715740520624e-05,
+      "loss": 1.0944,
+      "step": 1412
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.718300281420643e-05,
+      "loss": 1.1489,
+      "step": 1413
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.717828617791308e-05,
+      "loss": 1.0732,
+      "step": 1414
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.717356583242943e-05,
+      "loss": 1.0772,
+      "step": 1415
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.716884177854491e-05,
+      "loss": 1.0208,
+      "step": 1416
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.716411401704961e-05,
+      "loss": 0.9993,
+      "step": 1417
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.71593825487342e-05,
+      "loss": 1.0876,
+      "step": 1418
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.7154647374390005e-05,
+      "loss": 1.0719,
+      "step": 1419
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.7149908494808956e-05,
+      "loss": 0.8838,
+      "step": 1420
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.7145165910783596e-05,
+      "loss": 1.1803,
+      "step": 1421
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.714041962310709e-05,
+      "loss": 1.0851,
+      "step": 1422
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.713566963257324e-05,
+      "loss": 1.1785,
+      "step": 1423
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.713091593997645e-05,
+      "loss": 0.9553,
+      "step": 1424
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.712615854611174e-05,
+      "loss": 1.1266,
+      "step": 1425
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.712139745177477e-05,
+      "loss": 1.0163,
+      "step": 1426
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.7116632657761805e-05,
+      "loss": 0.91,
+      "step": 1427
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.711186416486971e-05,
+      "loss": 1.0589,
+      "step": 1428
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.710709197389601e-05,
+      "loss": 1.2108,
+      "step": 1429
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.710231608563882e-05,
+      "loss": 0.9798,
+      "step": 1430
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.709753650089687e-05,
+      "loss": 1.1001,
+      "step": 1431
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.709275322046953e-05,
+      "loss": 1.1018,
+      "step": 1432
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.708796624515677e-05,
+      "loss": 1.1515,
+      "step": 1433
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.708317557575919e-05,
+      "loss": 1.1088,
+      "step": 1434
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.707838121307799e-05,
+      "loss": 0.9428,
+      "step": 1435
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.7073583157915005e-05,
+      "loss": 1.0276,
+      "step": 1436
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.7068781411072686e-05,
+      "loss": 1.1672,
+      "step": 1437
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.7063975973354095e-05,
+      "loss": 1.1755,
+      "step": 1438
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.705916684556292e-05,
+      "loss": 1.0445,
+      "step": 1439
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.705435402850345e-05,
+      "loss": 1.1323,
+      "step": 1440
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.70495375229806e-05,
+      "loss": 1.1121,
+      "step": 1441
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.704471732979991e-05,
+      "loss": 0.9304,
+      "step": 1442
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.7039893449767526e-05,
+      "loss": 0.8747,
+      "step": 1443
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.703506588369021e-05,
+      "loss": 1.1185,
+      "step": 1444
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.703023463237536e-05,
+      "loss": 1.0789,
+      "step": 1445
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.7025399696630954e-05,
+      "loss": 0.9725,
+      "step": 1446
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.702056107726563e-05,
+      "loss": 1.0545,
+      "step": 1447
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.70157187750886e-05,
+      "loss": 1.0274,
+      "step": 1448
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.701087279090971e-05,
+      "loss": 0.8974,
+      "step": 1449
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.7006023125539444e-05,
+      "loss": 1.0592,
+      "step": 1450
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.700116977978885e-05,
+      "loss": 1.1467,
+      "step": 1451
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.699631275446965e-05,
+      "loss": 1.0182,
+      "step": 1452
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.6991452050394145e-05,
+      "loss": 1.063,
+      "step": 1453
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.698658766837526e-05,
+      "loss": 1.1534,
+      "step": 1454
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.698171960922652e-05,
+      "loss": 0.9764,
+      "step": 1455
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.69768478737621e-05,
+      "loss": 1.1119,
+      "step": 1456
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.697197246279675e-05,
+      "loss": 0.9891,
+      "step": 1457
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.696709337714588e-05,
+      "loss": 1.0612,
+      "step": 1458
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.696221061762547e-05,
+      "loss": 1.0956,
+      "step": 1459
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.695732418505214e-05,
+      "loss": 0.9689,
+      "step": 1460
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.695243408024311e-05,
+      "loss": 0.9985,
+      "step": 1461
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.694754030401623e-05,
+      "loss": 0.9084,
+      "step": 1462
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.694264285718995e-05,
+      "loss": 0.9869,
+      "step": 1463
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.6937741740583344e-05,
+      "loss": 0.978,
+      "step": 1464
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.69328369550161e-05,
+      "loss": 0.9736,
+      "step": 1465
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.6927928501308506e-05,
+      "loss": 1.0464,
+      "step": 1466
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.6923016380281474e-05,
+      "loss": 1.1218,
+      "step": 1467
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.691810059275653e-05,
+      "loss": 1.0586,
+      "step": 1468
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.691318113955581e-05,
+      "loss": 0.9336,
+      "step": 1469
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.690825802150207e-05,
+      "loss": 1.1408,
+      "step": 1470
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.690333123941867e-05,
+      "loss": 1.0009,
+      "step": 1471
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.689840079412958e-05,
+      "loss": 1.1192,
+      "step": 1472
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.689346668645939e-05,
+      "loss": 1.108,
+      "step": 1473
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.68885289172333e-05,
+      "loss": 1.0432,
+      "step": 1474
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6883587487277133e-05,
+      "loss": 1.0352,
+      "step": 1475
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.687864239741731e-05,
+      "loss": 1.1371,
+      "step": 1476
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.687369364848087e-05,
+      "loss": 1.068,
+      "step": 1477
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.686874124129545e-05,
+      "loss": 1.0048,
+      "step": 1478
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.686378517668932e-05,
+      "loss": 0.9304,
+      "step": 1479
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.6858825455491365e-05,
+      "loss": 1.093,
+      "step": 1480
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.685386207853106e-05,
+      "loss": 1.0743,
+      "step": 1481
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.68488950466385e-05,
+      "loss": 1.011,
+      "step": 1482
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.684392436064439e-05,
+      "loss": 0.903,
+      "step": 1483
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.6838950021380047e-05,
+      "loss": 0.842,
+      "step": 1484
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.683397202967741e-05,
+      "loss": 1.2155,
+      "step": 1485
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.682899038636901e-05,
+      "loss": 0.9828,
+      "step": 1486
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.6824005092288005e-05,
+      "loss": 1.0466,
+      "step": 1487
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.6819016148268154e-05,
+      "loss": 1.0897,
+      "step": 1488
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.681402355514383e-05,
+      "loss": 0.973,
+      "step": 1489
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.680902731375001e-05,
+      "loss": 1.0956,
+      "step": 1490
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.68040274249223e-05,
+      "loss": 0.9887,
+      "step": 1491
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.6799023889496884e-05,
+      "loss": 0.7988,
+      "step": 1492
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.67940167083106e-05,
+      "loss": 1.2009,
+      "step": 1493
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.678900588220083e-05,
+      "loss": 1.1429,
+      "step": 1494
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.678399141200565e-05,
+      "loss": 1.1,
+      "step": 1495
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.677897329856367e-05,
+      "loss": 1.2595,
+      "step": 1496
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.677395154271416e-05,
+      "loss": 1.1394,
+      "step": 1497
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.6768926145296965e-05,
+      "loss": 0.9947,
+      "step": 1498
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.676389710715256e-05,
+      "loss": 1.0428,
+      "step": 1499
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.675886442912202e-05,
+      "loss": 1.0582,
+      "step": 1500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.675382811204704e-05,
+      "loss": 1.1181,
+      "step": 1501
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.6748788156769904e-05,
+      "loss": 1.222,
+      "step": 1502
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.674374456413353e-05,
+      "loss": 0.9651,
+      "step": 1503
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.673869733498141e-05,
+      "loss": 1.0042,
+      "step": 1504
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.6733646470157674e-05,
+      "loss": 0.9929,
+      "step": 1505
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.6728591970507055e-05,
+      "loss": 1.0749,
+      "step": 1506
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.6723533836874885e-05,
+      "loss": 1.0398,
+      "step": 1507
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.67184720701071e-05,
+      "loss": 1.1068,
+      "step": 1508
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.671340667105027e-05,
+      "loss": 1.0522,
+      "step": 1509
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.6708337640551525e-05,
+      "loss": 1.1008,
+      "step": 1510
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.670326497945866e-05,
+      "loss": 0.9944,
+      "step": 1511
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.6698188688620026e-05,
+      "loss": 0.927,
+      "step": 1512
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.669310876888462e-05,
+      "loss": 0.7742,
+      "step": 1513
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.668802522110202e-05,
+      "loss": 1.114,
+      "step": 1514
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.6682938046122425e-05,
+      "loss": 1.0832,
+      "step": 1515
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.667784724479663e-05,
+      "loss": 1.2499,
+      "step": 1516
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.667275281797605e-05,
+      "loss": 1.1963,
+      "step": 1517
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.666765476651269e-05,
+      "loss": 1.1552,
+      "step": 1518
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.666255309125917e-05,
+      "loss": 1.1226,
+      "step": 1519
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.665744779306872e-05,
+      "loss": 1.1816,
+      "step": 1520
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.6652338872795174e-05,
+      "loss": 0.9878,
+      "step": 1521
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.664722633129297e-05,
+      "loss": 1.0442,
+      "step": 1522
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.664211016941714e-05,
+      "loss": 0.8199,
+      "step": 1523
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.663699038802334e-05,
+      "loss": 0.9547,
+      "step": 1524
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.663186698796783e-05,
+      "loss": 1.1323,
+      "step": 1525
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.6626739970107457e-05,
+      "loss": 1.0468,
+      "step": 1526
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.662160933529969e-05,
+      "loss": 1.0646,
+      "step": 1527
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.66164750844026e-05,
+      "loss": 0.9612,
+      "step": 1528
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.661133721827486e-05,
+      "loss": 1.1999,
+      "step": 1529
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.6606195737775754e-05,
+      "loss": 1.0432,
+      "step": 1530
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.660105064376515e-05,
+      "loss": 1.0318,
+      "step": 1531
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.6595901937103546e-05,
+      "loss": 1.0296,
+      "step": 1532
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.659074961865204e-05,
+      "loss": 1.13,
+      "step": 1533
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.658559368927231e-05,
+      "loss": 1.0877,
+      "step": 1534
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.658043414982667e-05,
+      "loss": 1.0586,
+      "step": 1535
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.6575271001178015e-05,
+      "loss": 1.0737,
+      "step": 1536
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.657010424418986e-05,
+      "loss": 1.0113,
+      "step": 1537
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.65649338797263e-05,
+      "loss": 1.1448,
+      "step": 1538
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.655975990865207e-05,
+      "loss": 1.0069,
+      "step": 1539
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.655458233183247e-05,
+      "loss": 1.0674,
+      "step": 1540
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.654940115013342e-05,
+      "loss": 1.1004,
+      "step": 1541
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.6544216364421456e-05,
+      "loss": 1.0437,
+      "step": 1542
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.6539027975563695e-05,
+      "loss": 1.0083,
+      "step": 1543
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.6533835984427865e-05,
+      "loss": 1.0732,
+      "step": 1544
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.65286403918823e-05,
+      "loss": 1.0632,
+      "step": 1545
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.652344119879592e-05,
+      "loss": 1.0763,
+      "step": 1546
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.6518238406038276e-05,
+      "loss": 1.1734,
+      "step": 1547
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.6513032014479494e-05,
+      "loss": 0.9477,
+      "step": 1548
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.650782202499032e-05,
+      "loss": 0.9864,
+      "step": 1549
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.6502608438442095e-05,
+      "loss": 1.0233,
+      "step": 1550
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.649739125570676e-05,
+      "loss": 1.0434,
+      "step": 1551
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.649217047765685e-05,
+      "loss": 1.1159,
+      "step": 1552
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.6486946105165527e-05,
+      "loss": 1.1057,
+      "step": 1553
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.648171813910652e-05,
+      "loss": 0.9491,
+      "step": 1554
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.647648658035418e-05,
+      "loss": 1.0315,
+      "step": 1555
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.647125142978347e-05,
+      "loss": 1.0047,
+      "step": 1556
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.646601268826992e-05,
+      "loss": 1.0648,
+      "step": 1557
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.64607703566897e-05,
+      "loss": 1.0301,
+      "step": 1558
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.645552443591953e-05,
+      "loss": 0.9755,
+      "step": 1559
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.645027492683679e-05,
+      "loss": 1.0124,
+      "step": 1560
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.64450218303194e-05,
+      "loss": 1.0394,
+      "step": 1561
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.643976514724594e-05,
+      "loss": 0.8555,
+      "step": 1562
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.643450487849554e-05,
+      "loss": 1.0523,
+      "step": 1563
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.642924102494796e-05,
+      "loss": 1.0864,
+      "step": 1564
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.642397358748354e-05,
+      "loss": 0.9599,
+      "step": 1565
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.6418702566983244e-05,
+      "loss": 1.0393,
+      "step": 1566
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.641342796432859e-05,
+      "loss": 1.0588,
+      "step": 1567
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.6408149780401766e-05,
+      "loss": 0.9993,
+      "step": 1568
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.640286801608548e-05,
+      "loss": 0.9457,
+      "step": 1569
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.63975826722631e-05,
+      "loss": 1.0866,
+      "step": 1570
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.639229374981856e-05,
+      "loss": 0.9941,
+      "step": 1571
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.638700124963641e-05,
+      "loss": 1.1592,
+      "step": 1572
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.638170517260178e-05,
+      "loss": 0.9347,
+      "step": 1573
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.63764055196004e-05,
+      "loss": 1.1162,
+      "step": 1574
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.637110229151863e-05,
+      "loss": 0.9072,
+      "step": 1575
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.636579548924339e-05,
+      "loss": 0.9627,
+      "step": 1576
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.6360485113662216e-05,
+      "loss": 0.9894,
+      "step": 1577
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.635517116566323e-05,
+      "loss": 1.0857,
+      "step": 1578
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.634985364613518e-05,
+      "loss": 1.0719,
+      "step": 1579
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.6344532555967365e-05,
+      "loss": 1.2283,
+      "step": 1580
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.633920789604972e-05,
+      "loss": 0.8812,
+      "step": 1581
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.633387966727276e-05,
+      "loss": 1.0882,
+      "step": 1582
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.632854787052759e-05,
+      "loss": 1.0044,
+      "step": 1583
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.6323212506705955e-05,
+      "loss": 1.0492,
+      "step": 1584
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.631787357670013e-05,
+      "loss": 0.9388,
+      "step": 1585
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.6312531081403036e-05,
+      "loss": 0.9195,
+      "step": 1586
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.630718502170817e-05,
+      "loss": 1.0637,
+      "step": 1587
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.6301835398509635e-05,
+      "loss": 0.9471,
+      "step": 1588
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.629648221270212e-05,
+      "loss": 1.0243,
+      "step": 1589
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.629112546518091e-05,
+      "loss": 1.0816,
+      "step": 1590
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.62857651568419e-05,
+      "loss": 1.0342,
+      "step": 1591
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.6280401288581565e-05,
+      "loss": 0.9733,
+      "step": 1592
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.627503386129698e-05,
+      "loss": 1.0089,
+      "step": 1593
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.626966287588581e-05,
+      "loss": 1.0702,
+      "step": 1594
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.626428833324634e-05,
+      "loss": 1.0446,
+      "step": 1595
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.625891023427742e-05,
+      "loss": 0.9803,
+      "step": 1596
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.62535285798785e-05,
+      "loss": 1.036,
+      "step": 1597
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.6248143370949636e-05,
+      "loss": 0.9928,
+      "step": 1598
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.624275460839148e-05,
+      "loss": 0.9934,
+      "step": 1599
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.623736229310527e-05,
+      "loss": 1.0465,
+      "step": 1600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.6231966425992826e-05,
+      "loss": 1.0697,
+      "step": 1601
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.622656700795659e-05,
+      "loss": 1.1755,
+      "step": 1602
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.622116403989958e-05,
+      "loss": 1.0018,
+      "step": 1603
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.62157575227254e-05,
+      "loss": 1.0457,
+      "step": 1604
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.621034745733828e-05,
+      "loss": 1.0439,
+      "step": 1605
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.6204933844643e-05,
+      "loss": 1.0403,
+      "step": 1606
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.619951668554497e-05,
+      "loss": 0.9987,
+      "step": 1607
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.619409598095018e-05,
+      "loss": 0.9967,
+      "step": 1608
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.61886717317652e-05,
+      "loss": 0.9577,
+      "step": 1609
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.618324393889721e-05,
+      "loss": 1.062,
+      "step": 1610
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.617781260325398e-05,
+      "loss": 0.9268,
+      "step": 1611
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.617237772574386e-05,
+      "loss": 0.9411,
+      "step": 1612
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.616693930727582e-05,
+      "loss": 1.2081,
+      "step": 1613
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.6161497348759386e-05,
+      "loss": 1.1915,
+      "step": 1614
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.61560518511047e-05,
+      "loss": 1.1637,
+      "step": 1615
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.61506028152225e-05,
+      "loss": 0.9762,
+      "step": 1616
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.614515024202408e-05,
+      "loss": 1.083,
+      "step": 1617
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.613969413242138e-05,
+      "loss": 0.9561,
+      "step": 1618
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.61342344873269e-05,
+      "loss": 1.2096,
+      "step": 1619
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.612877130765371e-05,
+      "loss": 0.9829,
+      "step": 1620
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.612330459431552e-05,
+      "loss": 1.0052,
+      "step": 1621
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.611783434822658e-05,
+      "loss": 1.0777,
+      "step": 1622
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.6112360570301794e-05,
+      "loss": 0.9773,
+      "step": 1623
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.610688326145659e-05,
+      "loss": 1.1043,
+      "step": 1624
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.610140242260703e-05,
+      "loss": 1.1077,
+      "step": 1625
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.6095918054669745e-05,
+      "loss": 0.9311,
+      "step": 1626
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.609043015856197e-05,
+      "loss": 0.9375,
+      "step": 1627
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.608493873520152e-05,
+      "loss": 1.0933,
+      "step": 1628
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.607944378550681e-05,
+      "loss": 0.9508,
+      "step": 1629
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.607394531039682e-05,
+      "loss": 0.9675,
+      "step": 1630
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.6068443310791166e-05,
+      "loss": 1.0797,
+      "step": 1631
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.6062937787610005e-05,
+      "loss": 0.8921,
+      "step": 1632
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.605742874177411e-05,
+      "loss": 0.9831,
+      "step": 1633
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.605191617420484e-05,
+      "loss": 0.8817,
+      "step": 1634
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.6046400085824136e-05,
+      "loss": 0.8432,
+      "step": 1635
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.604088047755454e-05,
+      "loss": 1.2064,
+      "step": 1636
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.603535735031916e-05,
+      "loss": 1.0536,
+      "step": 1637
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.6029830705041724e-05,
+      "loss": 0.8865,
+      "step": 1638
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.6024300542646514e-05,
+      "loss": 0.9038,
+      "step": 1639
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.6018766864058426e-05,
+      "loss": 0.9841,
+      "step": 1640
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.601322967020294e-05,
+      "loss": 1.117,
+      "step": 1641
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.600768896200612e-05,
+      "loss": 1.1086,
+      "step": 1642
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.6002144740394615e-05,
+      "loss": 0.9533,
+      "step": 1643
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.5996597006295655e-05,
+      "loss": 0.998,
+      "step": 1644
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.5991045760637074e-05,
+      "loss": 1.0845,
+      "step": 1645
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.598549100434729e-05,
+      "loss": 1.0007,
+      "step": 1646
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5979932738355304e-05,
+      "loss": 0.9234,
+      "step": 1647
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5974370963590694e-05,
+      "loss": 0.8997,
+      "step": 1648
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.596880568098364e-05,
+      "loss": 1.1907,
+      "step": 1649
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.5963236891464913e-05,
+      "loss": 1.0824,
+      "step": 1650
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.5957664595965846e-05,
+      "loss": 1.0457,
+      "step": 1651
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.5952088795418375e-05,
+      "loss": 1.0141,
+      "step": 1652
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.594650949075503e-05,
+      "loss": 1.1884,
+      "step": 1653
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.5940926682908906e-05,
+      "loss": 1.028,
+      "step": 1654
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.593534037281371e-05,
+      "loss": 1.025,
+      "step": 1655
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.5929750561403705e-05,
+      "loss": 1.1187,
+      "step": 1656
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.5924157249613764e-05,
+      "loss": 1.2872,
+      "step": 1657
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.5918560438379335e-05,
+      "loss": 0.7991,
+      "step": 1658
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.591296012863644e-05,
+      "loss": 1.0954,
+      "step": 1659
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.590735632132172e-05,
+      "loss": 1.0632,
+      "step": 1660
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.590174901737236e-05,
+      "loss": 0.8886,
+      "step": 1661
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.589613821772617e-05,
+      "loss": 0.896,
+      "step": 1662
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.5890523923321505e-05,
+      "loss": 1.1347,
+      "step": 1663
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.5884906135097325e-05,
+      "loss": 1.0818,
+      "step": 1664
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.587928485399319e-05,
+      "loss": 1.1205,
+      "step": 1665
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.58736600809492e-05,
+      "loss": 1.1043,
+      "step": 1666
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.586803181690609e-05,
+      "loss": 1.0571,
+      "step": 1667
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.5862400062805145e-05,
+      "loss": 1.0962,
+      "step": 1668
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.585676481958824e-05,
+      "loss": 1.0765,
+      "step": 1669
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.585112608819784e-05,
+      "loss": 1.0384,
+      "step": 1670
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.584548386957699e-05,
+      "loss": 1.1319,
+      "step": 1671
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.5839838164669316e-05,
+      "loss": 1.0276,
+      "step": 1672
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.583418897441904e-05,
+      "loss": 0.8623,
+      "step": 1673
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.582853629977094e-05,
+      "loss": 1.1704,
+      "step": 1674
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.582288014167041e-05,
+      "loss": 1.1645,
+      "step": 1675
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.5817220501063404e-05,
+      "loss": 0.9435,
+      "step": 1676
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.5811557378896454e-05,
+      "loss": 1.038,
+      "step": 1677
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.580589077611669e-05,
+      "loss": 1.105,
+      "step": 1678
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.580022069367183e-05,
+      "loss": 1.1464,
+      "step": 1679
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.579454713251015e-05,
+      "loss": 0.9403,
+      "step": 1680
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.578887009358053e-05,
+      "loss": 1.0316,
+      "step": 1681
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.578318957783241e-05,
+      "loss": 1.2689,
+      "step": 1682
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.577750558621583e-05,
+      "loss": 1.0983,
+      "step": 1683
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.5771818119681395e-05,
+      "loss": 1.0725,
+      "step": 1684
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.576612717918032e-05,
+      "loss": 1.0117,
+      "step": 1685
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.576043276566437e-05,
+      "loss": 0.8661,
+      "step": 1686
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.575473488008591e-05,
+      "loss": 0.8693,
+      "step": 1687
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.574903352339787e-05,
+      "loss": 1.1221,
+      "step": 1688
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.574332869655378e-05,
+      "loss": 1.0544,
+      "step": 1689
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.573762040050772e-05,
+      "loss": 1.0459,
+      "step": 1690
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.5731908636214385e-05,
+      "loss": 1.0625,
+      "step": 1691
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.572619340462903e-05,
+      "loss": 0.8806,
+      "step": 1692
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.572047470670749e-05,
+      "loss": 1.1524,
+      "step": 1693
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.57147525434062e-05,
+      "loss": 0.8975,
+      "step": 1694
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.570902691568215e-05,
+      "loss": 0.9132,
+      "step": 1695
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.570329782449291e-05,
+      "loss": 0.9592,
+      "step": 1696
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.569756527079664e-05,
+      "loss": 0.9896,
+      "step": 1697
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.569182925555209e-05,
+      "loss": 1.0241,
+      "step": 1698
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.568608977971857e-05,
+      "loss": 1.0981,
+      "step": 1699
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.568034684425595e-05,
+      "loss": 0.9934,
+      "step": 1700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.567460045012474e-05,
+      "loss": 0.9825,
+      "step": 1701
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.566885059828597e-05,
+      "loss": 0.9224,
+      "step": 1702
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.5663097289701265e-05,
+      "loss": 0.8818,
+      "step": 1703
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.565734052533285e-05,
+      "loss": 0.9475,
+      "step": 1704
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.5651580306143505e-05,
+      "loss": 1.0814,
+      "step": 1705
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.564581663309658e-05,
+      "loss": 1.071,
+      "step": 1706
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.5640049507156035e-05,
+      "loss": 0.8966,
+      "step": 1707
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.5634278929286376e-05,
+      "loss": 1.1917,
+      "step": 1708
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.562850490045271e-05,
+      "loss": 0.9214,
+      "step": 1709
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.562272742162069e-05,
+      "loss": 1.1835,
+      "step": 1710
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.5616946493756585e-05,
+      "loss": 1.1567,
+      "step": 1711
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.5611162117827213e-05,
+      "loss": 1.0497,
+      "step": 1712
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.560537429479998e-05,
+      "loss": 1.1596,
+      "step": 1713
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.5599583025642875e-05,
+      "loss": 0.8334,
+      "step": 1714
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.5593788311324437e-05,
+      "loss": 1.1051,
+      "step": 1715
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.558799015281381e-05,
+      "loss": 0.9444,
+      "step": 1716
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.558799015281381e-05,
+      "loss": 0.9599,
+      "step": 1717
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.55821885510807e-05,
+      "loss": 1.1053,
+      "step": 1718
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.557638350709539e-05,
+      "loss": 0.9987,
+      "step": 1719
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.5570575021828736e-05,
+      "loss": 0.8446,
+      "step": 1720
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.556476309625218e-05,
+      "loss": 1.0011,
+      "step": 1721
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.555894773133774e-05,
+      "loss": 1.1697,
+      "step": 1722
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.5553128928057984e-05,
+      "loss": 1.0111,
+      "step": 1723
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.554730668738608e-05,
+      "loss": 1.01,
+      "step": 1724
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.554148101029578e-05,
+      "loss": 1.0552,
+      "step": 1725
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.553565189776137e-05,
+      "loss": 1.0278,
+      "step": 1726
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.552981935075775e-05,
+      "loss": 1.1728,
+      "step": 1727
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.552398337026038e-05,
+      "loss": 0.9908,
+      "step": 1728
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.551814395724529e-05,
+      "loss": 1.0857,
+      "step": 1729
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.551230111268909e-05,
+      "loss": 1.0018,
+      "step": 1730
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.550645483756896e-05,
+      "loss": 0.9041,
+      "step": 1731
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.5500605132862664e-05,
+      "loss": 1.1267,
+      "step": 1732
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.549475199954851e-05,
+      "loss": 0.9196,
+      "step": 1733
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.548889543860543e-05,
+      "loss": 1.0953,
+      "step": 1734
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.548303545101287e-05,
+      "loss": 1.0087,
+      "step": 1735
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.5477172037750906e-05,
+      "loss": 1.2129,
+      "step": 1736
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.547130519980014e-05,
+      "loss": 0.9201,
+      "step": 1737
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.546543493814178e-05,
+      "loss": 1.0686,
+      "step": 1738
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.5459561253757594e-05,
+      "loss": 1.1143,
+      "step": 1739
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.5453684147629905e-05,
+      "loss": 1.0649,
+      "step": 1740
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.544780362074165e-05,
+      "loss": 1.0227,
+      "step": 1741
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.544191967407628e-05,
+      "loss": 1.0397,
+      "step": 1742
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.543603230861789e-05,
+      "loss": 1.1357,
+      "step": 1743
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.543014152535108e-05,
+      "loss": 1.0493,
+      "step": 1744
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.542424732526105e-05,
+      "loss": 1.0339,
+      "step": 1745
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.541834970933359e-05,
+      "loss": 0.9943,
+      "step": 1746
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.541244867855503e-05,
+      "loss": 1.1035,
+      "step": 1747
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.5406544233912275e-05,
+      "loss": 1.071,
+      "step": 1748
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.5400636376392824e-05,
+      "loss": 1.2484,
+      "step": 1749
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.5394725106984736e-05,
+      "loss": 0.9157,
+      "step": 1750
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.5388810426676614e-05,
+      "loss": 0.8682,
+      "step": 1751
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.538289233645767e-05,
+      "loss": 1.0336,
+      "step": 1752
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.5376970837317677e-05,
+      "loss": 0.8258,
+      "step": 1753
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.5371045930246954e-05,
+      "loss": 1.0163,
+      "step": 1754
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.536511761623642e-05,
+      "loss": 0.9519,
+      "step": 1755
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.535918589627754e-05,
+      "loss": 1.0026,
+      "step": 1756
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.535325077136238e-05,
+      "loss": 0.9246,
+      "step": 1757
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.534731224248353e-05,
+      "loss": 1.092,
+      "step": 1758
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.534137031063419e-05,
+      "loss": 0.9413,
+      "step": 1759
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.533542497680812e-05,
+      "loss": 1.0753,
+      "step": 1760
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.5329476241999625e-05,
+      "loss": 0.9246,
+      "step": 1761
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.5323524107203607e-05,
+      "loss": 1.0964,
+      "step": 1762
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.531756857341552e-05,
+      "loss": 1.1197,
+      "step": 1763
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.53116096416314e-05,
+      "loss": 0.9978,
+      "step": 1764
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.530564731284784e-05,
+      "loss": 0.9576,
+      "step": 1765
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.529968158806201e-05,
+      "loss": 1.0069,
+      "step": 1766
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.5293712468271644e-05,
+      "loss": 1.1162,
+      "step": 1767
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.528773995447503e-05,
+      "loss": 1.0483,
+      "step": 1768
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.528176404767105e-05,
+      "loss": 0.8916,
+      "step": 1769
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.527578474885913e-05,
+      "loss": 1.0478,
+      "step": 1770
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.526980205903928e-05,
+      "loss": 0.8489,
+      "step": 1771
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.5263815979212066e-05,
+      "loss": 1.2499,
+      "step": 1772
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.525782651037863e-05,
+      "loss": 1.041,
+      "step": 1773
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.525183365354067e-05,
+      "loss": 1.0959,
+      "step": 1774
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.524583740970045e-05,
+      "loss": 1.1331,
+      "step": 1775
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.523983777986084e-05,
+      "loss": 0.9978,
+      "step": 1776
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.52338347650252e-05,
+      "loss": 1.069,
+      "step": 1777
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.5227828366197535e-05,
+      "loss": 0.94,
+      "step": 1778
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.522181858438236e-05,
+      "loss": 1.0017,
+      "step": 1779
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.521580542058478e-05,
+      "loss": 1.024,
+      "step": 1780
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.520978887581046e-05,
+      "loss": 1.108,
+      "step": 1781
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.520376895106564e-05,
+      "loss": 1.0203,
+      "step": 1782
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.519774564735711e-05,
+      "loss": 1.0646,
+      "step": 1783
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.519171896569225e-05,
+      "loss": 0.9132,
+      "step": 1784
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.518568890707896e-05,
+      "loss": 0.9828,
+      "step": 1785
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.517965547252575e-05,
+      "loss": 1.1144,
+      "step": 1786
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.5173618663041684e-05,
+      "loss": 1.0539,
+      "step": 1787
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.516757847963636e-05,
+      "loss": 1.0402,
+      "step": 1788
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.516153492331999e-05,
+      "loss": 0.9799,
+      "step": 1789
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.515548799510331e-05,
+      "loss": 1.0005,
+      "step": 1790
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.514943769599763e-05,
+      "loss": 0.9751,
+      "step": 1791
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.514338402701483e-05,
+      "loss": 0.989,
+      "step": 1792
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5137326989167356e-05,
+      "loss": 1.0221,
+      "step": 1793
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.513126658346821e-05,
+      "loss": 1.1054,
+      "step": 1794
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.512520281093096e-05,
+      "loss": 1.1439,
+      "step": 1795
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5119135672569736e-05,
+      "loss": 0.8284,
+      "step": 1796
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5113065169399235e-05,
+      "loss": 0.8295,
+      "step": 1797
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5106991302434715e-05,
+      "loss": 0.9679,
+      "step": 1798
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5100914072691994e-05,
+      "loss": 1.0527,
+      "step": 1799
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.5094833481187434e-05,
+      "loss": 1.0758,
+      "step": 1800
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.508874952893802e-05,
+      "loss": 1.0212,
+      "step": 1801
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.508266221696122e-05,
+      "loss": 1.0293,
+      "step": 1802
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.507657154627512e-05,
+      "loss": 1.0767,
+      "step": 1803
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.507047751789834e-05,
+      "loss": 0.8789,
+      "step": 1804
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.5064380132850084e-05,
+      "loss": 0.988,
+      "step": 1805
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.5058279392150096e-05,
+      "loss": 1.1527,
+      "step": 1806
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.505217529681869e-05,
+      "loss": 1.0662,
+      "step": 1807
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.504606784787674e-05,
+      "loss": 1.054,
+      "step": 1808
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.503995704634568e-05,
+      "loss": 1.0982,
+      "step": 1809
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.503384289324751e-05,
+      "loss": 1.0274,
+      "step": 1810
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.502772538960479e-05,
+      "loss": 1.0273,
+      "step": 1811
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.502160453644062e-05,
+      "loss": 0.9737,
+      "step": 1812
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.501548033477871e-05,
+      "loss": 0.9506,
+      "step": 1813
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.5009352785643264e-05,
+      "loss": 1.0444,
+      "step": 1814
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.50032218900591e-05,
+      "loss": 1.0198,
+      "step": 1815
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.499708764905157e-05,
+      "loss": 0.831,
+      "step": 1816
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.4990950063646584e-05,
+      "loss": 1.1305,
+      "step": 1817
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.498480913487062e-05,
+      "loss": 1.0273,
+      "step": 1818
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.497866486375072e-05,
+      "loss": 1.1244,
+      "step": 1819
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.4972517251314476e-05,
+      "loss": 0.91,
+      "step": 1820
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.496636629859002e-05,
+      "loss": 1.0845,
+      "step": 1821
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.4960212006606094e-05,
+      "loss": 0.9611,
+      "step": 1822
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.4954054376391956e-05,
+      "loss": 0.8568,
+      "step": 1823
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.494789340897743e-05,
+      "loss": 1.0131,
+      "step": 1824
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.4941729105392904e-05,
+      "loss": 1.1537,
+      "step": 1825
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.493556146666933e-05,
+      "loss": 0.9255,
+      "step": 1826
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.4929390493838196e-05,
+      "loss": 1.0474,
+      "step": 1827
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.4923216187931574e-05,
+      "loss": 1.0458,
+      "step": 1828
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.491703854998207e-05,
+      "loss": 1.1119,
+      "step": 1829
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.491085758102286e-05,
+      "loss": 1.0646,
+      "step": 1830
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.4904673282087684e-05,
+      "loss": 0.8963,
+      "step": 1831
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.489848565421083e-05,
+      "loss": 0.9823,
+      "step": 1832
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.489229469842714e-05,
+      "loss": 1.1175,
+      "step": 1833
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.4886100415772e-05,
+      "loss": 1.1223,
+      "step": 1834
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.4879902807281395e-05,
+      "loss": 0.9941,
+      "step": 1835
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.4873701873991824e-05,
+      "loss": 1.0793,
+      "step": 1836
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.4867497616940366e-05,
+      "loss": 1.051,
+      "step": 1837
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.486129003716463e-05,
+      "loss": 0.8835,
+      "step": 1838
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.4855079135702816e-05,
+      "loss": 0.8274,
+      "step": 1839
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.484886491359365e-05,
+      "loss": 1.0766,
+      "step": 1840
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.484264737187643e-05,
+      "loss": 1.0281,
+      "step": 1841
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.483642651159101e-05,
+      "loss": 0.9128,
+      "step": 1842
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.483020233377777e-05,
+      "loss": 1.002,
+      "step": 1843
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.48239748394777e-05,
+      "loss": 0.9773,
+      "step": 1844
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.481774402973229e-05,
+      "loss": 0.8487,
+      "step": 1845
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.481150990558361e-05,
+      "loss": 1.0105,
+      "step": 1846
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.480527246807429e-05,
+      "loss": 1.1119,
+      "step": 1847
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.47990317182475e-05,
+      "loss": 0.9716,
+      "step": 1848
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.4792787657146965e-05,
+      "loss": 1.0255,
+      "step": 1849
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.478654028581697e-05,
+      "loss": 1.1118,
+      "step": 1850
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.478028960530236e-05,
+      "loss": 0.9287,
+      "step": 1851
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.477403561664852e-05,
+      "loss": 1.0675,
+      "step": 1852
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.476777832090139e-05,
+      "loss": 0.9325,
+      "step": 1853
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.4761517719107474e-05,
+      "loss": 1.0169,
+      "step": 1854
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.4755253812313805e-05,
+      "loss": 1.0541,
+      "step": 1855
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.474898660156801e-05,
+      "loss": 0.9254,
+      "step": 1856
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.474271608791822e-05,
+      "loss": 0.9559,
+      "step": 1857
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.4736442272413164e-05,
+      "loss": 0.861,
+      "step": 1858
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.4730165156102085e-05,
+      "loss": 0.9457,
+      "step": 1859
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.472388474003481e-05,
+      "loss": 0.9162,
+      "step": 1860
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.4717601025261676e-05,
+      "loss": 0.9348,
+      "step": 1861
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.4711314012833625e-05,
+      "loss": 0.9985,
+      "step": 1862
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.4705023703802104e-05,
+      "loss": 1.0837,
+      "step": 1863
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.469873009921916e-05,
+      "loss": 1.006,
+      "step": 1864
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.469243320013732e-05,
+      "loss": 0.8869,
+      "step": 1865
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.468613300760973e-05,
+      "loss": 1.1028,
+      "step": 1866
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4679829522690064e-05,
+      "loss": 0.9552,
+      "step": 1867
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.467352274643254e-05,
+      "loss": 1.0632,
+      "step": 1868
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.466721267989191e-05,
+      "loss": 1.0639,
+      "step": 1869
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.466089932412353e-05,
+      "loss": 0.9969,
+      "step": 1870
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.465458268018324e-05,
+      "loss": 0.9895,
+      "step": 1871
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.464826274912749e-05,
+      "loss": 1.0921,
+      "step": 1872
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.464193953201323e-05,
+      "loss": 1.0225,
+      "step": 1873
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.463561302989799e-05,
+      "loss": 0.9572,
+      "step": 1874
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.462928324383985e-05,
+      "loss": 0.8776,
+      "step": 1875
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.4622950174897416e-05,
+      "loss": 1.0471,
+      "step": 1876
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.461661382412986e-05,
+      "loss": 1.0317,
+      "step": 1877
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.46102741925969e-05,
+      "loss": 0.9692,
+      "step": 1878
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.460393128135881e-05,
+      "loss": 0.8569,
+      "step": 1879
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.459758509147641e-05,
+      "loss": 0.7915,
+      "step": 1880
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.459123562401104e-05,
+      "loss": 1.1659,
+      "step": 1881
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.4584882880024635e-05,
+      "loss": 0.9317,
+      "step": 1882
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.457852686057964e-05,
+      "loss": 0.9966,
+      "step": 1883
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.4572167566739075e-05,
+      "loss": 1.044,
+      "step": 1884
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.456580499956649e-05,
+      "loss": 0.923,
+      "step": 1885
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.4559439160125974e-05,
+      "loss": 1.0478,
+      "step": 1886
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.4553070049482204e-05,
+      "loss": 0.9384,
+      "step": 1887
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.4546697668700355e-05,
+      "loss": 0.7466,
+      "step": 1888
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.4540322018846176e-05,
+      "loss": 1.1552,
+      "step": 1889
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.4533943100985967e-05,
+      "loss": 1.0958,
+      "step": 1890
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.452756091618655e-05,
+      "loss": 1.0544,
+      "step": 1891
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.4521175465515316e-05,
+      "loss": 1.2076,
+      "step": 1892
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.4514786750040205e-05,
+      "loss": 1.0989,
+      "step": 1893
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.4508394770829675e-05,
+      "loss": 0.9365,
+      "step": 1894
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.4501999528952755e-05,
+      "loss": 0.993,
+      "step": 1895
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.449560102547902e-05,
+      "loss": 1.0149,
+      "step": 1896
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.448919926147857e-05,
+      "loss": 1.0702,
+      "step": 1897
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.448279423802207e-05,
+      "loss": 1.18,
+      "step": 1898
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.4476385956180724e-05,
+      "loss": 0.9172,
+      "step": 1899
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.446997441702627e-05,
+      "loss": 0.9624,
+      "step": 1900
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.4463559621631016e-05,
+      "loss": 0.9424,
+      "step": 1901
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.4457141571067794e-05,
+      "loss": 1.0255,
+      "step": 1902
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.445072026640999e-05,
+      "loss": 0.9884,
+      "step": 1903
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.444429570873151e-05,
+      "loss": 1.0615,
+      "step": 1904
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.4437867899106845e-05,
+      "loss": 1.0013,
+      "step": 1905
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.4431436838611e-05,
+      "loss": 1.0561,
+      "step": 1906
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.442500252831954e-05,
+      "loss": 0.9397,
+      "step": 1907
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.441856496930856e-05,
+      "loss": 0.8876,
+      "step": 1908
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.4412124162654714e-05,
+      "loss": 0.7322,
+      "step": 1909
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.4405680109435165e-05,
+      "loss": 1.0714,
+      "step": 1910
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.439923281072767e-05,
+      "loss": 1.0364,
+      "step": 1911
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.43927822676105e-05,
+      "loss": 1.1931,
+      "step": 1912
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.4386328481162455e-05,
+      "loss": 1.1539,
+      "step": 1913
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.437987145246291e-05,
+      "loss": 1.1141,
+      "step": 1914
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.437341118259175e-05,
+      "loss": 1.0845,
+      "step": 1915
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.436694767262943e-05,
+      "loss": 1.1455,
+      "step": 1916
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.436048092365693e-05,
+      "loss": 0.9319,
+      "step": 1917
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.435401093675578e-05,
+      "loss": 1.0054,
+      "step": 1918
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4347537713008045e-05,
+      "loss": 0.7604,
+      "step": 1919
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.434106125349634e-05,
+      "loss": 0.8996,
+      "step": 1920
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.43345815593038e-05,
+      "loss": 1.0908,
+      "step": 1921
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4328098631514136e-05,
+      "loss": 1.0009,
+      "step": 1922
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.432161247121156e-05,
+      "loss": 1.0158,
+      "step": 1923
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.431512307948087e-05,
+      "loss": 0.9205,
+      "step": 1924
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.430863045740736e-05,
+      "loss": 1.1482,
+      "step": 1925
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.4302134606076885e-05,
+      "loss": 0.9957,
+      "step": 1926
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.429563552657584e-05,
+      "loss": 0.9747,
+      "step": 1927
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.4289133219991166e-05,
+      "loss": 0.9724,
+      "step": 1928
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.428262768741033e-05,
+      "loss": 1.0864,
+      "step": 1929
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.4276118929921346e-05,
+      "loss": 1.0434,
+      "step": 1930
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.426960694861277e-05,
+      "loss": 1.0159,
+      "step": 1931
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.426309174457368e-05,
+      "loss": 1.0123,
+      "step": 1932
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.425657331889372e-05,
+      "loss": 0.9664,
+      "step": 1933
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.4250051672663054e-05,
+      "loss": 1.0955,
+      "step": 1934
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.424352680697239e-05,
+      "loss": 0.9491,
+      "step": 1935
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.423699872291298e-05,
+      "loss": 1.0209,
+      "step": 1936
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.423046742157659e-05,
+      "loss": 1.048,
+      "step": 1937
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.422393290405557e-05,
+      "loss": 0.9957,
+      "step": 1938
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.421739517144276e-05,
+      "loss": 0.9557,
+      "step": 1939
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.421085422483157e-05,
+      "loss": 1.0181,
+      "step": 1940
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.420431006531591e-05,
+      "loss": 1.0145,
+      "step": 1941
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.4197762693990284e-05,
+      "loss": 1.023,
+      "step": 1942
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.419121211194969e-05,
+      "loss": 1.1339,
+      "step": 1943
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.4184658320289675e-05,
+      "loss": 0.9017,
+      "step": 1944
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.417810132010633e-05,
+      "loss": 0.9422,
+      "step": 1945
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.417154111249625e-05,
+      "loss": 0.9912,
+      "step": 1946
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.416497769855662e-05,
+      "loss": 1.0001,
+      "step": 1947
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.415841107938513e-05,
+      "loss": 1.0772,
+      "step": 1948
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.415184125607999e-05,
+      "loss": 1.0648,
+      "step": 1949
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.414526822973999e-05,
+      "loss": 0.9032,
+      "step": 1950
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.413869200146441e-05,
+      "loss": 0.9825,
+      "step": 1951
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.4132112572353095e-05,
+      "loss": 0.9469,
+      "step": 1952
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.412552994350642e-05,
+      "loss": 1.0209,
+      "step": 1953
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.411894411602528e-05,
+      "loss": 0.9763,
+      "step": 1954
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.411235509101113e-05,
+      "loss": 0.9222,
+      "step": 1955
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.410576286956594e-05,
+      "loss": 0.9547,
+      "step": 1956
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.409916745279222e-05,
+      "loss": 0.992,
+      "step": 1957
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.4092568841793016e-05,
+      "loss": 0.8066,
+      "step": 1958
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.4085967037671906e-05,
+      "loss": 0.9993,
+      "step": 1959
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.407936204153301e-05,
+      "loss": 1.0403,
+      "step": 1960
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.4072753854480976e-05,
+      "loss": 0.909,
+      "step": 1961
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.406614247762098e-05,
+      "loss": 0.9898,
+      "step": 1962
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.4059527912058726e-05,
+      "loss": 1.0132,
+      "step": 1963
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.405291015890048e-05,
+      "loss": 0.9477,
+      "step": 1964
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.404628921925301e-05,
+      "loss": 0.8879,
+      "step": 1965
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.403966509422365e-05,
+      "loss": 1.0321,
+      "step": 1966
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.403303778492022e-05,
+      "loss": 0.9355,
+      "step": 1967
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.4026407292451124e-05,
+      "loss": 1.1093,
+      "step": 1968
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.4019773617925254e-05,
+      "loss": 0.8684,
+      "step": 1969
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.401313676245206e-05,
+      "loss": 1.0747,
+      "step": 1970
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.400649672714152e-05,
+      "loss": 0.855,
+      "step": 1971
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.3999853513104137e-05,
+      "loss": 0.9003,
+      "step": 1972
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.399320712145096e-05,
+      "loss": 0.9348,
+      "step": 1973
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.398655755329355e-05,
+      "loss": 1.0418,
+      "step": 1974
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.397990480974402e-05,
+      "loss": 1.0164,
+      "step": 1975
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.3973248891914986e-05,
+      "loss": 1.1829,
+      "step": 1976
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.396658980091962e-05,
+      "loss": 0.8416,
+      "step": 1977
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.395992753787163e-05,
+      "loss": 1.0298,
+      "step": 1978
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.395326210388522e-05,
+      "loss": 0.9489,
+      "step": 1979
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.394659350007515e-05,
+      "loss": 0.9784,
+      "step": 1980
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.393992172755672e-05,
+      "loss": 0.8715,
+      "step": 1981
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.3933246787445725e-05,
+      "loss": 0.8659,
+      "step": 1982
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.392656868085853e-05,
+      "loss": 1.0191,
+      "step": 1983
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.391988740891199e-05,
+      "loss": 0.8876,
+      "step": 1984
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.391320297272352e-05,
+      "loss": 0.9732,
+      "step": 1985
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.3906515373411065e-05,
+      "loss": 1.0367,
+      "step": 1986
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.389982461209306e-05,
+      "loss": 0.981,
+      "step": 1987
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.389313068988852e-05,
+      "loss": 0.9307,
+      "step": 1988
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.3886433607916946e-05,
+      "loss": 0.9651,
+      "step": 1989
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.3879733367298405e-05,
+      "loss": 1.0248,
+      "step": 1990
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.3873029969153466e-05,
+      "loss": 0.9942,
+      "step": 1991
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.386632341460323e-05,
+      "loss": 0.9261,
+      "step": 1992
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.3859613704769324e-05,
+      "loss": 0.9873,
+      "step": 1993
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.3852900840773924e-05,
+      "loss": 0.94,
+      "step": 1994
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.3846184823739714e-05,
+      "loss": 0.9446,
+      "step": 1995
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.383946565478989e-05,
+      "loss": 0.9954,
+      "step": 1996
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.383274333504822e-05,
+      "loss": 1.0165,
+      "step": 1997
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.3826017865638964e-05,
+      "loss": 1.1284,
+      "step": 1998
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.3819289247686914e-05,
+      "loss": 0.96,
+      "step": 1999
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.3812557482317396e-05,
+      "loss": 0.9836,
+      "step": 2000
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.3805822570656254e-05,
+      "loss": 0.9974,
+      "step": 2001
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.3799084513829866e-05,
+      "loss": 0.9876,
+      "step": 2002
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.379234331296514e-05,
+      "loss": 0.948,
+      "step": 2003
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.378559896918949e-05,
+      "loss": 0.9507,
+      "step": 2004
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.377885148363087e-05,
+      "loss": 0.917,
+      "step": 2005
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.3772100857417775e-05,
+      "loss": 1.012,
+      "step": 2006
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.3765347091679195e-05,
+      "loss": 0.8763,
+      "step": 2007
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.375859018754466e-05,
+      "loss": 0.895,
+      "step": 2008
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.375183014614421e-05,
+      "loss": 1.1534,
+      "step": 2009
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.374506696860845e-05,
+      "loss": 1.1528,
+      "step": 2010
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.373830065606846e-05,
+      "loss": 1.1235,
+      "step": 2011
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.3731531209655885e-05,
+      "loss": 0.9277,
+      "step": 2012
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.372475863050286e-05,
+      "loss": 1.0293,
+      "step": 2013
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.371798291974206e-05,
+      "loss": 0.9068,
+      "step": 2014
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.37112040785067e-05,
+      "loss": 1.165,
+      "step": 2015
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.3704422107930485e-05,
+      "loss": 0.9346,
+      "step": 2016
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.369763700914767e-05,
+      "loss": 0.9439,
+      "step": 2017
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.3690848783293016e-05,
+      "loss": 1.0294,
+      "step": 2018
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.3684057431501826e-05,
+      "loss": 0.9282,
+      "step": 2019
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.3677262954909903e-05,
+      "loss": 1.056,
+      "step": 2020
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.367046535465358e-05,
+      "loss": 1.0573,
+      "step": 2021
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.366366463186974e-05,
+      "loss": 0.8866,
+      "step": 2022
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.365686078769574e-05,
+      "loss": 0.8818,
+      "step": 2023
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.365005382326949e-05,
+      "loss": 1.0418,
+      "step": 2024
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.3643243739729424e-05,
+      "loss": 0.9096,
+      "step": 2025
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.363643053821448e-05,
+      "loss": 0.9047,
+      "step": 2026
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.362961421986413e-05,
+      "loss": 1.0349,
+      "step": 2027
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.3622794785818366e-05,
+      "loss": 0.8385,
+      "step": 2028
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.3615972237217696e-05,
+      "loss": 0.9425,
+      "step": 2029
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.3609146575203156e-05,
+      "loss": 0.8219,
+      "step": 2030
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.3602317800916293e-05,
+      "loss": 0.7941,
+      "step": 2031
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.359548591549918e-05,
+      "loss": 1.1694,
+      "step": 2032
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.358865092009441e-05,
+      "loss": 1.0006,
+      "step": 2033
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.3581812815845116e-05,
+      "loss": 0.8398,
+      "step": 2034
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.35749716038949e-05,
+      "loss": 0.8459,
+      "step": 2035
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.3568127285387925e-05,
+      "loss": 0.931,
+      "step": 2036
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.3561279861468884e-05,
+      "loss": 1.072,
+      "step": 2037
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.3554429333282945e-05,
+      "loss": 1.0626,
+      "step": 2038
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.354757570197583e-05,
+      "loss": 0.8993,
+      "step": 2039
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.3540718968693764e-05,
+      "loss": 0.9534,
+      "step": 2040
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.353385913458351e-05,
+      "loss": 1.0367,
+      "step": 2041
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.352699620079231e-05,
+      "loss": 0.9366,
+      "step": 2042
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.3520130168467974e-05,
+      "loss": 0.8698,
+      "step": 2043
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.35132610387588e-05,
+      "loss": 0.8566,
+      "step": 2044
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.3506388812813603e-05,
+      "loss": 1.1404,
+      "step": 2045
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.349951349178173e-05,
+      "loss": 1.0298,
+      "step": 2046
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.349263507681304e-05,
+      "loss": 0.9984,
+      "step": 2047
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.34857535690579e-05,
+      "loss": 0.9617,
+      "step": 2048
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.3478868969667205e-05,
+      "loss": 1.1486,
+      "step": 2049
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.3471981279792365e-05,
+      "loss": 0.9825,
+      "step": 2050
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.3465090500585316e-05,
+      "loss": 0.9813,
+      "step": 2051
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.345819663319849e-05,
+      "loss": 1.0731,
+      "step": 2052
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.345129967878485e-05,
+      "loss": 1.2427,
+      "step": 2053
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.3444399638497865e-05,
+      "loss": 0.7396,
+      "step": 2054
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.343749651349154e-05,
+      "loss": 1.0409,
+      "step": 2055
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.343059030492037e-05,
+      "loss": 1.0141,
+      "step": 2056
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.34236810139394e-05,
+      "loss": 0.829,
+      "step": 2057
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.341676864170414e-05,
+      "loss": 0.8438,
+      "step": 2058
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.340985318937066e-05,
+      "loss": 1.085,
+      "step": 2059
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.340293465809553e-05,
+      "loss": 1.0291,
+      "step": 2060
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.3396013049035826e-05,
+      "loss": 1.0693,
+      "step": 2061
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.3389088363349164e-05,
+      "loss": 1.057,
+      "step": 2062
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.338216060219363e-05,
+      "loss": 0.9991,
+      "step": 2063
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.3375229766727885e-05,
+      "loss": 1.0474,
+      "step": 2064
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.336829585811105e-05,
+      "loss": 1.0301,
+      "step": 2065
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.336135887750279e-05,
+      "loss": 0.9866,
+      "step": 2066
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.335441882606326e-05,
+      "loss": 1.0866,
+      "step": 2067
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.334747570495316e-05,
+      "loss": 0.9707,
+      "step": 2068
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.334052951533368e-05,
+      "loss": 0.8108,
+      "step": 2069
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.333358025836654e-05,
+      "loss": 1.1265,
+      "step": 2070
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.332662793521395e-05,
+      "loss": 1.1186,
+      "step": 2071
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.3319672547038646e-05,
+      "loss": 0.8959,
+      "step": 2072
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.3312714095003894e-05,
+      "loss": 0.9816,
+      "step": 2073
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.330575258027343e-05,
+      "loss": 1.0419,
+      "step": 2074
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.3298788004011535e-05,
+      "loss": 1.0947,
+      "step": 2075
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.329182036738301e-05,
+      "loss": 0.8909,
+      "step": 2076
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.328484967155314e-05,
+      "loss": 0.9763,
+      "step": 2077
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.3277875917687726e-05,
+      "loss": 1.2191,
+      "step": 2078
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.3270899106953105e-05,
+      "loss": 1.0513,
+      "step": 2079
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.326391924051609e-05,
+      "loss": 1.0197,
+      "step": 2080
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.325693631954405e-05,
+      "loss": 0.9591,
+      "step": 2081
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.3249950345204806e-05,
+      "loss": 0.8085,
+      "step": 2082
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.324296131866674e-05,
+      "loss": 0.812,
+      "step": 2083
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.3235969241098726e-05,
+      "loss": 1.0724,
+      "step": 2084
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.322897411367014e-05,
+      "loss": 1.007,
+      "step": 2085
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.3221975937550885e-05,
+      "loss": 0.9944,
+      "step": 2086
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.321497471391136e-05,
+      "loss": 1.02,
+      "step": 2087
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.320797044392249e-05,
+      "loss": 0.8257,
+      "step": 2088
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.320096312875569e-05,
+      "loss": 1.1136,
+      "step": 2089
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.319395276958289e-05,
+      "loss": 0.8464,
+      "step": 2090
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.318693936757653e-05,
+      "loss": 0.864,
+      "step": 2091
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.317992292390958e-05,
+      "loss": 0.914,
+      "step": 2092
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.317290343975549e-05,
+      "loss": 0.9462,
+      "step": 2093
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.316588091628821e-05,
+      "loss": 0.9766,
+      "step": 2094
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.3158855354682234e-05,
+      "loss": 1.0446,
+      "step": 2095
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.315182675611255e-05,
+      "loss": 0.9367,
+      "step": 2096
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.3144795121754644e-05,
+      "loss": 0.9258,
+      "step": 2097
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.3137760452784514e-05,
+      "loss": 0.8673,
+      "step": 2098
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.313072275037867e-05,
+      "loss": 0.825,
+      "step": 2099
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.312368201571414e-05,
+      "loss": 0.8916,
+      "step": 2100
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.311663824996842e-05,
+      "loss": 1.0418,
+      "step": 2101
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.310959145431956e-05,
+      "loss": 1.0265,
+      "step": 2102
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.310254162994608e-05,
+      "loss": 0.8426,
+      "step": 2103
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.3095488778027045e-05,
+      "loss": 1.1474,
+      "step": 2104
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.3088432899741985e-05,
+      "loss": 0.8672,
+      "step": 2105
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.3081373996270954e-05,
+      "loss": 1.1356,
+      "step": 2106
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.307431206879452e-05,
+      "loss": 1.1089,
+      "step": 2107
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.306724711849376e-05,
+      "loss": 1.0002,
+      "step": 2108
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.306017914655022e-05,
+      "loss": 1.1167,
+      "step": 2109
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.3053108154146e-05,
+      "loss": 0.7861,
+      "step": 2110
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.304603414246368e-05,
+      "loss": 1.0578,
+      "step": 2111
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.303895711268633e-05,
+      "loss": 0.9026,
+      "step": 2112
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.303187706599756e-05,
+      "loss": 0.9789,
+      "step": 2113
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.3024794003581464e-05,
+      "loss": 1.062,
+      "step": 2114
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.301770792662264e-05,
+      "loss": 0.951,
+      "step": 2115
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.301061883630619e-05,
+      "loss": 0.7825,
+      "step": 2116
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.300352673381773e-05,
+      "loss": 0.9572,
+      "step": 2117
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.299643162034338e-05,
+      "loss": 1.1311,
+      "step": 2118
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.298933349706974e-05,
+      "loss": 0.9564,
+      "step": 2119
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.298223236518394e-05,
+      "loss": 0.9606,
+      "step": 2120
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.2975128225873605e-05,
+      "loss": 1.0066,
+      "step": 2121
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.296802108032685e-05,
+      "loss": 0.9788,
+      "step": 2122
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.296091092973231e-05,
+      "loss": 1.1259,
+      "step": 2123
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.295379777527913e-05,
+      "loss": 0.9375,
+      "step": 2124
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.294668161815693e-05,
+      "loss": 1.0405,
+      "step": 2125
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.2939562459555836e-05,
+      "loss": 0.9528,
+      "step": 2126
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.2932440300666507e-05,
+      "loss": 0.8533,
+      "step": 2127
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.292531514268008e-05,
+      "loss": 1.0818,
+      "step": 2128
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.2918186986788186e-05,
+      "loss": 0.8657,
+      "step": 2129
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.291105583418297e-05,
+      "loss": 1.0435,
+      "step": 2130
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.2903921686057085e-05,
+      "loss": 0.9539,
+      "step": 2131
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.289678454360367e-05,
+      "loss": 1.1706,
+      "step": 2132
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.2889644408016364e-05,
+      "loss": 0.8639,
+      "step": 2133
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.288250128048933e-05,
+      "loss": 1.0177,
+      "step": 2134
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.28753551622172e-05,
+      "loss": 1.0674,
+      "step": 2135
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.286820605439513e-05,
+      "loss": 1.02,
+      "step": 2136
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.2861053958218754e-05,
+      "loss": 0.981,
+      "step": 2137
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.285389887488423e-05,
+      "loss": 0.9802,
+      "step": 2138
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.2846740805588215e-05,
+      "loss": 1.081,
+      "step": 2139
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.283957975152784e-05,
+      "loss": 1.0095,
+      "step": 2140
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.283241571390074e-05,
+      "loss": 0.9899,
+      "step": 2141
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.282524869390509e-05,
+      "loss": 0.9411,
+      "step": 2142
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.2818078692739495e-05,
+      "loss": 1.0677,
+      "step": 2143
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.281090571160312e-05,
+      "loss": 1.0198,
+      "step": 2144
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.280372975169561e-05,
+      "loss": 1.2048,
+      "step": 2145
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.279655081421709e-05,
+      "loss": 0.8644,
+      "step": 2146
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.2789368900368186e-05,
+      "loss": 0.8165,
+      "step": 2147
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.278218401135006e-05,
+      "loss": 0.9792,
+      "step": 2148
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.277499614836432e-05,
+      "loss": 0.7673,
+      "step": 2149
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.2767805312613096e-05,
+      "loss": 0.9629,
+      "step": 2150
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.276061150529903e-05,
+      "loss": 0.9071,
+      "step": 2151
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.275341472762523e-05,
+      "loss": 0.9466,
+      "step": 2152
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.274621498079532e-05,
+      "loss": 0.8757,
+      "step": 2153
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.273901226601342e-05,
+      "loss": 1.0485,
+      "step": 2154
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.2731806584484134e-05,
+      "loss": 0.8866,
+      "step": 2155
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.272459793741258e-05,
+      "loss": 1.0214,
+      "step": 2156
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.271738632600435e-05,
+      "loss": 0.8803,
+      "step": 2157
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.271017175146556e-05,
+      "loss": 1.0529,
+      "step": 2158
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.27029542150028e-05,
+      "loss": 1.0647,
+      "step": 2159
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.269573371782315e-05,
+      "loss": 0.9456,
+      "step": 2160
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.268851026113421e-05,
+      "loss": 0.9077,
+      "step": 2161
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.268128384614406e-05,
+      "loss": 0.9624,
+      "step": 2162
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.2674054474061274e-05,
+      "loss": 1.0584,
+      "step": 2163
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.266682214609492e-05,
+      "loss": 0.9913,
+      "step": 2164
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.265958686345457e-05,
+      "loss": 0.8429,
+      "step": 2165
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.2652348627350266e-05,
+      "loss": 1.0018,
+      "step": 2166
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.264510743899258e-05,
+      "loss": 0.79,
+      "step": 2167
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.2637863299592564e-05,
+      "loss": 1.2138,
+      "step": 2168
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.263061621036174e-05,
+      "loss": 0.9926,
+      "step": 2169
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.262336617251214e-05,
+      "loss": 1.0631,
+      "step": 2170
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.2616113187256314e-05,
+      "loss": 1.089,
+      "step": 2171
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.260885725580726e-05,
+      "loss": 0.9534,
+      "step": 2172
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.26015983793785e-05,
+      "loss": 1.0199,
+      "step": 2173
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.259433655918404e-05,
+      "loss": 0.8924,
+      "step": 2174
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.258707179643837e-05,
+      "loss": 0.9475,
+      "step": 2175
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.257980409235649e-05,
+      "loss": 0.9721,
+      "step": 2176
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.257253344815387e-05,
+      "loss": 1.0619,
+      "step": 2177
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.25652598650465e-05,
+      "loss": 0.9709,
+      "step": 2178
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.255798334425083e-05,
+      "loss": 1.018,
+      "step": 2179
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.255070388698382e-05,
+      "loss": 0.8687,
+      "step": 2180
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.254342149446292e-05,
+      "loss": 0.9337,
+      "step": 2181
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.253613616790606e-05,
+      "loss": 1.0685,
+      "step": 2182
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.252884790853168e-05,
+      "loss": 1.0079,
+      "step": 2183
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.2521556717558685e-05,
+      "loss": 1.0001,
+      "step": 2184
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.2514262596206505e-05,
+      "loss": 0.9272,
+      "step": 2185
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.250696554569502e-05,
+      "loss": 0.9528,
+      "step": 2186
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.249966556724464e-05,
+      "loss": 0.9227,
+      "step": 2187
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.249236266207622e-05,
+      "loss": 0.9462,
+      "step": 2188
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.248505683141114e-05,
+      "loss": 0.9816,
+      "step": 2189
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.2477748076471266e-05,
+      "loss": 1.0493,
+      "step": 2190
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.247043639847894e-05,
+      "loss": 1.093,
+      "step": 2191
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.246312179865699e-05,
+      "loss": 0.77,
+      "step": 2192
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.245580427822875e-05,
+      "loss": 0.7836,
+      "step": 2193
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.244848383841803e-05,
+      "loss": 0.9187,
+      "step": 2194
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.244116048044913e-05,
+      "loss": 1.0049,
+      "step": 2195
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.2433834205546836e-05,
+      "loss": 1.0276,
+      "step": 2196
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.242650501493642e-05,
+      "loss": 0.9692,
+      "step": 2197
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.241917290984367e-05,
+      "loss": 0.9757,
+      "step": 2198
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.241183789149482e-05,
+      "loss": 1.0204,
+      "step": 2199
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.2404499961116624e-05,
+      "loss": 0.824,
+      "step": 2200
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.239715911993629e-05,
+      "loss": 0.9317,
+      "step": 2201
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.238981536918154e-05,
+      "loss": 1.1145,
+      "step": 2202
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.238246871008057e-05,
+      "loss": 1.0172,
+      "step": 2203
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.2375119143862074e-05,
+      "loss": 1.0105,
+      "step": 2204
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.236776667175522e-05,
+      "loss": 1.0563,
+      "step": 2205
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.2360411294989665e-05,
+      "loss": 0.9801,
+      "step": 2206
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.235305301479555e-05,
+      "loss": 0.9731,
+      "step": 2207
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.234569183240352e-05,
+      "loss": 0.9383,
+      "step": 2208
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.233832774904468e-05,
+      "loss": 0.9008,
+      "step": 2209
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.233096076595062e-05,
+      "loss": 0.9989,
+      "step": 2210
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.232359088435345e-05,
+      "loss": 0.965,
+      "step": 2211
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.231621810548572e-05,
+      "loss": 0.7795,
+      "step": 2212
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.230884243058048e-05,
+      "loss": 1.0789,
+      "step": 2213
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.23014638608713e-05,
+      "loss": 0.9722,
+      "step": 2214
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.229408239759217e-05,
+      "loss": 1.0721,
+      "step": 2215
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.2286698041977615e-05,
+      "loss": 0.8586,
+      "step": 2216
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.227931079526262e-05,
+      "loss": 1.0351,
+      "step": 2217
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.227192065868266e-05,
+      "loss": 0.9039,
+      "step": 2218
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.22645276334737e-05,
+      "loss": 0.8089,
+      "step": 2219
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.2257131720872164e-05,
+      "loss": 0.9644,
+      "step": 2220
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.224973292211498e-05,
+      "loss": 1.1008,
+      "step": 2221
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.224233123843958e-05,
+      "loss": 0.8743,
+      "step": 2222
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.2234926671083817e-05,
+      "loss": 0.9993,
+      "step": 2223
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.2227519221286084e-05,
+      "loss": 0.9854,
+      "step": 2224
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.222010889028523e-05,
+      "loss": 1.0744,
+      "step": 2225
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.221269567932058e-05,
+      "loss": 1.0225,
+      "step": 2226
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.2205279589631964e-05,
+      "loss": 0.8465,
+      "step": 2227
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.219786062245967e-05,
+      "loss": 0.9327,
+      "step": 2228
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.219043877904449e-05,
+      "loss": 1.0655,
+      "step": 2229
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.218301406062768e-05,
+      "loss": 1.0715,
+      "step": 2230
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.2175586468450967e-05,
+      "loss": 0.9503,
+      "step": 2231
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.216815600375659e-05,
+      "loss": 1.031,
+      "step": 2232
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.216072266778724e-05,
+      "loss": 0.9954,
+      "step": 2233
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.215328646178612e-05,
+      "loss": 0.8421,
+      "step": 2234
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.2145847386996866e-05,
+      "loss": 0.776,
+      "step": 2235
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.213840544466364e-05,
+      "loss": 1.0404,
+      "step": 2236
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.213096063603105e-05,
+      "loss": 0.9776,
+      "step": 2237
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.21235129623442e-05,
+      "loss": 0.8568,
+      "step": 2238
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.211606242484868e-05,
+      "loss": 0.9484,
+      "step": 2239
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.210860902479054e-05,
+      "loss": 0.9259,
+      "step": 2240
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.210115276341632e-05,
+      "loss": 0.8018,
+      "step": 2241
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.2093693641973034e-05,
+      "loss": 0.9692,
+      "step": 2242
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.2086231661708185e-05,
+      "loss": 1.0646,
+      "step": 2243
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.207876682386974e-05,
+      "loss": 0.924,
+      "step": 2244
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.207129912970616e-05,
+      "loss": 0.9833,
+      "step": 2245
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.206382858046636e-05,
+      "loss": 1.0739,
+      "step": 2246
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.2056355177399744e-05,
+      "loss": 0.8747,
+      "step": 2247
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.204887892175621e-05,
+      "loss": 1.0175,
+      "step": 2248
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.2041399814786116e-05,
+      "loss": 0.8796,
+      "step": 2249
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.203391785774029e-05,
+      "loss": 0.9655,
+      "step": 2250
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.202643305187005e-05,
+      "loss": 1.004,
+      "step": 2251
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.201894539842719e-05,
+      "loss": 0.8714,
+      "step": 2252
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.201145489866398e-05,
+      "loss": 0.9125,
+      "step": 2253
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.200396155383315e-05,
+      "loss": 0.812,
+      "step": 2254
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.199646536518793e-05,
+      "loss": 0.9004,
+      "step": 2255
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.1988966333982e-05,
+      "loss": 0.8599,
+      "step": 2256
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.198146446146956e-05,
+      "loss": 0.8888,
+      "step": 2257
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.197395974890522e-05,
+      "loss": 0.9508,
+      "step": 2258
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.1966452197544124e-05,
+      "loss": 1.0465,
+      "step": 2259
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.195894180864186e-05,
+      "loss": 0.9595,
+      "step": 2260
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.1951428583454484e-05,
+      "loss": 0.8418,
+      "step": 2261
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.194391252323855e-05,
+      "loss": 1.0666,
+      "step": 2262
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.193639362925109e-05,
+      "loss": 0.9165,
+      "step": 2263
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.192887190274957e-05,
+      "loss": 1.0018,
+      "step": 2264
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.192134734499197e-05,
+      "loss": 1.0162,
+      "step": 2265
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.191381995723672e-05,
+      "loss": 0.9482,
+      "step": 2266
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.1906289740742746e-05,
+      "loss": 0.9428,
+      "step": 2267
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.1898756696769423e-05,
+      "loss": 1.0483,
+      "step": 2268
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.1891220826576605e-05,
+      "loss": 0.9718,
+      "step": 2269
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.188368213142464e-05,
+      "loss": 0.9091,
+      "step": 2270
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.187614061257431e-05,
+      "loss": 0.827,
+      "step": 2271
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.18685962712869e-05,
+      "loss": 0.9917,
+      "step": 2272
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.186104910882417e-05,
+      "loss": 0.9849,
+      "step": 2273
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.1853499126448316e-05,
+      "loss": 0.9193,
+      "step": 2274
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.1845946325422047e-05,
+      "loss": 0.8098,
+      "step": 2275
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.1838390707008516e-05,
+      "loss": 0.7428,
+      "step": 2276
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.1830832272471356e-05,
+      "loss": 1.1149,
+      "step": 2277
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.1823271023074686e-05,
+      "loss": 0.8751,
+      "step": 2278
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.181570696008307e-05,
+      "loss": 0.9483,
+      "step": 2279
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.1808140084761537e-05,
+      "loss": 0.9962,
+      "step": 2280
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.180057039837564e-05,
+      "loss": 0.8806,
+      "step": 2281
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.1792997902191336e-05,
+      "loss": 0.9932,
+      "step": 2282
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.178542259747509e-05,
+      "loss": 0.8824,
+      "step": 2283
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.177784448549383e-05,
+      "loss": 0.6859,
+      "step": 2284
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.1770263567514956e-05,
+      "loss": 1.1137,
+      "step": 2285
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.1762679844806316e-05,
+      "loss": 1.0508,
+      "step": 2286
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.175509331863627e-05,
+      "loss": 1.0037,
+      "step": 2287
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.174750399027359e-05,
+      "loss": 1.1645,
+      "step": 2288
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.173991186098757e-05,
+      "loss": 1.0638,
+      "step": 2289
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.173231693204794e-05,
+      "loss": 0.8788,
+      "step": 2290
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.172471920472492e-05,
+      "loss": 0.9422,
+      "step": 2291
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.1717118680289175e-05,
+      "loss": 0.9713,
+      "step": 2292
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.1709515360011844e-05,
+      "loss": 1.0208,
+      "step": 2293
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.170190924516455e-05,
+      "loss": 1.1352,
+      "step": 2294
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.1694300337019365e-05,
+      "loss": 0.8698,
+      "step": 2295
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.168668863684884e-05,
+      "loss": 0.916,
+      "step": 2296
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.1679074145925977e-05,
+      "loss": 0.8905,
+      "step": 2297
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.1671456865524275e-05,
+      "loss": 0.9749,
+      "step": 2298
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.166383679691766e-05,
+      "loss": 0.9362,
+      "step": 2299
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.165621394138055e-05,
+      "loss": 1.0117,
+      "step": 2300
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.164858830018783e-05,
+      "loss": 0.952,
+      "step": 2301
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.164095987461485e-05,
+      "loss": 1.0106,
+      "step": 2302
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.16333286659374e-05,
+      "loss": 0.8858,
+      "step": 2303
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.162569467543177e-05,
+      "loss": 0.8445,
+      "step": 2304
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.16180579043747e-05,
+      "loss": 0.681,
+      "step": 2305
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.1610418354043394e-05,
+      "loss": 1.022,
+      "step": 2306
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.1602776025715516e-05,
+      "loss": 0.9896,
+      "step": 2307
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.159513092066922e-05,
+      "loss": 1.1374,
+      "step": 2308
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.158748304018308e-05,
+      "loss": 1.1068,
+      "step": 2309
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.1579832385536186e-05,
+      "loss": 1.0715,
+      "step": 2310
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.157217895800804e-05,
+      "loss": 1.0378,
+      "step": 2311
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.1564522758878656e-05,
+      "loss": 1.1072,
+      "step": 2312
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.1556863789428467e-05,
+      "loss": 0.883,
+      "step": 2313
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.154920205093842e-05,
+      "loss": 0.9612,
+      "step": 2314
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.1541537544689866e-05,
+      "loss": 0.6991,
+      "step": 2315
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.153387027196468e-05,
+      "loss": 0.8407,
+      "step": 2316
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.152620023404514e-05,
+      "loss": 1.0448,
+      "step": 2317
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.1518527432214024e-05,
+      "loss": 0.9557,
+      "step": 2318
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.151085186775459e-05,
+      "loss": 0.97,
+      "step": 2319
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.150317354195049e-05,
+      "loss": 0.877,
+      "step": 2320
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.149549245608591e-05,
+      "loss": 1.0916,
+      "step": 2321
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.148780861144545e-05,
+      "loss": 0.943,
+      "step": 2322
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.14801220093142e-05,
+      "loss": 0.9253,
+      "step": 2323
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.14724326509777e-05,
+      "loss": 0.9147,
+      "step": 2324
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.146474053772194e-05,
+      "loss": 1.0461,
+      "step": 2325
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.145704567083339e-05,
+      "loss": 0.9991,
+      "step": 2326
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.144934805159897e-05,
+      "loss": 0.9723,
+      "step": 2327
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.144164768130606e-05,
+      "loss": 0.9555,
+      "step": 2328
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.143394456124251e-05,
+      "loss": 0.9455,
+      "step": 2329
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.142623869269661e-05,
+      "loss": 1.0522,
+      "step": 2330
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.141853007695713e-05,
+      "loss": 0.8946,
+      "step": 2331
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.1410818715313296e-05,
+      "loss": 0.9756,
+      "step": 2332
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.140310460905478e-05,
+      "loss": 0.9948,
+      "step": 2333
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.139538775947173e-05,
+      "loss": 0.9486,
+      "step": 2334
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.138766816785474e-05,
+      "loss": 0.9099,
+      "step": 2335
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.137994583549486e-05,
+      "loss": 0.9703,
+      "step": 2336
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.137222076368361e-05,
+      "loss": 0.9615,
+      "step": 2337
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.136449295371298e-05,
+      "loss": 0.9748,
+      "step": 2338
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.135676240687537e-05,
+      "loss": 1.0924,
+      "step": 2339
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.13490291244637e-05,
+      "loss": 0.8596,
+      "step": 2340
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.134129310777129e-05,
+      "loss": 0.8918,
+      "step": 2341
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.1333554358091975e-05,
+      "loss": 0.9532,
+      "step": 2342
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.132581287671999e-05,
+      "loss": 0.9579,
+      "step": 2343
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.131806866495006e-05,
+      "loss": 1.0378,
+      "step": 2344
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.1310321724077374e-05,
+      "loss": 1.0129,
+      "step": 2345
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.130257205539754e-05,
+      "loss": 0.8498,
+      "step": 2346
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.129481966020665e-05,
+      "loss": 0.9381,
+      "step": 2347
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.128706453980127e-05,
+      "loss": 0.8889,
+      "step": 2348
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.127930669547837e-05,
+      "loss": 0.9755,
+      "step": 2349
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.127154612853542e-05,
+      "loss": 0.9233,
+      "step": 2350
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.126378284027034e-05,
+      "loss": 0.8688,
+      "step": 2351
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.1256016831981466e-05,
+      "loss": 0.8938,
+      "step": 2352
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.124824810496764e-05,
+      "loss": 0.9477,
+      "step": 2353
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.124047666052813e-05,
+      "loss": 0.756,
+      "step": 2354
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.123270249996267e-05,
+      "loss": 0.942,
+      "step": 2355
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.1224925624571433e-05,
+      "loss": 0.9917,
+      "step": 2356
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.1217146035655074e-05,
+      "loss": 0.8601,
+      "step": 2357
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.1209363734514674e-05,
+      "loss": 0.9443,
+      "step": 2358
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.120157872245177e-05,
+      "loss": 0.963,
+      "step": 2359
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.119379100076837e-05,
+      "loss": 0.8969,
+      "step": 2360
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.1186000570766926e-05,
+      "loss": 0.8405,
+      "step": 2361
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.1178207433750337e-05,
+      "loss": 0.9772,
+      "step": 2362
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.117041159102196e-05,
+      "loss": 0.8761,
+      "step": 2363
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.116261304388561e-05,
+      "loss": 1.0611,
+      "step": 2364
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.1154811793645556e-05,
+      "loss": 0.7995,
+      "step": 2365
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.114700784160649e-05,
+      "loss": 1.0342,
+      "step": 2366
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.113920118907361e-05,
+      "loss": 0.7995,
+      "step": 2367
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.11313918373525e-05,
+      "loss": 0.8385,
+      "step": 2368
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.1123579787749236e-05,
+      "loss": 0.8827,
+      "step": 2369
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.111576504157036e-05,
+      "loss": 1.0012,
+      "step": 2370
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.110794760012282e-05,
+      "loss": 0.9709,
+      "step": 2371
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.110012746471405e-05,
+      "loss": 1.1336,
+      "step": 2372
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.109230463665192e-05,
+      "loss": 0.799,
+      "step": 2373
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.108447911724475e-05,
+      "loss": 0.973,
+      "step": 2374
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.1076650907801316e-05,
+      "loss": 0.8917,
+      "step": 2375
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.1068820009630836e-05,
+      "loss": 0.9056,
+      "step": 2376
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.106098642404299e-05,
+      "loss": 0.8,
+      "step": 2377
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.10531501523479e-05,
+      "loss": 0.8151,
+      "step": 2378
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.104531119585613e-05,
+      "loss": 0.9757,
+      "step": 2379
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.1037469555878704e-05,
+      "loss": 0.8341,
+      "step": 2380
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.102962523372709e-05,
+      "loss": 0.9224,
+      "step": 2381
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.10217782307132e-05,
+      "loss": 0.995,
+      "step": 2382
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.1013928548149424e-05,
+      "loss": 0.9252,
+      "step": 2383
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.1006076187348544e-05,
+      "loss": 0.89,
+      "step": 2384
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.099822114962384e-05,
+      "loss": 0.9272,
+      "step": 2385
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.099036343628902e-05,
+      "loss": 0.9804,
+      "step": 2386
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.098250304865823e-05,
+      "loss": 0.9479,
+      "step": 2387
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.097463998804609e-05,
+      "loss": 0.8719,
+      "step": 2388
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.096677425576765e-05,
+      "loss": 0.9393,
+      "step": 2389
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.0958905853138393e-05,
+      "loss": 0.8855,
+      "step": 2390
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.0951034781474286e-05,
+      "loss": 0.8941,
+      "step": 2391
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.094316104209171e-05,
+      "loss": 0.9361,
+      "step": 2392
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.093528463630749e-05,
+      "loss": 0.9653,
+      "step": 2393
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.092740556543893e-05,
+      "loss": 1.0874,
+      "step": 2394
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.091952383080374e-05,
+      "loss": 0.9049,
+      "step": 2395
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.091163943372011e-05,
+      "loss": 0.9287,
+      "step": 2396
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.090375237550665e-05,
+      "loss": 0.9442,
+      "step": 2397
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.089586265748243e-05,
+      "loss": 0.9347,
+      "step": 2398
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.088797028096696e-05,
+      "loss": 0.8959,
+      "step": 2399
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.088007524728019e-05,
+      "loss": 0.8991,
+      "step": 2400
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.087217755774253e-05,
+      "loss": 0.867,
+      "step": 2401
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.086427721367481e-05,
+      "loss": 0.9606,
+      "step": 2402
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.0856374216398324e-05,
+      "loss": 0.8265,
+      "step": 2403
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.0848468567234796e-05,
+      "loss": 0.8549,
+      "step": 2404
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.084056026750641e-05,
+      "loss": 1.1057,
+      "step": 2405
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.083264931853578e-05,
+      "loss": 1.1101,
+      "step": 2406
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.082473572164596e-05,
+      "loss": 1.0841,
+      "step": 2407
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.0816819478160464e-05,
+      "loss": 0.8708,
+      "step": 2408
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.080890058940323e-05,
+      "loss": 0.9801,
+      "step": 2409
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.080097905669864e-05,
+      "loss": 0.858,
+      "step": 2410
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.079305488137154e-05,
+      "loss": 1.1201,
+      "step": 2411
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.07851280647472e-05,
+      "loss": 0.8838,
+      "step": 2412
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.077719860815132e-05,
+      "loss": 0.8865,
+      "step": 2413
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.076926651291008e-05,
+      "loss": 0.9903,
+      "step": 2414
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.076133178035005e-05,
+      "loss": 0.8747,
+      "step": 2415
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.0753394411798287e-05,
+      "loss": 0.9993,
+      "step": 2416
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.0745454408582265e-05,
+      "loss": 1.0073,
+      "step": 2417
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.07375117720299e-05,
+      "loss": 0.8486,
+      "step": 2418
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.072956650346955e-05,
+      "loss": 0.8231,
+      "step": 2419
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.0721618604230034e-05,
+      "loss": 0.9963,
+      "step": 2420
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.071366807564057e-05,
+      "loss": 0.8733,
+      "step": 2421
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.0705714919030847e-05,
+      "loss": 0.8477,
+      "step": 2422
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.069775913573098e-05,
+      "loss": 0.9876,
+      "step": 2423
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.0689800727071545e-05,
+      "loss": 0.7811,
+      "step": 2424
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.068183969438352e-05,
+      "loss": 0.8892,
+      "step": 2425
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.067387603899835e-05,
+      "loss": 0.7672,
+      "step": 2426
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.066590976224791e-05,
+      "loss": 0.7422,
+      "step": 2427
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.0657940865464505e-05,
+      "loss": 1.1254,
+      "step": 2428
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.06499693499809e-05,
+      "loss": 0.9444,
+      "step": 2429
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.064199521713028e-05,
+      "loss": 0.7962,
+      "step": 2430
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.063401846824628e-05,
+      "loss": 0.7877,
+      "step": 2431
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.062603910466295e-05,
+      "loss": 0.8819,
+      "step": 2432
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.06180571277148e-05,
+      "loss": 1.0224,
+      "step": 2433
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.061007253873677e-05,
+      "loss": 1.0094,
+      "step": 2434
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.0602085339064234e-05,
+      "loss": 0.8451,
+      "step": 2435
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.059409553003301e-05,
+      "loss": 0.9027,
+      "step": 2436
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.0586103112979344e-05,
+      "loss": 0.9887,
+      "step": 2437
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.057810808923993e-05,
+      "loss": 0.8719,
+      "step": 2438
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.057011046015187e-05,
+      "loss": 0.8253,
+      "step": 2439
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.056211022705274e-05,
+      "loss": 0.8048,
+      "step": 2440
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.055410739128052e-05,
+      "loss": 1.0937,
+      "step": 2441
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.054610195417364e-05,
+      "loss": 0.9776,
+      "step": 2442
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.053809391707098e-05,
+      "loss": 0.9513,
+      "step": 2443
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.053008328131182e-05,
+      "loss": 0.9064,
+      "step": 2444
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.052207004823589e-05,
+      "loss": 1.104,
+      "step": 2445
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.0514054219183375e-05,
+      "loss": 0.9287,
+      "step": 2446
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.0506035795494866e-05,
+      "loss": 0.9304,
+      "step": 2447
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.049801477851139e-05,
+      "loss": 1.0268,
+      "step": 2448
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.0489991169574426e-05,
+      "loss": 1.2014,
+      "step": 2449
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.048196497002588e-05,
+      "loss": 0.6877,
+      "step": 2450
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.047393618120807e-05,
+      "loss": 0.9824,
+      "step": 2451
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.046590480446379e-05,
+      "loss": 0.9644,
+      "step": 2452
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.045787084113622e-05,
+      "loss": 0.7764,
+      "step": 2453
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.0449834292569e-05,
+      "loss": 0.7921,
+      "step": 2454
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.04417951601062e-05,
+      "loss": 1.0439,
+      "step": 2455
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.043375344509232e-05,
+      "loss": 0.9775,
+      "step": 2456
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.042570914887227e-05,
+      "loss": 1.0217,
+      "step": 2457
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.041766227279145e-05,
+      "loss": 1.0044,
+      "step": 2458
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.0409612818195615e-05,
+      "loss": 0.9443,
+      "step": 2459
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.040156078643101e-05,
+      "loss": 0.9936,
+      "step": 2460
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.03935061788443e-05,
+      "loss": 0.9826,
+      "step": 2461
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.038544899678255e-05,
+      "loss": 0.9334,
+      "step": 2462
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.0377389241593286e-05,
+      "loss": 1.0452,
+      "step": 2463
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.036932691462446e-05,
+      "loss": 0.9215,
+      "step": 2464
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.036126201722444e-05,
+      "loss": 0.7615,
+      "step": 2465
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.0353194550742044e-05,
+      "loss": 1.0799,
+      "step": 2466
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.0345124516526505e-05,
+      "loss": 1.0735,
+      "step": 2467
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.033705191592749e-05,
+      "loss": 0.8379,
+      "step": 2468
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.03289767502951e-05,
+      "loss": 0.9294,
+      "step": 2469
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.032089902097984e-05,
+      "loss": 0.9757,
+      "step": 2470
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.031281872933268e-05,
+      "loss": 1.0481,
+      "step": 2471
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.0304735876705e-05,
+      "loss": 0.8399,
+      "step": 2472
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.0296650464448616e-05,
+      "loss": 0.9293,
+      "step": 2473
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.028856249391576e-05,
+      "loss": 1.175,
+      "step": 2474
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.0280471966459086e-05,
+      "loss": 1.0138,
+      "step": 2475
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.027237888343172e-05,
+      "loss": 0.9715,
+      "step": 2476
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.026428324618715e-05,
+      "loss": 0.9069,
+      "step": 2477
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.025618505607934e-05,
+      "loss": 0.7474,
+      "step": 2478
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.024808431446268e-05,
+      "loss": 0.7597,
+      "step": 2479
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.023998102269194e-05,
+      "loss": 1.0216,
+      "step": 2480
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.023187518212237e-05,
+      "loss": 0.9659,
+      "step": 2481
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.022376679410962e-05,
+      "loss": 0.9385,
+      "step": 2482
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.021565586000977e-05,
+      "loss": 0.9778,
+      "step": 2483
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.020754238117933e-05,
+      "loss": 0.7697,
+      "step": 2484
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.019942635897523e-05,
+      "loss": 1.0756,
+      "step": 2485
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.019130779475483e-05,
+      "loss": 0.7973,
+      "step": 2486
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.018318668987592e-05,
+      "loss": 0.8077,
+      "step": 2487
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.017506304569669e-05,
+      "loss": 0.8654,
+      "step": 2488
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.016693686357578e-05,
+      "loss": 0.8934,
+      "step": 2489
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.0158808144872254e-05,
+      "loss": 0.928,
+      "step": 2490
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.015067689094559e-05,
+      "loss": 0.9961,
+      "step": 2491
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.0142543103155685e-05,
+      "loss": 0.8854,
+      "step": 2492
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.013440678286288e-05,
+      "loss": 0.8727,
+      "step": 2493
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.012626793142792e-05,
+      "loss": 0.8108,
+      "step": 2494
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.0118126550211985e-05,
+      "loss": 0.7792,
+      "step": 2495
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.0109982640576674e-05,
+      "loss": 0.8325,
+      "step": 2496
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.0101836203884e-05,
+      "loss": 0.995,
+      "step": 2497
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.009368724149642e-05,
+      "loss": 0.9872,
+      "step": 2498
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.0085535754776806e-05,
+      "loss": 0.7877,
+      "step": 2499
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.0077381745088424e-05,
+      "loss": 1.0989,
+      "step": 2500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.0069225213795006e-05,
+      "loss": 0.8102,
+      "step": 2501
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.006106616226066e-05,
+      "loss": 1.0875,
+      "step": 2502
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.005290459184997e-05,
+      "loss": 1.0653,
+      "step": 2503
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.00447405039279e-05,
+      "loss": 0.9546,
+      "step": 2504
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.003657389985984e-05,
+      "loss": 1.0803,
+      "step": 2505
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.002840478101161e-05,
+      "loss": 0.745,
+      "step": 2506
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.002023314874945e-05,
+      "loss": 1.0072,
+      "step": 2507
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.001205900444003e-05,
+      "loss": 0.8534,
+      "step": 2508
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.0003882349450404e-05,
+      "loss": 0.9269,
+      "step": 2509
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.999570318514809e-05,
+      "loss": 1.0152,
+      "step": 2510
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9987521512900995e-05,
+      "loss": 0.9025,
+      "step": 2511
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.9979337334077465e-05,
+      "loss": 0.7207,
+      "step": 2512
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.997115065004625e-05,
+      "loss": 0.9093,
+      "step": 2513
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.996296146217653e-05,
+      "loss": 1.0819,
+      "step": 2514
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.9954769771837896e-05,
+      "loss": 0.9063,
+      "step": 2515
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.994657558040036e-05,
+      "loss": 0.904,
+      "step": 2516
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.993837888923436e-05,
+      "loss": 0.959,
+      "step": 2517
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.993017969971074e-05,
+      "loss": 0.9289,
+      "step": 2518
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.9921978013200766e-05,
+      "loss": 1.0818,
+      "step": 2519
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.991377383107613e-05,
+      "loss": 0.8946,
+      "step": 2520
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.9905567154708924e-05,
+      "loss": 0.9988,
+      "step": 2521
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.989735798547167e-05,
+      "loss": 0.8991,
+      "step": 2522
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.988914632473731e-05,
+      "loss": 0.8014,
+      "step": 2523
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.988093217387919e-05,
+      "loss": 1.0373,
+      "step": 2524
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.987271553427109e-05,
+      "loss": 0.8167,
+      "step": 2525
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.986449640728718e-05,
+      "loss": 0.9937,
+      "step": 2526
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.9856274794302065e-05,
+      "loss": 0.8952,
+      "step": 2527
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.984805069669078e-05,
+      "loss": 1.1238,
+      "step": 2528
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.983982411582873e-05,
+      "loss": 0.795,
+      "step": 2529
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.983159505309179e-05,
+      "loss": 0.9714,
+      "step": 2530
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.98233635098562e-05,
+      "loss": 1.0153,
+      "step": 2531
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.9815129487498656e-05,
+      "loss": 0.9766,
+      "step": 2532
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.9806892987396245e-05,
+      "loss": 0.9302,
+      "step": 2533
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.9798654010926464e-05,
+      "loss": 0.929,
+      "step": 2534
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.979041255946726e-05,
+      "loss": 1.0304,
+      "step": 2535
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.978216863439693e-05,
+      "loss": 0.9565,
+      "step": 2536
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.9773922237094255e-05,
+      "loss": 0.9453,
+      "step": 2537
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.97656733689384e-05,
+      "loss": 0.8884,
+      "step": 2538
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.9757422031308906e-05,
+      "loss": 1.0238,
+      "step": 2539
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.974916822558579e-05,
+      "loss": 0.9686,
+      "step": 2540
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.974091195314944e-05,
+      "loss": 1.1718,
+      "step": 2541
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.973265321538069e-05,
+      "loss": 0.8159,
+      "step": 2542
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.972439201366074e-05,
+      "loss": 0.7676,
+      "step": 2543
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.971612834937124e-05,
+      "loss": 0.9233,
+      "step": 2544
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.970786222389424e-05,
+      "loss": 0.7154,
+      "step": 2545
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.9699593638612206e-05,
+      "loss": 0.909,
+      "step": 2546
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.969132259490802e-05,
+      "loss": 0.8548,
+      "step": 2547
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.968304909416494e-05,
+      "loss": 0.8908,
+      "step": 2548
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.967477313776667e-05,
+      "loss": 0.8261,
+      "step": 2549
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.9666494727097325e-05,
+      "loss": 0.9923,
+      "step": 2550
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.965821386354142e-05,
+      "loss": 0.8332,
+      "step": 2551
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.9649930548483876e-05,
+      "loss": 0.9704,
+      "step": 2552
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.964164478331003e-05,
+      "loss": 0.8306,
+      "step": 2553
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.9633356569405646e-05,
+      "loss": 1.0113,
+      "step": 2554
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.962506590815685e-05,
+      "loss": 1.016,
+      "step": 2555
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.961677280095022e-05,
+      "loss": 0.8964,
+      "step": 2556
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.960847724917274e-05,
+      "loss": 0.8633,
+      "step": 2557
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.960017925421179e-05,
+      "loss": 0.9111,
+      "step": 2558
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.959187881745514e-05,
+      "loss": 1.0105,
+      "step": 2559
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.958357594029102e-05,
+      "loss": 0.942,
+      "step": 2560
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.957527062410802e-05,
+      "loss": 0.7961,
+      "step": 2561
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.956696287029517e-05,
+      "loss": 0.9543,
+      "step": 2562
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.955865268024188e-05,
+      "loss": 0.7284,
+      "step": 2563
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.9550340055337986e-05,
+      "loss": 1.1683,
+      "step": 2564
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.954202499697373e-05,
+      "loss": 0.9383,
+      "step": 2565
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.953370750653975e-05,
+      "loss": 1.0252,
+      "step": 2566
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.952538758542711e-05,
+      "loss": 1.0473,
+      "step": 2567
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.951706523502726e-05,
+      "loss": 0.9113,
+      "step": 2568
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9508740456732064e-05,
+      "loss": 0.9705,
+      "step": 2569
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9500413251933797e-05,
+      "loss": 0.8456,
+      "step": 2570
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.949208362202513e-05,
+      "loss": 0.8923,
+      "step": 2571
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.9483751568399164e-05,
+      "loss": 0.9201,
+      "step": 2572
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.947541709244936e-05,
+      "loss": 1.0164,
+      "step": 2573
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.9467080195569634e-05,
+      "loss": 0.9291,
+      "step": 2574
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.9458740879154264e-05,
+      "loss": 0.9698,
+      "step": 2575
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.945039914459798e-05,
+      "loss": 0.8241,
+      "step": 2576
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.944205499329585e-05,
+      "loss": 0.8875,
+      "step": 2577
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.9433708426643425e-05,
+      "loss": 1.0246,
+      "step": 2578
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.942535944603659e-05,
+      "loss": 0.9621,
+      "step": 2579
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.941700805287168e-05,
+      "loss": 0.9409,
+      "step": 2580
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.9408654248545415e-05,
+      "loss": 0.8812,
+      "step": 2581
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.940029803445492e-05,
+      "loss": 0.8875,
+      "step": 2582
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.939193941199771e-05,
+      "loss": 0.8695,
+      "step": 2583
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.938357838257174e-05,
+      "loss": 0.902,
+      "step": 2584
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.9375214947575325e-05,
+      "loss": 0.9405,
+      "step": 2585
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.936684910840722e-05,
+      "loss": 0.9935,
+      "step": 2586
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.935848086646654e-05,
+      "loss": 1.0417,
+      "step": 2587
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.9350110223152844e-05,
+      "loss": 0.7054,
+      "step": 2588
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.934173717986607e-05,
+      "loss": 0.7373,
+      "step": 2589
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.9333361738006544e-05,
+      "loss": 0.8689,
+      "step": 2590
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.9324983898975034e-05,
+      "loss": 0.9547,
+      "step": 2591
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.931660366417267e-05,
+      "loss": 0.973,
+      "step": 2592
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.9308221035001016e-05,
+      "loss": 0.9219,
+      "step": 2593
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.929983601286199e-05,
+      "loss": 0.9292,
+      "step": 2594
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.9291448599157963e-05,
+      "loss": 0.9721,
+      "step": 2595
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.9283058795291664e-05,
+      "loss": 0.7702,
+      "step": 2596
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.927466660266625e-05,
+      "loss": 0.8804,
+      "step": 2597
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.926627202268528e-05,
+      "loss": 1.0733,
+      "step": 2598
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.925787505675267e-05,
+      "loss": 0.9656,
+      "step": 2599
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9249475706272784e-05,
+      "loss": 0.9602,
+      "step": 2600
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.924107397265036e-05,
+      "loss": 1.0012,
+      "step": 2601
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.9232669857290526e-05,
+      "loss": 0.9422,
+      "step": 2602
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.922426336159884e-05,
+      "loss": 0.9255,
+      "step": 2603
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.921585448698123e-05,
+      "loss": 0.8942,
+      "step": 2604
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.920744323484404e-05,
+      "loss": 0.8511,
+      "step": 2605
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.919902960659401e-05,
+      "loss": 0.9509,
+      "step": 2606
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.919061360363824e-05,
+      "loss": 0.9086,
+      "step": 2607
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.9182195227384286e-05,
+      "loss": 0.7354,
+      "step": 2608
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.9173774479240064e-05,
+      "loss": 1.0264,
+      "step": 2609
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.91653513606139e-05,
+      "loss": 0.9089,
+      "step": 2610
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.9156925872914506e-05,
+      "loss": 1.0188,
+      "step": 2611
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.914849801755099e-05,
+      "loss": 0.8153,
+      "step": 2612
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.914006779593288e-05,
+      "loss": 0.984,
+      "step": 2613
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.913163520947007e-05,
+      "loss": 0.847,
+      "step": 2614
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.912320025957286e-05,
+      "loss": 0.7606,
+      "step": 2615
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.9114762947651953e-05,
+      "loss": 0.9229,
+      "step": 2616
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.910632327511844e-05,
+      "loss": 1.0508,
+      "step": 2617
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.9097881243383796e-05,
+      "loss": 0.8166,
+      "step": 2618
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.9089436853859915e-05,
+      "loss": 0.956,
+      "step": 2619
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.908099010795907e-05,
+      "loss": 0.9337,
+      "step": 2620
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.907254100709393e-05,
+      "loss": 1.0354,
+      "step": 2621
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.906408955267756e-05,
+      "loss": 0.9777,
+      "step": 2622
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.905563574612341e-05,
+      "loss": 0.7902,
+      "step": 2623
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.904717958884534e-05,
+      "loss": 0.8866,
+      "step": 2624
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.903872108225758e-05,
+      "loss": 1.0145,
+      "step": 2625
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.903026022777478e-05,
+      "loss": 1.0132,
+      "step": 2626
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.9021797026811965e-05,
+      "loss": 0.9019,
+      "step": 2627
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.9013331480784554e-05,
+      "loss": 0.987,
+      "step": 2628
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.900486359110836e-05,
+      "loss": 0.9374,
+      "step": 2629
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.899639335919959e-05,
+      "loss": 0.7992,
+      "step": 2630
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.898792078647484e-05,
+      "loss": 0.7244,
+      "step": 2631
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.8979445874351106e-05,
+      "loss": 1.0014,
+      "step": 2632
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.897096862424577e-05,
+      "loss": 0.9279,
+      "step": 2633
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.8962489037576586e-05,
+      "loss": 0.7977,
+      "step": 2634
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.895400711576173e-05,
+      "loss": 0.8966,
+      "step": 2635
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.894552286021974e-05,
+      "loss": 0.8763,
+      "step": 2636
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.893703627236958e-05,
+      "loss": 0.7525,
+      "step": 2637
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.892854735363057e-05,
+      "loss": 0.9092,
+      "step": 2638
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.892005610542243e-05,
+      "loss": 1.0149,
+      "step": 2639
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.891156252916529e-05,
+      "loss": 0.8747,
+      "step": 2640
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8903066626279624e-05,
+      "loss": 0.9357,
+      "step": 2641
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8894568398186344e-05,
+      "loss": 1.0336,
+      "step": 2642
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.8886067846306725e-05,
+      "loss": 0.8272,
+      "step": 2643
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.887756497206243e-05,
+      "loss": 0.9614,
+      "step": 2644
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.886905977687552e-05,
+      "loss": 0.8227,
+      "step": 2645
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.8860552262168434e-05,
+      "loss": 0.9104,
+      "step": 2646
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.885204242936401e-05,
+      "loss": 0.9524,
+      "step": 2647
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.884353027988547e-05,
+      "loss": 0.8193,
+      "step": 2648
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.883501581515642e-05,
+      "loss": 0.8707,
+      "step": 2649
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.882649903660085e-05,
+      "loss": 0.7586,
+      "step": 2650
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.881797994564315e-05,
+      "loss": 0.8622,
+      "step": 2651
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.880945854370809e-05,
+      "loss": 0.8359,
+      "step": 2652
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.880093483222082e-05,
+      "loss": 0.8384,
+      "step": 2653
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.879240881260687e-05,
+      "loss": 0.901,
+      "step": 2654
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.878388048629219e-05,
+      "loss": 1.0029,
+      "step": 2655
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.877534985470308e-05,
+      "loss": 0.9123,
+      "step": 2656
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.876681691926624e-05,
+      "loss": 0.7901,
+      "step": 2657
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.875828168140876e-05,
+      "loss": 1.0258,
+      "step": 2658
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.8749744142558105e-05,
+      "loss": 0.8761,
+      "step": 2659
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8741204304142134e-05,
+      "loss": 0.9442,
+      "step": 2660
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8732662167589066e-05,
+      "loss": 0.9725,
+      "step": 2661
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.872411773432755e-05,
+      "loss": 0.8973,
+      "step": 2662
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.8715571005786577e-05,
+      "loss": 0.8967,
+      "step": 2663
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.870702198339555e-05,
+      "loss": 0.9979,
+      "step": 2664
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.8698470668584234e-05,
+      "loss": 0.9246,
+      "step": 2665
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.868991706278279e-05,
+      "loss": 0.8596,
+      "step": 2666
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.868136116742176e-05,
+      "loss": 0.782,
+      "step": 2667
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.8672802983932065e-05,
+      "loss": 0.9743,
+      "step": 2668
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.866424251374501e-05,
+      "loss": 0.927,
+      "step": 2669
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.865567975829229e-05,
+      "loss": 0.882,
+      "step": 2670
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.864711471900598e-05,
+      "loss": 0.7602,
+      "step": 2671
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.8638547397318525e-05,
+      "loss": 0.6951,
+      "step": 2672
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.862997779466276e-05,
+      "loss": 1.0534,
+      "step": 2673
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.862140591247191e-05,
+      "loss": 0.8264,
+      "step": 2674
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.861283175217957e-05,
+      "loss": 0.9067,
+      "step": 2675
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.860425531521971e-05,
+      "loss": 0.9453,
+      "step": 2676
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.859567660302669e-05,
+      "loss": 0.8415,
+      "step": 2677
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.858709561703527e-05,
+      "loss": 0.9402,
+      "step": 2678
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.857851235868055e-05,
+      "loss": 0.8283,
+      "step": 2679
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.856992682939803e-05,
+      "loss": 0.6305,
+      "step": 2680
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.856133903062361e-05,
+      "loss": 1.0739,
+      "step": 2681
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.8552748963793524e-05,
+      "loss": 1.005,
+      "step": 2682
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.854415663034443e-05,
+      "loss": 0.9593,
+      "step": 2683
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.8535562031713334e-05,
+      "loss": 1.1126,
+      "step": 2684
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.852696516933765e-05,
+      "loss": 1.0217,
+      "step": 2685
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.851836604465513e-05,
+      "loss": 0.8341,
+      "step": 2686
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.850976465910394e-05,
+      "loss": 0.9009,
+      "step": 2687
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8501161014122624e-05,
+      "loss": 0.9244,
+      "step": 2688
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.849255511115008e-05,
+      "loss": 0.966,
+      "step": 2689
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.8483946951625584e-05,
+      "loss": 1.0913,
+      "step": 2690
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.847533653698882e-05,
+      "loss": 0.8176,
+      "step": 2691
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.846672386867982e-05,
+      "loss": 0.8588,
+      "step": 2692
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.8458108948139013e-05,
+      "loss": 0.8411,
+      "step": 2693
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.844949177680718e-05,
+      "loss": 0.9258,
+      "step": 2694
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.84408723561255e-05,
+      "loss": 0.8789,
+      "step": 2695
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.843225068753553e-05,
+      "loss": 0.9636,
+      "step": 2696
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.842362677247919e-05,
+      "loss": 0.9063,
+      "step": 2697
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.841500061239877e-05,
+      "loss": 0.9632,
+      "step": 2698
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.8406372208736954e-05,
+      "loss": 0.8298,
+      "step": 2699
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.839774156293678e-05,
+      "loss": 0.7935,
+      "step": 2700
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.8389108676441694e-05,
+      "loss": 0.6354,
+      "step": 2701
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.838047355069549e-05,
+      "loss": 0.9734,
+      "step": 2702
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.837183618714233e-05,
+      "loss": 0.9384,
+      "step": 2703
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.836319658722678e-05,
+      "loss": 1.0895,
+      "step": 2704
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.8354554752393754e-05,
+      "loss": 1.061,
+      "step": 2705
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.834591068408855e-05,
+      "loss": 1.0287,
+      "step": 2706
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.8337264383756847e-05,
+      "loss": 0.9892,
+      "step": 2707
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.8328615852844676e-05,
+      "loss": 1.063,
+      "step": 2708
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.831996509279846e-05,
+      "loss": 0.8261,
+      "step": 2709
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.831131210506499e-05,
+      "loss": 0.9226,
+      "step": 2710
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.830265689109143e-05,
+      "loss": 0.6392,
+      "step": 2711
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.82939994523253e-05,
+      "loss": 0.7829,
+      "step": 2712
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.828533979021452e-05,
+      "loss": 0.9931,
+      "step": 2713
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.827667790620737e-05,
+      "loss": 0.9129,
+      "step": 2714
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.8268013801752486e-05,
+      "loss": 0.92,
+      "step": 2715
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.8259347478298916e-05,
+      "loss": 0.8322,
+      "step": 2716
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.825067893729603e-05,
+      "loss": 1.0417,
+      "step": 2717
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.824200818019359e-05,
+      "loss": 0.8933,
+      "step": 2718
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.8233335208441745e-05,
+      "loss": 0.8737,
+      "step": 2719
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.8224660023490986e-05,
+      "loss": 0.8591,
+      "step": 2720
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.82159826267922e-05,
+      "loss": 1.0069,
+      "step": 2721
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.820730301979662e-05,
+      "loss": 0.9505,
+      "step": 2722
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.819862120395587e-05,
+      "loss": 0.9232,
+      "step": 2723
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.8189937180721926e-05,
+      "loss": 0.8966,
+      "step": 2724
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.818125095154714e-05,
+      "loss": 0.8913,
+      "step": 2725
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.817256251788425e-05,
+      "loss": 1.0081,
+      "step": 2726
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.816387188118632e-05,
+      "loss": 0.846,
+      "step": 2727
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.815517904290683e-05,
+      "loss": 0.9325,
+      "step": 2728
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.81464840044996e-05,
+      "loss": 0.9481,
+      "step": 2729
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.813778676741882e-05,
+      "loss": 0.8948,
+      "step": 2730
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.812908733311905e-05,
+      "loss": 0.8639,
+      "step": 2731
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.812038570305524e-05,
+      "loss": 0.9125,
+      "step": 2732
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.811168187868267e-05,
+      "loss": 0.9168,
+      "step": 2733
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.810297586145701e-05,
+      "loss": 0.9245,
+      "step": 2734
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.8094267652834295e-05,
+      "loss": 1.0497,
+      "step": 2735
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.808555725427091e-05,
+      "loss": 0.8139,
+      "step": 2736
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.807684466722363e-05,
+      "loss": 0.8422,
+      "step": 2737
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.8068129893149576e-05,
+      "loss": 0.8966,
+      "step": 2738
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.805941293350625e-05,
+      "loss": 0.9137,
+      "step": 2739
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.805069378975152e-05,
+      "loss": 0.9892,
+      "step": 2740
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.80419724633436e-05,
+      "loss": 0.9573,
+      "step": 2741
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.8033248955741086e-05,
+      "loss": 0.8089,
+      "step": 2742
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.802452326840294e-05,
+      "loss": 0.8996,
+      "step": 2743
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.801579540278847e-05,
+      "loss": 0.8301,
+      "step": 2744
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.8007065360357375e-05,
+      "loss": 0.9305,
+      "step": 2745
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.7998333142569694e-05,
+      "loss": 0.8663,
+      "step": 2746
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.798959875088584e-05,
+      "loss": 0.8224,
+      "step": 2747
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.7980862186766594e-05,
+      "loss": 0.8438,
+      "step": 2748
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.79721234516731e-05,
+      "loss": 0.8939,
+      "step": 2749
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.796338254706686e-05,
+      "loss": 0.7061,
+      "step": 2750
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.795463947440973e-05,
+      "loss": 0.887,
+      "step": 2751
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.794589423516394e-05,
+      "loss": 0.9481,
+      "step": 2752
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.793714683079209e-05,
+      "loss": 0.8081,
+      "step": 2753
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.792839726275713e-05,
+      "loss": 0.9033,
+      "step": 2754
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.7919645532522375e-05,
+      "loss": 0.9106,
+      "step": 2755
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.7910891641551494e-05,
+      "loss": 0.8518,
+      "step": 2756
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.790213559130852e-05,
+      "loss": 0.7814,
+      "step": 2757
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.7893377383257874e-05,
+      "loss": 0.9272,
+      "step": 2758
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.788461701886431e-05,
+      "loss": 0.8221,
+      "step": 2759
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.787585449959293e-05,
+      "loss": 1.0123,
+      "step": 2760
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.786708982690923e-05,
+      "loss": 0.7409,
+      "step": 2761
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.785832300227905e-05,
+      "loss": 0.993,
+      "step": 2762
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.784955402716858e-05,
+      "loss": 0.7502,
+      "step": 2763
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.78407829030444e-05,
+      "loss": 0.7739,
+      "step": 2764
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.783200963137341e-05,
+      "loss": 0.844,
+      "step": 2765
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.78232342136229e-05,
+      "loss": 0.9682,
+      "step": 2766
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.781445665126051e-05,
+      "loss": 0.9146,
+      "step": 2767
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.7805676945754235e-05,
+      "loss": 1.0908,
+      "step": 2768
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.779689509857241e-05,
+      "loss": 0.7511,
+      "step": 2769
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.778811111118379e-05,
+      "loss": 0.9236,
+      "step": 2770
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.77793249850574e-05,
+      "loss": 0.8414,
+      "step": 2771
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.77705367216627e-05,
+      "loss": 0.835,
+      "step": 2772
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.776174632246946e-05,
+      "loss": 0.7344,
+      "step": 2773
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.775295378894784e-05,
+      "loss": 0.7573,
+      "step": 2774
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.774415912256832e-05,
+      "loss": 0.9296,
+      "step": 2775
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.7735362324801784e-05,
+      "loss": 0.7822,
+      "step": 2776
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.772656339711942e-05,
+      "loss": 0.8711,
+      "step": 2777
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.771776234099281e-05,
+      "loss": 0.9477,
+      "step": 2778
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.7708959157893876e-05,
+      "loss": 0.8777,
+      "step": 2779
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.77001538492949e-05,
+      "loss": 0.8628,
+      "step": 2780
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.769134641666852e-05,
+      "loss": 0.8759,
+      "step": 2781
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.768253686148773e-05,
+      "loss": 0.9417,
+      "step": 2782
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.7673725185225865e-05,
+      "loss": 0.9085,
+      "step": 2783
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.766491138935665e-05,
+      "loss": 0.8199,
+      "step": 2784
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.7656095475354124e-05,
+      "loss": 0.8903,
+      "step": 2785
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.76472774446927e-05,
+      "loss": 0.8327,
+      "step": 2786
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.7638457298847154e-05,
+      "loss": 0.8454,
+      "step": 2787
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.762963503929259e-05,
+      "loss": 0.8803,
+      "step": 2788
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.762081066750448e-05,
+      "loss": 0.9243,
+      "step": 2789
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.7611984184958665e-05,
+      "loss": 1.0365,
+      "step": 2790
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.7603155593131314e-05,
+      "loss": 0.8519,
+      "step": 2791
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.7594324893498945e-05,
+      "loss": 0.8713,
+      "step": 2792
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.758549208753846e-05,
+      "loss": 0.8996,
+      "step": 2793
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.757665717672709e-05,
+      "loss": 0.8806,
+      "step": 2794
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.756782016254242e-05,
+      "loss": 0.8496,
+      "step": 2795
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.755898104646239e-05,
+      "loss": 0.8419,
+      "step": 2796
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.755013982996529e-05,
+      "loss": 0.821,
+      "step": 2797
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.754129651452977e-05,
+      "loss": 0.9107,
+      "step": 2798
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.7532451101634826e-05,
+      "loss": 0.7732,
+      "step": 2799
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.752360359275978e-05,
+      "loss": 0.8088,
+      "step": 2800
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.751475398938434e-05,
+      "loss": 1.0604,
+      "step": 2801
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.750590229298856e-05,
+      "loss": 1.0531,
+      "step": 2802
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.749704850505282e-05,
+      "loss": 1.0429,
+      "step": 2803
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.7488192627057884e-05,
+      "loss": 0.8191,
+      "step": 2804
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.747933466048482e-05,
+      "loss": 0.9306,
+      "step": 2805
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.74704746068151e-05,
+      "loss": 0.8088,
+      "step": 2806
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.7461612467530494e-05,
+      "loss": 1.0722,
+      "step": 2807
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.745274824411315e-05,
+      "loss": 0.8409,
+      "step": 2808
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.744388193804557e-05,
+      "loss": 0.825,
+      "step": 2809
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.743501355081058e-05,
+      "loss": 0.9459,
+      "step": 2810
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.742614308389137e-05,
+      "loss": 0.8229,
+      "step": 2811
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.741727053877148e-05,
+      "loss": 0.9542,
+      "step": 2812
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.7408395916934774e-05,
+      "loss": 0.9563,
+      "step": 2813
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.73995192198655e-05,
+      "loss": 0.799,
+      "step": 2814
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.7390640449048225e-05,
+      "loss": 0.7693,
+      "step": 2815
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.738175960596788e-05,
+      "loss": 0.9434,
+      "step": 2816
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.7372876692109724e-05,
+      "loss": 0.8316,
+      "step": 2817
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.7363991708959386e-05,
+      "loss": 0.7911,
+      "step": 2818
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.7355104658002815e-05,
+      "loss": 0.9398,
+      "step": 2819
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.7346215540726324e-05,
+      "loss": 0.7285,
+      "step": 2820
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.733732435861656e-05,
+      "loss": 0.8342,
+      "step": 2821
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.732843111316055e-05,
+      "loss": 0.7072,
+      "step": 2822
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.731953580584561e-05,
+      "loss": 0.7002,
+      "step": 2823
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.731063843815943e-05,
+      "loss": 1.0927,
+      "step": 2824
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.7301739011590056e-05,
+      "loss": 0.8857,
+      "step": 2825
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.729283752762586e-05,
+      "loss": 0.7408,
+      "step": 2826
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.7283933987755546e-05,
+      "loss": 0.7243,
+      "step": 2827
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.727502839346821e-05,
+      "loss": 0.8258,
+      "step": 2828
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.726612074625325e-05,
+      "loss": 0.9751,
+      "step": 2829
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.725721104760041e-05,
+      "loss": 0.9669,
+      "step": 2830
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.72482992989998e-05,
+      "loss": 0.7983,
+      "step": 2831
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.7239385501941845e-05,
+      "loss": 0.8547,
+      "step": 2832
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.723046965791733e-05,
+      "loss": 0.9419,
+      "step": 2833
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.722155176841738e-05,
+      "loss": 0.8016,
+      "step": 2834
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.721263183493346e-05,
+      "loss": 0.7753,
+      "step": 2835
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.7203709858957384e-05,
+      "loss": 0.7498,
+      "step": 2836
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.71947858419813e-05,
+      "loss": 1.0419,
+      "step": 2837
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.718585978549768e-05,
+      "loss": 0.9355,
+      "step": 2838
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.717693169099937e-05,
+      "loss": 0.8946,
+      "step": 2839
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.716800155997955e-05,
+      "loss": 0.8923,
+      "step": 2840
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.715906939393172e-05,
+      "loss": 1.0624,
+      "step": 2841
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.7150135194349736e-05,
+      "loss": 0.8855,
+      "step": 2842
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.7141198962727786e-05,
+      "loss": 0.8918,
+      "step": 2843
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.713226070056042e-05,
+      "loss": 0.9933,
+      "step": 2844
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.7123320409342494e-05,
+      "loss": 1.1614,
+      "step": 2845
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.7114378090569226e-05,
+      "loss": 0.6259,
+      "step": 2846
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.710543374573617e-05,
+      "loss": 0.9446,
+      "step": 2847
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.709648737633921e-05,
+      "loss": 0.9053,
+      "step": 2848
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.708753898387458e-05,
+      "loss": 0.7212,
+      "step": 2849
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.707858856983885e-05,
+      "loss": 0.7493,
+      "step": 2850
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.706963613572891e-05,
+      "loss": 0.9996,
+      "step": 2851
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.706068168304202e-05,
+      "loss": 0.9253,
+      "step": 2852
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.705172521327576e-05,
+      "loss": 0.9703,
+      "step": 2853
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.704276672792803e-05,
+      "loss": 0.9536,
+      "step": 2854
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.70338062284971e-05,
+      "loss": 0.8937,
+      "step": 2855
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.702484371648155e-05,
+      "loss": 0.9451,
+      "step": 2856
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.7015879193380334e-05,
+      "loss": 0.9564,
+      "step": 2857
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.700691266069269e-05,
+      "loss": 0.8836,
+      "step": 2858
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.699794411991822e-05,
+      "loss": 0.9953,
+      "step": 2859
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.698897357255688e-05,
+      "loss": 0.8707,
+      "step": 2860
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.698000102010893e-05,
+      "loss": 0.7066,
+      "step": 2861
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.697102646407497e-05,
+      "loss": 1.044,
+      "step": 2862
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.6962049905955964e-05,
+      "loss": 1.0261,
+      "step": 2863
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.695307134725317e-05,
+      "loss": 0.7772,
+      "step": 2864
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.6944090789468206e-05,
+      "loss": 0.8759,
+      "step": 2865
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.693510823410302e-05,
+      "loss": 0.918,
+      "step": 2866
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.692612368265989e-05,
+      "loss": 0.9968,
+      "step": 2867
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.691713713664144e-05,
+      "loss": 0.7842,
+      "step": 2868
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.6908148597550605e-05,
+      "loss": 0.8805,
+      "step": 2869
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.689915806689067e-05,
+      "loss": 1.1374,
+      "step": 2870
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.689016554616525e-05,
+      "loss": 0.9659,
+      "step": 2871
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.68811710368783e-05,
+      "loss": 0.9185,
+      "step": 2872
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.687217454053409e-05,
+      "loss": 0.858,
+      "step": 2873
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.686317605863723e-05,
+      "loss": 0.6962,
+      "step": 2874
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.685417559269267e-05,
+      "loss": 0.7051,
+      "step": 2875
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.684517314420568e-05,
+      "loss": 0.9736,
+      "step": 2876
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.6836168714681885e-05,
+      "loss": 0.9331,
+      "step": 2877
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.68271623056272e-05,
+      "loss": 0.9016,
+      "step": 2878
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.6818153918547916e-05,
+      "loss": 0.9316,
+      "step": 2879
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.680914355495062e-05,
+      "loss": 0.7095,
+      "step": 2880
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.680013121634224e-05,
+      "loss": 1.034,
+      "step": 2881
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.679111690423006e-05,
+      "loss": 0.7397,
+      "step": 2882
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.678210062012164e-05,
+      "loss": 0.7594,
+      "step": 2883
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.677308236552493e-05,
+      "loss": 0.816,
+      "step": 2884
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.6764062141948165e-05,
+      "loss": 0.8436,
+      "step": 2885
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.6755039950899925e-05,
+      "loss": 0.8824,
+      "step": 2886
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.674601579388913e-05,
+      "loss": 0.9391,
+      "step": 2887
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.673698967242501e-05,
+      "loss": 0.8248,
+      "step": 2888
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.672796158801713e-05,
+      "loss": 0.821,
+      "step": 2889
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.671893154217539e-05,
+      "loss": 0.7597,
+      "step": 2890
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.670989953641001e-05,
+      "loss": 0.7231,
+      "step": 2891
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.670086557223155e-05,
+      "loss": 0.7755,
+      "step": 2892
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.669182965115087e-05,
+      "loss": 0.9508,
+      "step": 2893
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.6682791774679185e-05,
+      "loss": 0.941,
+      "step": 2894
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.667375194432804e-05,
+      "loss": 0.7391,
+      "step": 2895
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.666471016160928e-05,
+      "loss": 1.052,
+      "step": 2896
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.6655666428035085e-05,
+      "loss": 0.7572,
+      "step": 2897
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.664662074511799e-05,
+      "loss": 1.0389,
+      "step": 2898
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.663757311437081e-05,
+      "loss": 1.0169,
+      "step": 2899
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.6628523537306724e-05,
+      "loss": 0.8967,
+      "step": 2900
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.6619472015439216e-05,
+      "loss": 1.0468,
+      "step": 2901
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.66104185502821e-05,
+      "loss": 0.6953,
+      "step": 2902
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.6601363143349515e-05,
+      "loss": 0.9582,
+      "step": 2903
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.659230579615593e-05,
+      "loss": 0.8132,
+      "step": 2904
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.658324651021613e-05,
+      "loss": 0.8718,
+      "step": 2905
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.657418528704523e-05,
+      "loss": 0.961,
+      "step": 2906
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.656512212815866e-05,
+      "loss": 0.8543,
+      "step": 2907
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.65560570350722e-05,
+      "loss": 0.6662,
+      "step": 2908
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.654699000930192e-05,
+      "loss": 0.8613,
+      "step": 2909
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.653792105236422e-05,
+      "loss": 1.0384,
+      "step": 2910
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.652885016577585e-05,
+      "loss": 0.8593,
+      "step": 2911
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.6519777351053855e-05,
+      "loss": 0.8583,
+      "step": 2912
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.651070260971561e-05,
+      "loss": 0.9163,
+      "step": 2913
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.6501625943278805e-05,
+      "loss": 0.8821,
+      "step": 2914
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.649254735326147e-05,
+      "loss": 1.0383,
+      "step": 2915
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.648346684118195e-05,
+      "loss": 0.8429,
+      "step": 2916
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.64743844085589e-05,
+      "loss": 0.9505,
+      "step": 2917
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.646530005691131e-05,
+      "loss": 0.8444,
+      "step": 2918
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.6456213787758483e-05,
+      "loss": 0.7566,
+      "step": 2919
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.6447125602620046e-05,
+      "loss": 0.9947,
+      "step": 2920
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.6438035503015945e-05,
+      "loss": 0.7639,
+      "step": 2921
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.642894349046644e-05,
+      "loss": 0.9425,
+      "step": 2922
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.6419849566492126e-05,
+      "loss": 0.8447,
+      "step": 2923
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.641075373261391e-05,
+      "loss": 1.0853,
+      "step": 2924
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.640165599035302e-05,
+      "loss": 0.738,
+      "step": 2925
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.6392556341230984e-05,
+      "loss": 0.9199,
+      "step": 2926
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.6383454786769686e-05,
+      "loss": 0.9649,
+      "step": 2927
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.63743513284913e-05,
+      "loss": 0.9239,
+      "step": 2928
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.636524596791832e-05,
+      "loss": 0.8803,
+      "step": 2929
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.635613870657357e-05,
+      "loss": 0.8712,
+      "step": 2930
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.6347029545980194e-05,
+      "loss": 0.9743,
+      "step": 2931
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.633791848766164e-05,
+      "loss": 0.9056,
+      "step": 2932
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.6328805533141684e-05,
+      "loss": 0.8972,
+      "step": 2933
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.63196906839444e-05,
+      "loss": 0.8398,
+      "step": 2934
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.631057394159422e-05,
+      "loss": 0.9892,
+      "step": 2935
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.630145530761585e-05,
+      "loss": 0.926,
+      "step": 2936
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.629233478353432e-05,
+      "loss": 1.1249,
+      "step": 2937
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.628321237087501e-05,
+      "loss": 0.7652,
+      "step": 2938
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.6274088071163574e-05,
+      "loss": 0.7163,
+      "step": 2939
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.6264961885925996e-05,
+      "loss": 0.8791,
+      "step": 2940
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.625583381668859e-05,
+      "loss": 0.6512,
+      "step": 2941
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.624670386497796e-05,
+      "loss": 0.8481,
+      "step": 2942
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.6237572032321046e-05,
+      "loss": 0.8027,
+      "step": 2943
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.6228438320245096e-05,
+      "loss": 0.8356,
+      "step": 2944
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.621930273027767e-05,
+      "loss": 0.7738,
+      "step": 2945
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.6210165263946625e-05,
+      "loss": 0.936,
+      "step": 2946
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.620102592278017e-05,
+      "loss": 0.7817,
+      "step": 2947
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.619188470830681e-05,
+      "loss": 0.9135,
+      "step": 2948
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.6182741622055346e-05,
+      "loss": 0.7878,
+      "step": 2949
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.61735966655549e-05,
+      "loss": 0.9596,
+      "step": 2950
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.6164449840334946e-05,
+      "loss": 0.9604,
+      "step": 2951
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.61553011479252e-05,
+      "loss": 0.8519,
+      "step": 2952
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.614615058985575e-05,
+      "loss": 0.8125,
+      "step": 2953
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.6136998167656976e-05,
+      "loss": 0.8611,
+      "step": 2954
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.6127843882859555e-05,
+      "loss": 0.9673,
+      "step": 2955
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.611868773699449e-05,
+      "loss": 0.8865,
+      "step": 2956
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.610952973159311e-05,
+      "loss": 0.746,
+      "step": 2957
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.6100369868187e-05,
+      "loss": 0.9074,
+      "step": 2958
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.6091208148308143e-05,
+      "loss": 0.6764,
+      "step": 2959
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.608204457348876e-05,
+      "loss": 1.1276,
+      "step": 2960
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.607287914526139e-05,
+      "loss": 0.8903,
+      "step": 2961
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.6063711865158926e-05,
+      "loss": 0.9881,
+      "step": 2962
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.6054542734714535e-05,
+      "loss": 0.9995,
+      "step": 2963
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.6045371755461685e-05,
+      "loss": 0.8763,
+      "step": 2964
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.6036198928934196e-05,
+      "loss": 0.9229,
+      "step": 2965
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.6027024256666146e-05,
+      "loss": 0.794,
+      "step": 2966
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.601784774019196e-05,
+      "loss": 0.8332,
+      "step": 2967
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.600866938104636e-05,
+      "loss": 0.867,
+      "step": 2968
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.5999489180764356e-05,
+      "loss": 0.9653,
+      "step": 2969
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.59903071408813e-05,
+      "loss": 0.8791,
+      "step": 2970
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.5981123262932826e-05,
+      "loss": 0.9296,
+      "step": 2971
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.597193754845489e-05,
+      "loss": 0.7801,
+      "step": 2972
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.596274999898375e-05,
+      "loss": 0.8436,
+      "step": 2973
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.595356061605597e-05,
+      "loss": 0.9673,
+      "step": 2974
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.5944369401208414e-05,
+      "loss": 0.9229,
+      "step": 2975
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.593517635597827e-05,
+      "loss": 0.8768,
+      "step": 2976
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.592598148190301e-05,
+      "loss": 0.8257,
+      "step": 2977
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.591678478052044e-05,
+      "loss": 0.8396,
+      "step": 2978
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.590758625336864e-05,
+      "loss": 0.8145,
+      "step": 2979
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.589838590198602e-05,
+      "loss": 0.8546,
+      "step": 2980
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.5889183727911284e-05,
+      "loss": 0.902,
+      "step": 2981
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.5879979732683436e-05,
+      "loss": 0.953,
+      "step": 2982
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.58707739178418e-05,
+      "loss": 0.9902,
+      "step": 2983
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.586156628492598e-05,
+      "loss": 0.6504,
+      "step": 2984
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.585235683547592e-05,
+      "loss": 0.6986,
+      "step": 2985
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.584314557103184e-05,
+      "loss": 0.8277,
+      "step": 2986
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.583393249313427e-05,
+      "loss": 0.9077,
+      "step": 2987
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.5824717603324034e-05,
+      "loss": 0.9279,
+      "step": 2988
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.581550090314228e-05,
+      "loss": 0.8839,
+      "step": 2989
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.580628239413045e-05,
+      "loss": 0.8804,
+      "step": 2990
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.579706207783028e-05,
+      "loss": 0.9166,
+      "step": 2991
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.578783995578382e-05,
+      "loss": 0.7159,
+      "step": 2992
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.57786160295334e-05,
+      "loss": 0.8286,
+      "step": 2993
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.57693903006217e-05,
+      "loss": 1.0367,
+      "step": 2994
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.5760162770591645e-05,
+      "loss": 0.9121,
+      "step": 2995
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.575093344098648e-05,
+      "loss": 0.9109,
+      "step": 2996
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.5741702313349785e-05,
+      "loss": 0.9482,
+      "step": 2997
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.57324693892254e-05,
+      "loss": 0.8926,
+      "step": 2998
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.572323467015746e-05,
+      "loss": 0.8693,
+      "step": 2999
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.571399815769045e-05,
+      "loss": 0.8701,
+      "step": 3000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.57047598533691e-05,
+      "loss": 0.8069,
+      "step": 3001
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.569551975873847e-05,
+      "loss": 0.8991,
+      "step": 3002
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.568627787534392e-05,
+      "loss": 0.8477,
+      "step": 3003
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.5677034204731086e-05,
+      "loss": 0.6867,
+      "step": 3004
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.5667788748445934e-05,
+      "loss": 0.9723,
+      "step": 3005
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.5658541508034705e-05,
+      "loss": 0.8558,
+      "step": 3006
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.564929248504395e-05,
+      "loss": 0.9647,
+      "step": 3007
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.564004168102051e-05,
+      "loss": 0.7664,
+      "step": 3008
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.563078909751154e-05,
+      "loss": 0.9348,
+      "step": 3009
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.562153473606446e-05,
+      "loss": 0.7831,
+      "step": 3010
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.5612278598227025e-05,
+      "loss": 0.715,
+      "step": 3011
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.5603020685547274e-05,
+      "loss": 0.8753,
+      "step": 3012
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.559376099957352e-05,
+      "loss": 0.996,
+      "step": 3013
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.558449954185442e-05,
+      "loss": 0.7598,
+      "step": 3014
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.557523631393887e-05,
+      "loss": 0.9118,
+      "step": 3015
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.556597131737611e-05,
+      "loss": 0.8811,
+      "step": 3016
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.5556704553715645e-05,
+      "loss": 0.9994,
+      "step": 3017
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.554743602450731e-05,
+      "loss": 0.936,
+      "step": 3018
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.5538165731301185e-05,
+      "loss": 0.7409,
+      "step": 3019
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.552889367564769e-05,
+      "loss": 0.8438,
+      "step": 3020
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.551961985909752e-05,
+      "loss": 0.9719,
+      "step": 3021
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.551034428320167e-05,
+      "loss": 0.9644,
+      "step": 3022
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.550106694951141e-05,
+      "loss": 0.8553,
+      "step": 3023
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.549178785957835e-05,
+      "loss": 0.9393,
+      "step": 3024
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.548250701495432e-05,
+      "loss": 0.8852,
+      "step": 3025
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.547322441719154e-05,
+      "loss": 0.7542,
+      "step": 3026
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.546394006784245e-05,
+      "loss": 0.6818,
+      "step": 3027
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.545465396845977e-05,
+      "loss": 0.9582,
+      "step": 3028
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.54453661205966e-05,
+      "loss": 0.8762,
+      "step": 3029
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.543607652580626e-05,
+      "loss": 0.7431,
+      "step": 3030
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.542678518564235e-05,
+      "loss": 0.8528,
+      "step": 3031
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.541749210165884e-05,
+      "loss": 0.8216,
+      "step": 3032
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.540819727540991e-05,
+      "loss": 0.7054,
+      "step": 3033
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.5398900708450094e-05,
+      "loss": 0.8595,
+      "step": 3034
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.538960240233417e-05,
+      "loss": 0.977,
+      "step": 3035
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.5380302358617225e-05,
+      "loss": 0.8401,
+      "step": 3036
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.537100057885465e-05,
+      "loss": 0.8825,
+      "step": 3037
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.536169706460211e-05,
+      "loss": 0.9894,
+      "step": 3038
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.535239181741555e-05,
+      "loss": 0.7753,
+      "step": 3039
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.534308483885124e-05,
+      "loss": 0.9167,
+      "step": 3040
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.53337761304657e-05,
+      "loss": 0.7685,
+      "step": 3041
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.532446569381578e-05,
+      "loss": 0.8525,
+      "step": 3042
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.531515353045857e-05,
+      "loss": 0.9052,
+      "step": 3043
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.53058396419515e-05,
+      "loss": 0.7709,
+      "step": 3044
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.529652402985225e-05,
+      "loss": 0.8274,
+      "step": 3045
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.52872066957188e-05,
+      "loss": 0.7072,
+      "step": 3046
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.527788764110943e-05,
+      "loss": 0.824,
+      "step": 3047
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.526856686758269e-05,
+      "loss": 0.7762,
+      "step": 3048
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.525924437669743e-05,
+      "loss": 0.7952,
+      "step": 3049
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.5249920170012785e-05,
+      "loss": 0.8536,
+      "step": 3050
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.5240594249088164e-05,
+      "loss": 0.9682,
+      "step": 3051
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.523126661548329e-05,
+      "loss": 0.8648,
+      "step": 3052
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.522193727075815e-05,
+      "loss": 0.7456,
+      "step": 3053
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.521260621647301e-05,
+      "loss": 0.9932,
+      "step": 3054
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.5203273454188446e-05,
+      "loss": 0.8399,
+      "step": 3055
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.5193938985465316e-05,
+      "loss": 0.8863,
+      "step": 3056
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.518460281186474e-05,
+      "loss": 0.9197,
+      "step": 3057
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.517526493494815e-05,
+      "loss": 0.8452,
+      "step": 3058
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.516592535627725e-05,
+      "loss": 0.8501,
+      "step": 3059
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.5156584077414025e-05,
+      "loss": 0.9554,
+      "step": 3060
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.514724109992076e-05,
+      "loss": 0.8857,
+      "step": 3061
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.513789642536001e-05,
+      "loss": 0.819,
+      "step": 3062
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.5128550055294607e-05,
+      "loss": 0.7397,
+      "step": 3063
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.511920199128769e-05,
+      "loss": 0.916,
+      "step": 3064
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.5109852234902676e-05,
+      "loss": 0.8723,
+      "step": 3065
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.5100500787703236e-05,
+      "loss": 0.8454,
+      "step": 3066
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.509114765125335e-05,
+      "loss": 0.715,
+      "step": 3067
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.50817928271173e-05,
+      "loss": 0.6448,
+      "step": 3068
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.5072436316859594e-05,
+      "loss": 1.0047,
+      "step": 3069
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.506307812204508e-05,
+      "loss": 0.7814,
+      "step": 3070
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.505371824423885e-05,
+      "loss": 0.858,
+      "step": 3071
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.5044356685006284e-05,
+      "loss": 0.8961,
+      "step": 3072
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.503499344591307e-05,
+      "loss": 0.7937,
+      "step": 3073
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.502562852852513e-05,
+      "loss": 0.9047,
+      "step": 3074
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.50162619344087e-05,
+      "loss": 0.7698,
+      "step": 3075
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.5006893665130304e-05,
+      "loss": 0.5742,
+      "step": 3076
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.4997523722256715e-05,
+      "loss": 1.038,
+      "step": 3077
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.4988152107355e-05,
+      "loss": 0.955,
+      "step": 3078
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.497877882199252e-05,
+      "loss": 0.9025,
+      "step": 3079
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.49694038677369e-05,
+      "loss": 1.0669,
+      "step": 3080
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.496002724615604e-05,
+      "loss": 0.9845,
+      "step": 3081
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.495064895881813e-05,
+      "loss": 0.779,
+      "step": 3082
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.494126900729163e-05,
+      "loss": 0.8545,
+      "step": 3083
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.493188739314529e-05,
+      "loss": 0.8883,
+      "step": 3084
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.492250411794813e-05,
+      "loss": 0.9135,
+      "step": 3085
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.4913119183269444e-05,
+      "loss": 1.0435,
+      "step": 3086
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.490373259067881e-05,
+      "loss": 0.765,
+      "step": 3087
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.4894344341746084e-05,
+      "loss": 0.824,
+      "step": 3088
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.488495443804139e-05,
+      "loss": 0.7906,
+      "step": 3089
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.4875562881135135e-05,
+      "loss": 0.875,
+      "step": 3090
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.486616967259802e-05,
+      "loss": 0.8311,
+      "step": 3091
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.4856774814000983e-05,
+      "loss": 0.9213,
+      "step": 3092
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.484737830691527e-05,
+      "loss": 0.8634,
+      "step": 3093
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.483798015291239e-05,
+      "loss": 0.9147,
+      "step": 3094
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.4828580353564135e-05,
+      "loss": 0.7738,
+      "step": 3095
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.481917891044256e-05,
+      "loss": 0.7575,
+      "step": 3096
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.480977582512001e-05,
+      "loss": 0.5948,
+      "step": 3097
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.480037109916908e-05,
+      "loss": 0.9221,
+      "step": 3098
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.479096473416268e-05,
+      "loss": 0.8967,
+      "step": 3099
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.478155673167395e-05,
+      "loss": 1.0373,
+      "step": 3100
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.477214709327634e-05,
+      "loss": 1.0166,
+      "step": 3101
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.476273582054355e-05,
+      "loss": 0.9877,
+      "step": 3102
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.4753322915049556e-05,
+      "loss": 0.9484,
+      "step": 3103
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.474390837836863e-05,
+      "loss": 1.0242,
+      "step": 3104
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.4734492212075273e-05,
+      "loss": 0.7763,
+      "step": 3105
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.47250744177443e-05,
+      "loss": 0.8874,
+      "step": 3106
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.4715654996950784e-05,
+      "loss": 0.5805,
+      "step": 3107
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.470623395127007e-05,
+      "loss": 0.7337,
+      "step": 3108
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.469681128227776e-05,
+      "loss": 0.9437,
+      "step": 3109
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.4687386991549756e-05,
+      "loss": 0.8751,
+      "step": 3110
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.4677961080662205e-05,
+      "loss": 0.8798,
+      "step": 3111
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.466853355119154e-05,
+      "loss": 0.7907,
+      "step": 3112
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.465910440471446e-05,
+      "loss": 0.9932,
+      "step": 3113
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.4649673642807944e-05,
+      "loss": 0.8422,
+      "step": 3114
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.464024126704923e-05,
+      "loss": 0.8353,
+      "step": 3115
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.463080727901582e-05,
+      "loss": 0.8036,
+      "step": 3116
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.462137168028549e-05,
+      "loss": 0.9672,
+      "step": 3117
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.4611934472436305e-05,
+      "loss": 0.9016,
+      "step": 3118
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.4602495657046567e-05,
+      "loss": 0.8689,
+      "step": 3119
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.459305523569489e-05,
+      "loss": 0.8371,
+      "step": 3120
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.45836132099601e-05,
+      "loss": 0.8496,
+      "step": 3121
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.457416958142133e-05,
+      "loss": 0.9665,
+      "step": 3122
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.456472435165799e-05,
+      "loss": 0.7968,
+      "step": 3123
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.455527752224972e-05,
+      "loss": 0.8879,
+      "step": 3124
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.454582909477645e-05,
+      "loss": 0.9014,
+      "step": 3125
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.4536379070818384e-05,
+      "loss": 0.8353,
+      "step": 3126
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.452692745195597e-05,
+      "loss": 0.8151,
+      "step": 3127
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.4517474239769953e-05,
+      "loss": 0.8617,
+      "step": 3128
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.4508019435841326e-05,
+      "loss": 0.8701,
+      "step": 3129
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.449856304175134e-05,
+      "loss": 0.8802,
+      "step": 3130
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.448910505908152e-05,
+      "loss": 1.0065,
+      "step": 3131
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.447964548941368e-05,
+      "loss": 0.7724,
+      "step": 3132
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.447018433432985e-05,
+      "loss": 0.801,
+      "step": 3133
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.4460721595412375e-05,
+      "loss": 0.8505,
+      "step": 3134
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.445125727424384e-05,
+      "loss": 0.8705,
+      "step": 3135
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.44417913724071e-05,
+      "loss": 0.9378,
+      "step": 3136
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.443232389148525e-05,
+      "loss": 0.9109,
+      "step": 3137
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.44228548330617e-05,
+      "loss": 0.7666,
+      "step": 3138
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.4413384198720083e-05,
+      "loss": 0.8481,
+      "step": 3139
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.440391199004431e-05,
+      "loss": 0.7733,
+      "step": 3140
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.439443820861855e-05,
+      "loss": 0.8903,
+      "step": 3141
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.4384962856027244e-05,
+      "loss": 0.8103,
+      "step": 3142
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.437548593385508e-05,
+      "loss": 0.7715,
+      "step": 3143
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.4366007443687034e-05,
+      "loss": 0.7845,
+      "step": 3144
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.435652738710831e-05,
+      "loss": 0.8449,
+      "step": 3145
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.4347045765704406e-05,
+      "loss": 0.6531,
+      "step": 3146
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.433756258106107e-05,
+      "loss": 0.8298,
+      "step": 3147
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.4328077834764296e-05,
+      "loss": 0.8932,
+      "step": 3148
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.4318591528400365e-05,
+      "loss": 0.7641,
+      "step": 3149
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.430910366355581e-05,
+      "loss": 0.8868,
+      "step": 3150
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.42996142418174e-05,
+      "loss": 0.8683,
+      "step": 3151
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.429012326477222e-05,
+      "loss": 0.8025,
+      "step": 3152
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.4280630734007555e-05,
+      "loss": 0.7384,
+      "step": 3153
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.427113665111098e-05,
+      "loss": 0.8773,
+      "step": 3154
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.426164101767034e-05,
+      "loss": 0.7704,
+      "step": 3155
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.425214383527371e-05,
+      "loss": 0.9644,
+      "step": 3156
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.424264510550943e-05,
+      "loss": 0.6836,
+      "step": 3157
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.423314482996614e-05,
+      "loss": 0.9491,
+      "step": 3158
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.4223643010232666e-05,
+      "loss": 0.7041,
+      "step": 3159
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.4214139647898166e-05,
+      "loss": 0.7086,
+      "step": 3160
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.420463474455201e-05,
+      "loss": 0.7976,
+      "step": 3161
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.4195128301783837e-05,
+      "loss": 0.9184,
+      "step": 3162
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.4185620321183545e-05,
+      "loss": 0.8591,
+      "step": 3163
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.417611080434129e-05,
+      "loss": 1.0393,
+      "step": 3164
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.416659975284748e-05,
+      "loss": 0.7026,
+      "step": 3165
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.415708716829279e-05,
+      "loss": 0.8786,
+      "step": 3166
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.4147573052268144e-05,
+      "loss": 0.7792,
+      "step": 3167
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.413805740636471e-05,
+      "loss": 0.7698,
+      "step": 3168
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.4128540232173946e-05,
+      "loss": 0.6601,
+      "step": 3169
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.411902153128753e-05,
+      "loss": 0.7037,
+      "step": 3170
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.410950130529741e-05,
+      "loss": 0.8828,
+      "step": 3171
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.4099979555795804e-05,
+      "loss": 0.7291,
+      "step": 3172
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3.4090456284375145e-05,
+      "loss": 0.8305,
+      "step": 3173
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.408093149262817e-05,
+      "loss": 0.9037,
+      "step": 3174
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.407140518214782e-05,
+      "loss": 0.8428,
+      "step": 3175
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.406187735452734e-05,
+      "loss": 0.8239,
+      "step": 3176
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.405234801136019e-05,
+      "loss": 0.8222,
+      "step": 3177
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.404281715424009e-05,
+      "loss": 0.8919,
+      "step": 3178
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.403328478476106e-05,
+      "loss": 0.8673,
+      "step": 3179
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.402375090451728e-05,
+      "loss": 0.7814,
+      "step": 3180
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.4014215515103264e-05,
+      "loss": 0.839,
+      "step": 3181
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.400467861811376e-05,
+      "loss": 0.7814,
+      "step": 3182
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.399514021514374e-05,
+      "loss": 0.7978,
+      "step": 3183
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.398560030778846e-05,
+      "loss": 0.8294,
+      "step": 3184
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.39760588976434e-05,
+      "loss": 0.8716,
+      "step": 3185
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.396651598630432e-05,
+      "loss": 0.9872,
+      "step": 3186
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.3956971575367215e-05,
+      "loss": 0.8076,
+      "step": 3187
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.3947425666428326e-05,
+      "loss": 0.8279,
+      "step": 3188
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3.393787826108415e-05,
+      "loss": 0.8565,
+      "step": 3189
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.392832936093144e-05,
+      "loss": 0.8341,
+      "step": 3190
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.3918778967567204e-05,
+      "loss": 0.8079,
+      "step": 3191
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.390922708258867e-05,
+      "loss": 0.7896,
+      "step": 3192
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.389967370759335e-05,
+      "loss": 0.7713,
+      "step": 3193
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.389011884417899e-05,
+      "loss": 0.8648,
+      "step": 3194
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.3880562493943575e-05,
+      "loss": 0.7286,
+      "step": 3195
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.3871004658485364e-05,
+      "loss": 0.7639,
+      "step": 3196
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.386144533940284e-05,
+      "loss": 1.0101,
+      "step": 3197
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.385188453829474e-05,
+      "loss": 1.0057,
+      "step": 3198
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.384232225676007e-05,
+      "loss": 0.997,
+      "step": 3199
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.3832758496398045e-05,
+      "loss": 0.7768,
+      "step": 3200
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.3823193258808165e-05,
+      "loss": 0.8796,
+      "step": 3201
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.381362654559016e-05,
+      "loss": 0.7622,
+      "step": 3202
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.380405835834399e-05,
+      "loss": 1.0321,
+      "step": 3203
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.37944886986699e-05,
+      "loss": 0.8033,
+      "step": 3204
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.378491756816836e-05,
+      "loss": 0.7743,
+      "step": 3205
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.377534496844007e-05,
+      "loss": 0.9059,
+      "step": 3206
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.3765770901085995e-05,
+      "loss": 0.7716,
+      "step": 3207
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.375619536770736e-05,
+      "loss": 0.9087,
+      "step": 3208
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.37466183699056e-05,
+      "loss": 0.9199,
+      "step": 3209
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.373703990928241e-05,
+      "loss": 0.7572,
+      "step": 3210
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.372745998743975e-05,
+      "loss": 0.7095,
+      "step": 3211
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.3717878605979796e-05,
+      "loss": 0.9076,
+      "step": 3212
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.370829576650497e-05,
+      "loss": 0.7871,
+      "step": 3213
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.369871147061797e-05,
+      "loss": 0.7356,
+      "step": 3214
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.36891257199217e-05,
+      "loss": 0.8968,
+      "step": 3215
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.367953851601931e-05,
+      "loss": 0.6896,
+      "step": 3216
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3.366994986051423e-05,
+      "loss": 0.7781,
+      "step": 3217
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.366035975501008e-05,
+      "loss": 0.6568,
+      "step": 3218
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.3650768201110774e-05,
+      "loss": 0.6497,
+      "step": 3219
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.3641175200420434e-05,
+      "loss": 1.0457,
+      "step": 3220
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.3631580754543424e-05,
+      "loss": 0.8356,
+      "step": 3221
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.362198486508438e-05,
+      "loss": 0.706,
+      "step": 3222
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.3612387533648146e-05,
+      "loss": 0.6636,
+      "step": 3223
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.3602788761839813e-05,
+      "loss": 0.7771,
+      "step": 3224
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.359318855126474e-05,
+      "loss": 0.9288,
+      "step": 3225
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.3583586903528496e-05,
+      "loss": 0.9164,
+      "step": 3226
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.357398382023689e-05,
+      "loss": 0.7517,
+      "step": 3227
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.356437930299601e-05,
+      "loss": 0.8065,
+      "step": 3228
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.355477335341213e-05,
+      "loss": 0.8928,
+      "step": 3229
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.35451659730918e-05,
+      "loss": 0.7509,
+      "step": 3230
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.35355571636418e-05,
+      "loss": 0.7326,
+      "step": 3231
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.352594692666915e-05,
+      "loss": 0.6971,
+      "step": 3232
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3.35163352637811e-05,
+      "loss": 0.9969,
+      "step": 3233
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.350672217658515e-05,
+      "loss": 0.885,
+      "step": 3234
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.349710766668903e-05,
+      "loss": 0.8508,
+      "step": 3235
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.348749173570071e-05,
+      "loss": 0.832,
+      "step": 3236
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.3477874385228405e-05,
+      "loss": 1.0177,
+      "step": 3237
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.346825561688055e-05,
+      "loss": 0.8385,
+      "step": 3238
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.345863543226585e-05,
+      "loss": 0.8422,
+      "step": 3239
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.344901383299321e-05,
+      "loss": 0.9471,
+      "step": 3240
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.343939082067178e-05,
+      "loss": 1.1117,
+      "step": 3241
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.342976639691096e-05,
+      "loss": 0.5727,
+      "step": 3242
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.342014056332039e-05,
+      "loss": 0.8928,
+      "step": 3243
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.3410513321509926e-05,
+      "loss": 0.8515,
+      "step": 3244
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3.340088467308966e-05,
+      "loss": 0.6724,
+      "step": 3245
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.3391254619669945e-05,
+      "loss": 0.6958,
+      "step": 3246
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.3381623162861344e-05,
+      "loss": 0.9585,
+      "step": 3247
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.3371990304274656e-05,
+      "loss": 0.8789,
+      "step": 3248
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.336235604552093e-05,
+      "loss": 0.9258,
+      "step": 3249
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.335272038821145e-05,
+      "loss": 0.9095,
+      "step": 3250
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.3343083333957684e-05,
+      "loss": 0.8471,
+      "step": 3251
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.333344488437142e-05,
+      "loss": 0.902,
+      "step": 3252
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.332380504106461e-05,
+      "loss": 0.9037,
+      "step": 3253
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.3314163805649464e-05,
+      "loss": 0.8398,
+      "step": 3254
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.3304521179738437e-05,
+      "loss": 0.9487,
+      "step": 3255
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.329487716494418e-05,
+      "loss": 0.8182,
+      "step": 3256
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.328523176287961e-05,
+      "loss": 0.6667,
+      "step": 3257
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.327558497515787e-05,
+      "loss": 0.9958,
+      "step": 3258
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.3265936803392314e-05,
+      "loss": 0.9849,
+      "step": 3259
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.325628724919656e-05,
+      "loss": 0.7223,
+      "step": 3260
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3.324663631418443e-05,
+      "loss": 0.8248,
+      "step": 3261
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.323698399996999e-05,
+      "loss": 0.8575,
+      "step": 3262
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.3227330308167535e-05,
+      "loss": 0.9424,
+      "step": 3263
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.3217675240391586e-05,
+      "loss": 0.7414,
+      "step": 3264
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.320801879825691e-05,
+      "loss": 0.8322,
+      "step": 3265
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.319836098337847e-05,
+      "loss": 1.0953,
+      "step": 3266
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.31887017973715e-05,
+      "loss": 0.9258,
+      "step": 3267
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3.317904124185142e-05,
+      "loss": 0.8798,
+      "step": 3268
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.316937931843392e-05,
+      "loss": 0.812,
+      "step": 3269
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.315971602873491e-05,
+      "loss": 0.6363,
+      "step": 3270
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.3150051374370496e-05,
+      "loss": 0.6479,
+      "step": 3271
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.314038535695704e-05,
+      "loss": 0.9256,
+      "step": 3272
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.3130717978111144e-05,
+      "loss": 0.8957,
+      "step": 3273
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.312104923944961e-05,
+      "loss": 0.8604,
+      "step": 3274
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.311137914258947e-05,
+      "loss": 0.8889,
+      "step": 3275
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.310170768914801e-05,
+      "loss": 0.6598,
+      "step": 3276
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.3092034880742704e-05,
+      "loss": 0.9883,
+      "step": 3277
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.3082360718991304e-05,
+      "loss": 0.6932,
+      "step": 3278
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.3072685205511726e-05,
+      "loss": 0.7149,
+      "step": 3279
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.3063008341922156e-05,
+      "loss": 0.7717,
+      "step": 3280
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.305333012984101e-05,
+      "loss": 0.7886,
+      "step": 3281
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.3043650570886884e-05,
+      "loss": 0.8381,
+      "step": 3282
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.303396966667864e-05,
+      "loss": 0.8943,
+      "step": 3283
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.302428741883536e-05,
+      "loss": 0.7683,
+      "step": 3284
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.301460382897634e-05,
+      "loss": 0.761,
+      "step": 3285
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.3004918898721096e-05,
+      "loss": 0.7163,
+      "step": 3286
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.29952326296894e-05,
+      "loss": 0.6751,
+      "step": 3287
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3.298554502350119e-05,
+      "loss": 0.7195,
+      "step": 3288
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.297585608177669e-05,
+      "loss": 0.9076,
+      "step": 3289
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.296616580613632e-05,
+      "loss": 0.8986,
+      "step": 3290
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.29564741982007e-05,
+      "loss": 0.6872,
+      "step": 3291
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.294678125959071e-05,
+      "loss": 1.0035,
+      "step": 3292
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.293708699192745e-05,
+      "loss": 0.6986,
+      "step": 3293
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.29273913968322e-05,
+      "loss": 1.0097,
+      "step": 3294
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.291769447592652e-05,
+      "loss": 0.9753,
+      "step": 3295
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3.290799623083215e-05,
+      "loss": 0.8447,
+      "step": 3296
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.289829666317107e-05,
+      "loss": 1.0025,
+      "step": 3297
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.288859577456547e-05,
+      "loss": 0.6452,
+      "step": 3298
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.287889356663778e-05,
+      "loss": 0.9139,
+      "step": 3299
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.286919004101063e-05,
+      "loss": 0.768,
+      "step": 3300
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.2859485199306885e-05,
+      "loss": 0.823,
+      "step": 3301
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.284977904314961e-05,
+      "loss": 0.9165,
+      "step": 3302
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.2840071574162115e-05,
+      "loss": 0.8138,
+      "step": 3303
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.283036279396792e-05,
+      "loss": 0.6097,
+      "step": 3304
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.282065270419076e-05,
+      "loss": 0.8164,
+      "step": 3305
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.281094130645458e-05,
+      "loss": 0.9914,
+      "step": 3306
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.280122860238357e-05,
+      "loss": 0.8101,
+      "step": 3307
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.279151459360211e-05,
+      "loss": 0.7984,
+      "step": 3308
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.2781799281734836e-05,
+      "loss": 0.8725,
+      "step": 3309
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.277208266840655e-05,
+      "loss": 0.8346,
+      "step": 3310
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.276236475524232e-05,
+      "loss": 0.9946,
+      "step": 3311
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3.2752645543867386e-05,
+      "loss": 0.7857,
+      "step": 3312
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.274292503590727e-05,
+      "loss": 0.9113,
+      "step": 3313
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.2733203232987635e-05,
+      "loss": 0.8029,
+      "step": 3314
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.2723480136734406e-05,
+      "loss": 0.7053,
+      "step": 3315
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.271375574877372e-05,
+      "loss": 0.947,
+      "step": 3316
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.2704030070731926e-05,
+      "loss": 0.7179,
+      "step": 3317
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.269430310423558e-05,
+      "loss": 0.8938,
+      "step": 3318
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.268457485091146e-05,
+      "loss": 0.7965,
+      "step": 3319
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.267484531238657e-05,
+      "loss": 1.0398,
+      "step": 3320
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.266511449028812e-05,
+      "loss": 0.6789,
+      "step": 3321
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.2655382386243524e-05,
+      "loss": 0.877,
+      "step": 3322
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.264564900188042e-05,
+      "loss": 0.9327,
+      "step": 3323
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3.263591433882666e-05,
+      "loss": 0.8807,
+      "step": 3324
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.262617839871033e-05,
+      "loss": 0.8445,
+      "step": 3325
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.2616441183159674e-05,
+      "loss": 0.8225,
+      "step": 3326
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.2606702693803216e-05,
+      "loss": 0.9252,
+      "step": 3327
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.259696293226965e-05,
+      "loss": 0.8566,
+      "step": 3328
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.2587221900187906e-05,
+      "loss": 0.857,
+      "step": 3329
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.2577479599187096e-05,
+      "loss": 0.7992,
+      "step": 3330
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.2577479599187096e-05,
+      "loss": 0.9514,
+      "step": 3331
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.256773603089657e-05,
+      "loss": 0.8798,
+      "step": 3332
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.255799119694589e-05,
+      "loss": 1.0809,
+      "step": 3333
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.254824509896483e-05,
+      "loss": 0.7245,
+      "step": 3334
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.253849773858335e-05,
+      "loss": 0.6721,
+      "step": 3335
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.2528749117431645e-05,
+      "loss": 0.8306,
+      "step": 3336
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.251899923714012e-05,
+      "loss": 0.6017,
+      "step": 3337
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.250924809933939e-05,
+      "loss": 0.7916,
+      "step": 3338
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.249949570566027e-05,
+      "loss": 0.7506,
+      "step": 3339
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3.2489742057733796e-05,
+      "loss": 0.7974,
+      "step": 3340
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.24799871571912e-05,
+      "loss": 0.7321,
+      "step": 3341
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.2470231005663944e-05,
+      "loss": 0.8931,
+      "step": 3342
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.2460473604783684e-05,
+      "loss": 0.7316,
+      "step": 3343
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.245071495618228e-05,
+      "loss": 0.8685,
+      "step": 3344
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.244095506149183e-05,
+      "loss": 0.7433,
+      "step": 3345
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.243119392234461e-05,
+      "loss": 0.9198,
+      "step": 3346
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.24214315403731e-05,
+      "loss": 0.9153,
+      "step": 3347
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3.241166791721001e-05,
+      "loss": 0.8192,
+      "step": 3348
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.240190305448827e-05,
+      "loss": 0.7667,
+      "step": 3349
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.239213695384097e-05,
+      "loss": 0.8098,
+      "step": 3350
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.238236961690144e-05,
+      "loss": 0.9224,
+      "step": 3351
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.237260104530323e-05,
+      "loss": 0.8347,
+      "step": 3352
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.236283124068006e-05,
+      "loss": 0.7019,
+      "step": 3353
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.2353060204665865e-05,
+      "loss": 0.8637,
+      "step": 3354
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.234328793889482e-05,
+      "loss": 0.623,
+      "step": 3355
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3.233351444500125e-05,
+      "loss": 1.0941,
+      "step": 3356
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.232373972461974e-05,
+      "loss": 0.8475,
+      "step": 3357
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.231396377938505e-05,
+      "loss": 0.9613,
+      "step": 3358
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.230418661093214e-05,
+      "loss": 0.9568,
+      "step": 3359
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.2294408220896186e-05,
+      "loss": 0.8349,
+      "step": 3360
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.228462861091259e-05,
+      "loss": 0.8812,
+      "step": 3361
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.2274847782616905e-05,
+      "loss": 0.7527,
+      "step": 3362
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.2265065737644925e-05,
+      "loss": 0.7867,
+      "step": 3363
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.2255282477632657e-05,
+      "loss": 0.8178,
+      "step": 3364
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.224549800421628e-05,
+      "loss": 0.9198,
+      "step": 3365
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.223571231903221e-05,
+      "loss": 0.8372,
+      "step": 3366
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3.222592542371702e-05,
+      "loss": 0.8894,
+      "step": 3367
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.221613731990752e-05,
+      "loss": 0.7325,
+      "step": 3368
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.2206348009240724e-05,
+      "loss": 0.8019,
+      "step": 3369
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.2196557493353824e-05,
+      "loss": 0.9266,
+      "step": 3370
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.2186765773884245e-05,
+      "loss": 0.879,
+      "step": 3371
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.2176972852469575e-05,
+      "loss": 0.823,
+      "step": 3372
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.2167178730747635e-05,
+      "loss": 0.7796,
+      "step": 3373
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.215738341035643e-05,
+      "loss": 0.7838,
+      "step": 3374
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3.214758689293418e-05,
+      "loss": 0.7564,
+      "step": 3375
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.213778918011929e-05,
+      "loss": 0.8079,
+      "step": 3376
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.2127990273550375e-05,
+      "loss": 0.8589,
+      "step": 3377
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.211819017486624e-05,
+      "loss": 0.9127,
+      "step": 3378
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.210838888570589e-05,
+      "loss": 0.9406,
+      "step": 3379
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.209858640770854e-05,
+      "loss": 0.595,
+      "step": 3380
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.208878274251361e-05,
+      "loss": 0.6554,
+      "step": 3381
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.207897789176069e-05,
+      "loss": 0.7774,
+      "step": 3382
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3.206917185708959e-05,
+      "loss": 0.8665,
+      "step": 3383
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.205936464014031e-05,
+      "loss": 0.8784,
+      "step": 3384
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.204955624255305e-05,
+      "loss": 0.8414,
+      "step": 3385
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.203974666596821e-05,
+      "loss": 0.8352,
+      "step": 3386
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.20299359120264e-05,
+      "loss": 0.87,
+      "step": 3387
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.202012398236838e-05,
+      "loss": 0.6693,
+      "step": 3388
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.2010310878635165e-05,
+      "loss": 0.7882,
+      "step": 3389
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.2000496602467934e-05,
+      "loss": 1.0009,
+      "step": 3390
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3.199068115550806e-05,
+      "loss": 0.8615,
+      "step": 3391
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.198086453939712e-05,
+      "loss": 0.8557,
+      "step": 3392
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.19710467557769e-05,
+      "loss": 0.898,
+      "step": 3393
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.196122780628936e-05,
+      "loss": 0.8398,
+      "step": 3394
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.195140769257665e-05,
+      "loss": 0.8261,
+      "step": 3395
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.1941586416281146e-05,
+      "loss": 0.8246,
+      "step": 3396
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.19317639790454e-05,
+      "loss": 0.7493,
+      "step": 3397
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.192194038251214e-05,
+      "loss": 0.8634,
+      "step": 3398
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.191211562832432e-05,
+      "loss": 0.7914,
+      "step": 3399
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.1902289718125074e-05,
+      "loss": 0.6338,
+      "step": 3400
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.189246265355772e-05,
+      "loss": 0.9231,
+      "step": 3401
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.188263443626578e-05,
+      "loss": 0.8061,
+      "step": 3402
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3.1872805067892975e-05,
+      "loss": 0.9149,
+      "step": 3403
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.1862974550083205e-05,
+      "loss": 0.7266,
+      "step": 3404
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.1853142884480555e-05,
+      "loss": 0.884,
+      "step": 3405
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.184331007272934e-05,
+      "loss": 0.7362,
+      "step": 3406
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.183347611647402e-05,
+      "loss": 0.6749,
+      "step": 3407
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.182364101735927e-05,
+      "loss": 0.8307,
+      "step": 3408
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.1813804777029965e-05,
+      "loss": 0.9514,
+      "step": 3409
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.180396739713114e-05,
+      "loss": 0.7083,
+      "step": 3410
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3.179412887930807e-05,
+      "loss": 0.8668,
+      "step": 3411
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.178428922520616e-05,
+      "loss": 0.8288,
+      "step": 3412
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.1774448436471046e-05,
+      "loss": 0.966,
+      "step": 3413
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.1764606514748557e-05,
+      "loss": 0.8869,
+      "step": 3414
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.1754763461684674e-05,
+      "loss": 0.6989,
+      "step": 3415
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.1744919278925605e-05,
+      "loss": 0.7987,
+      "step": 3416
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.173507396811774e-05,
+      "loss": 0.9205,
+      "step": 3417
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.172522753090763e-05,
+      "loss": 0.9102,
+      "step": 3418
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3.1715379968942036e-05,
+      "loss": 0.8104,
+      "step": 3419
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.170553128386794e-05,
+      "loss": 0.8928,
+      "step": 3420
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.169568147733243e-05,
+      "loss": 0.8364,
+      "step": 3421
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.168583055098286e-05,
+      "loss": 0.7189,
+      "step": 3422
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.167597850646673e-05,
+      "loss": 0.6344,
+      "step": 3423
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.1666125345431745e-05,
+      "loss": 0.9104,
+      "step": 3424
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.1656271069525783e-05,
+      "loss": 0.8309,
+      "step": 3425
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.164641568039691e-05,
+      "loss": 0.6985,
+      "step": 3426
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.16365591796934e-05,
+      "loss": 0.7929,
+      "step": 3427
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.162670156906368e-05,
+      "loss": 0.778,
+      "step": 3428
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.161684285015638e-05,
+      "loss": 0.6522,
+      "step": 3429
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.160698302462032e-05,
+      "loss": 0.8092,
+      "step": 3430
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.1597122094104496e-05,
+      "loss": 0.94,
+      "step": 3431
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.158726006025809e-05,
+      "loss": 0.7997,
+      "step": 3432
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.1577396924730476e-05,
+      "loss": 0.8346,
+      "step": 3433
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.15675326891712e-05,
+      "loss": 0.9502,
+      "step": 3434
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.155766735523e-05,
+      "loss": 0.7266,
+      "step": 3435
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.15478009245568e-05,
+      "loss": 0.872,
+      "step": 3436
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.15379333988017e-05,
+      "loss": 0.7136,
+      "step": 3437
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.152806477961498e-05,
+      "loss": 0.7977,
+      "step": 3438
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.151819506864711e-05,
+      "loss": 0.8549,
+      "step": 3439
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.150832426754877e-05,
+      "loss": 0.7228,
+      "step": 3440
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.1498452377970765e-05,
+      "loss": 0.7831,
+      "step": 3441
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.1488579401564115e-05,
+      "loss": 0.6538,
+      "step": 3442
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.147870533998004e-05,
+      "loss": 0.7789,
+      "step": 3443
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.1468830194869885e-05,
+      "loss": 0.7327,
+      "step": 3444
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.145895396788523e-05,
+      "loss": 0.7417,
+      "step": 3445
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.144907666067783e-05,
+      "loss": 0.8166,
+      "step": 3446
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.143919827489958e-05,
+      "loss": 0.9272,
+      "step": 3447
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.1429318812202603e-05,
+      "loss": 0.8208,
+      "step": 3448
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.141943827423918e-05,
+      "loss": 0.7013,
+      "step": 3449
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.140955666266176e-05,
+      "loss": 0.9533,
+      "step": 3450
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.1399673979122995e-05,
+      "loss": 0.8013,
+      "step": 3451
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.138979022527572e-05,
+      "loss": 0.844,
+      "step": 3452
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.137990540277291e-05,
+      "loss": 0.8831,
+      "step": 3453
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.137001951326776e-05,
+      "loss": 0.7932,
+      "step": 3454
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.136013255841363e-05,
+      "loss": 0.8029,
+      "step": 3455
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.135024453986404e-05,
+      "loss": 0.9132,
+      "step": 3456
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.134035545927272e-05,
+      "loss": 0.8436,
+      "step": 3457
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.133046531829357e-05,
+      "loss": 0.7773,
+      "step": 3458
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.132057411858063e-05,
+      "loss": 0.6979,
+      "step": 3459
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.131068186178817e-05,
+      "loss": 0.8667,
+      "step": 3460
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.1300788549570606e-05,
+      "loss": 0.8255,
+      "step": 3461
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.129089418358253e-05,
+      "loss": 0.8035,
+      "step": 3462
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.1280998765478727e-05,
+      "loss": 0.6782,
+      "step": 3463
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.127110229691414e-05,
+      "loss": 0.5994,
+      "step": 3464
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.1261204779543906e-05,
+      "loss": 0.9491,
+      "step": 3465
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.1251306215023325e-05,
+      "loss": 0.7302,
+      "step": 3466
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.124140660500787e-05,
+      "loss": 0.8112,
+      "step": 3467
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.123150595115319e-05,
+      "loss": 0.8501,
+      "step": 3468
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.122160425511513e-05,
+      "loss": 0.7542,
+      "step": 3469
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.121170151854968e-05,
+      "loss": 0.8425,
+      "step": 3470
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.120179774311302e-05,
+      "loss": 0.7175,
+      "step": 3471
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.119189293046148e-05,
+      "loss": 0.5137,
+      "step": 3472
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.11819870822516e-05,
+      "loss": 0.9991,
+      "step": 3473
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.117208020014007e-05,
+      "loss": 0.9075,
+      "step": 3474
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.116217228578378e-05,
+      "loss": 0.8506,
+      "step": 3475
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.1152263340839745e-05,
+      "loss": 1.0247,
+      "step": 3476
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.114235336696518e-05,
+      "loss": 0.9409,
+      "step": 3477
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.113244236581749e-05,
+      "loss": 0.7345,
+      "step": 3478
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.112253033905422e-05,
+      "loss": 0.8134,
+      "step": 3479
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.111261728833309e-05,
+      "loss": 0.8534,
+      "step": 3480
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.1102703215312026e-05,
+      "loss": 0.8758,
+      "step": 3481
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.109278812164907e-05,
+      "loss": 1.0078,
+      "step": 3482
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.108287200900249e-05,
+      "loss": 0.725,
+      "step": 3483
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.107295487903069e-05,
+      "loss": 0.7771,
+      "step": 3484
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.106303673339224e-05,
+      "loss": 0.7549,
+      "step": 3485
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.10531175737459e-05,
+      "loss": 0.8272,
+      "step": 3486
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.1043197401750606e-05,
+      "loss": 0.786,
+      "step": 3487
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.103327621906543e-05,
+      "loss": 0.8748,
+      "step": 3488
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.102335402734964e-05,
+      "loss": 0.8205,
+      "step": 3489
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.1013430828262676e-05,
+      "loss": 0.8637,
+      "step": 3490
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.1003506623464115e-05,
+      "loss": 0.7245,
+      "step": 3491
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.099358141461374e-05,
+      "loss": 0.7035,
+      "step": 3492
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.0983655203371475e-05,
+      "loss": 0.547,
+      "step": 3493
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.0973727991397426e-05,
+      "loss": 0.8871,
+      "step": 3494
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.0963799780351865e-05,
+      "loss": 0.8561,
+      "step": 3495
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.095387057189522e-05,
+      "loss": 0.9919,
+      "step": 3496
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.09439403676881e-05,
+      "loss": 0.9882,
+      "step": 3497
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.093400916939128e-05,
+      "loss": 0.9513,
+      "step": 3498
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.0924076978665674e-05,
+      "loss": 0.911,
+      "step": 3499
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.09141437971724e-05,
+      "loss": 0.9843,
+      "step": 3500
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.090420962657273e-05,
+      "loss": 0.732,
+      "step": 3501
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.0894274468528084e-05,
+      "loss": 0.8431,
+      "step": 3502
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.088433832470006e-05,
+      "loss": 0.5262,
+      "step": 3503
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.087440119675043e-05,
+      "loss": 0.696,
+      "step": 3504
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.086446308634112e-05,
+      "loss": 0.9017,
+      "step": 3505
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.0854523995134216e-05,
+      "loss": 0.843,
+      "step": 3506
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.084458392479198e-05,
+      "loss": 0.8382,
+      "step": 3507
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.083464287697683e-05,
+      "loss": 0.7506,
+      "step": 3508
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.082470085335133e-05,
+      "loss": 0.9573,
+      "step": 3509
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.081475785557826e-05,
+      "loss": 0.7902,
+      "step": 3510
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.0804813885320506e-05,
+      "loss": 0.7914,
+      "step": 3511
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.0794868944241136e-05,
+      "loss": 0.7487,
+      "step": 3512
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.078492303400341e-05,
+      "loss": 0.9339,
+      "step": 3513
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.07749761562707e-05,
+      "loss": 0.8616,
+      "step": 3514
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.0765028312706564e-05,
+      "loss": 0.8244,
+      "step": 3515
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.075507950497474e-05,
+      "loss": 0.8064,
+      "step": 3516
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.074512973473908e-05,
+      "loss": 0.8025,
+      "step": 3517
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.0735179003663654e-05,
+      "loss": 0.9306,
+      "step": 3518
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.0725227313412655e-05,
+      "loss": 0.7431,
+      "step": 3519
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.071527466565044e-05,
+      "loss": 0.8461,
+      "step": 3520
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.0705321062041534e-05,
+      "loss": 0.8466,
+      "step": 3521
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.069536650425063e-05,
+      "loss": 0.7939,
+      "step": 3522
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.0685410993942555e-05,
+      "loss": 0.7668,
+      "step": 3523
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.0675454532782317e-05,
+      "loss": 0.8092,
+      "step": 3524
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.066549712243508e-05,
+      "loss": 0.8237,
+      "step": 3525
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.0655538764566154e-05,
+      "loss": 0.8371,
+      "step": 3526
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.0655538764566154e-05,
+      "loss": 0.9655,
+      "step": 3527
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.064557946084103e-05,
+      "loss": 0.7246,
+      "step": 3528
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.0635619212925334e-05,
+      "loss": 0.7579,
+      "step": 3529
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.062565802248486e-05,
+      "loss": 0.8002,
+      "step": 3530
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.061569589118557e-05,
+      "loss": 0.8215,
+      "step": 3531
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.060573282069356e-05,
+      "loss": 0.9007,
+      "step": 3532
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.0595768812675104e-05,
+      "loss": 0.8782,
+      "step": 3533
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.058580386879663e-05,
+      "loss": 0.7209,
+      "step": 3534
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.0575837990724695e-05,
+      "loss": 0.8109,
+      "step": 3535
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.0565871180126055e-05,
+      "loss": 0.7217,
+      "step": 3536
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.05559034386676e-05,
+      "loss": 0.8577,
+      "step": 3537
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.054593476801637e-05,
+      "loss": 0.7583,
+      "step": 3538
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.053596516983957e-05,
+      "loss": 0.7331,
+      "step": 3539
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.052599464580456e-05,
+      "loss": 0.7405,
+      "step": 3540
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.051602319757885e-05,
+      "loss": 0.7924,
+      "step": 3541
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.0506050826830103e-05,
+      "loss": 0.5972,
+      "step": 3542
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.0496077535226158e-05,
+      "loss": 0.7832,
+      "step": 3543
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.0486103324434972e-05,
+      "loss": 0.8498,
+      "step": 3544
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.0476128196124676e-05,
+      "loss": 0.7225,
+      "step": 3545
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.046615215196356e-05,
+      "loss": 0.8438,
+      "step": 3546
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.0456175193620063e-05,
+      "loss": 0.8228,
+      "step": 3547
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.0446197322762753e-05,
+      "loss": 0.7624,
+      "step": 3548
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.0436218541060402e-05,
+      "loss": 0.6891,
+      "step": 3549
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.0426238850181878e-05,
+      "loss": 0.8378,
+      "step": 3550
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.041625825179623e-05,
+      "loss": 0.719,
+      "step": 3551
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.0406276747572677e-05,
+      "loss": 0.9226,
+      "step": 3552
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.0396294339180536e-05,
+      "loss": 0.6335,
+      "step": 3553
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.0386311028289333e-05,
+      "loss": 0.9092,
+      "step": 3554
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.0376326816568713e-05,
+      "loss": 0.6648,
+      "step": 3555
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.0366341705688468e-05,
+      "loss": 0.6555,
+      "step": 3556
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.0356355697318556e-05,
+      "loss": 0.7519,
+      "step": 3557
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.0346368793129082e-05,
+      "loss": 0.8811,
+      "step": 3558
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.033638099479029e-05,
+      "loss": 0.8111,
+      "step": 3559
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.0326392303972596e-05,
+      "loss": 0.9869,
+      "step": 3560
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.0316402722346542e-05,
+      "loss": 0.6504,
+      "step": 3561
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.0306412251582827e-05,
+      "loss": 0.8306,
+      "step": 3562
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0296420893352295e-05,
+      "loss": 0.7347,
+      "step": 3563
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0286428649325956e-05,
+      "loss": 0.7067,
+      "step": 3564
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0276435521174944e-05,
+      "loss": 0.603,
+      "step": 3565
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.0266441510570553e-05,
+      "loss": 0.6595,
+      "step": 3566
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.0256446619184232e-05,
+      "loss": 0.8565,
+      "step": 3567
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.0246450848687557e-05,
+      "loss": 0.6835,
+      "step": 3568
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.0236454200752272e-05,
+      "loss": 0.7825,
+      "step": 3569
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.0226456677050263e-05,
+      "loss": 0.8667,
+      "step": 3570
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.0216458279253545e-05,
+      "loss": 0.7933,
+      "step": 3571
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.0206459009034295e-05,
+      "loss": 0.7822,
+      "step": 3572
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.0196458868064836e-05,
+      "loss": 0.7713,
+      "step": 3573
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.0186457858017642e-05,
+      "loss": 0.8523,
+      "step": 3574
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.017645598056531e-05,
+      "loss": 0.8308,
+      "step": 3575
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.0166453237380605e-05,
+      "loss": 0.7388,
+      "step": 3576
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.0156449630136418e-05,
+      "loss": 0.7897,
+      "step": 3577
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.0146445160505814e-05,
+      "loss": 0.736,
+      "step": 3578
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.0136439830161967e-05,
+      "loss": 0.7526,
+      "step": 3579
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.0126433640778213e-05,
+      "loss": 0.7866,
+      "step": 3580
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.0116426594028034e-05,
+      "loss": 0.8227,
+      "step": 3581
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.0106418691585047e-05,
+      "loss": 0.9458,
+      "step": 3582
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.009640993512301e-05,
+      "loss": 0.7654,
+      "step": 3583
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.008640032631585e-05,
+      "loss": 0.7795,
+      "step": 3584
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.007638986683759e-05,
+      "loss": 0.8116,
+      "step": 3585
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.006637855836244e-05,
+      "loss": 0.7775,
+      "step": 3586
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0056366402564734e-05,
+      "loss": 0.7617,
+      "step": 3587
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0046353401118933e-05,
+      "loss": 0.7429,
+      "step": 3588
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.003633955569966e-05,
+      "loss": 0.718,
+      "step": 3589
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0026324867981688e-05,
+      "loss": 0.8237,
+      "step": 3590
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0016309339639887e-05,
+      "loss": 0.6859,
+      "step": 3591
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.000629297234932e-05,
+      "loss": 0.7264,
+      "step": 3592
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.9996275767785152e-05,
+      "loss": 0.9671,
+      "step": 3593
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.998625772762272e-05,
+      "loss": 0.9868,
+      "step": 3594
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9976238853537462e-05,
+      "loss": 0.9625,
+      "step": 3595
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9966219147204995e-05,
+      "loss": 0.7317,
+      "step": 3596
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9956198610301046e-05,
+      "loss": 0.8369,
+      "step": 3597
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.994617724450149e-05,
+      "loss": 0.7149,
+      "step": 3598
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.993615505148235e-05,
+      "loss": 0.9966,
+      "step": 3599
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9926132032919785e-05,
+      "loss": 0.7634,
+      "step": 3600
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.991610819049007e-05,
+      "loss": 0.7271,
+      "step": 3601
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.990608352586965e-05,
+      "loss": 0.8597,
+      "step": 3602
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9896058040735087e-05,
+      "loss": 0.7305,
+      "step": 3603
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.988603173676308e-05,
+      "loss": 0.8698,
+      "step": 3604
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9876004615630482e-05,
+      "loss": 0.8726,
+      "step": 3605
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.986597667901426e-05,
+      "loss": 0.7206,
+      "step": 3606
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.985594792859154e-05,
+      "loss": 0.6722,
+      "step": 3607
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.984591836603956e-05,
+      "loss": 0.8603,
+      "step": 3608
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9835887993035706e-05,
+      "loss": 0.7461,
+      "step": 3609
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.982585681125752e-05,
+      "loss": 0.6916,
+      "step": 3610
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.9815824822382637e-05,
+      "loss": 0.8593,
+      "step": 3611
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.980579202808886e-05,
+      "loss": 0.634,
+      "step": 3612
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.9795758430054117e-05,
+      "loss": 0.7211,
+      "step": 3613
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.9785724029956463e-05,
+      "loss": 0.6121,
+      "step": 3614
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.977568882947409e-05,
+      "loss": 0.6034,
+      "step": 3615
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.9765652830285345e-05,
+      "loss": 1.0161,
+      "step": 3616
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.975561603406867e-05,
+      "loss": 0.805,
+      "step": 3617
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.9745578442502675e-05,
+      "loss": 0.6624,
+      "step": 3618
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.9735540057266088e-05,
+      "loss": 0.6135,
+      "step": 3619
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.972550088003776e-05,
+      "loss": 0.7365,
+      "step": 3620
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.97154609124967e-05,
+      "loss": 0.8763,
+      "step": 3621
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.9705420156322034e-05,
+      "loss": 0.8757,
+      "step": 3622
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.9695378613193002e-05,
+      "loss": 0.693,
+      "step": 3623
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.968533628478901e-05,
+      "loss": 0.7728,
+      "step": 3624
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.9675293172789583e-05,
+      "loss": 0.8544,
+      "step": 3625
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.966524927887436e-05,
+      "loss": 0.7004,
+      "step": 3626
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.965520460472313e-05,
+      "loss": 0.6881,
+      "step": 3627
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.964515915201581e-05,
+      "loss": 0.6401,
+      "step": 3628
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.9635112922432445e-05,
+      "loss": 0.955,
+      "step": 3629
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.9625065917653204e-05,
+      "loss": 0.8412,
+      "step": 3630
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.961501813935839e-05,
+      "loss": 0.8,
+      "step": 3631
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.960496958922844e-05,
+      "loss": 0.7775,
+      "step": 3632
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.9594920268943916e-05,
+      "loss": 0.9801,
+      "step": 3633
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.9584870180185504e-05,
+      "loss": 0.7957,
+      "step": 3634
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.957481932463403e-05,
+      "loss": 0.794,
+      "step": 3635
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.9564767703970426e-05,
+      "loss": 0.9101,
+      "step": 3636
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.9554715319875793e-05,
+      "loss": 1.0745,
+      "step": 3637
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.9544662174031313e-05,
+      "loss": 0.5215,
+      "step": 3638
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.9534608268118324e-05,
+      "loss": 0.8407,
+      "step": 3639
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.952455360381829e-05,
+      "loss": 0.8063,
+      "step": 3640
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.9514498182812784e-05,
+      "loss": 0.6151,
+      "step": 3641
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.9504442006783517e-05,
+      "loss": 0.644,
+      "step": 3642
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.9494385077412346e-05,
+      "loss": 0.9138,
+      "step": 3643
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.9484327396381206e-05,
+      "loss": 0.8338,
+      "step": 3644
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.947426896537221e-05,
+      "loss": 0.8895,
+      "step": 3645
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.946420978606756e-05,
+      "loss": 0.8672,
+      "step": 3646
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.9454149860149595e-05,
+      "loss": 0.8024,
+      "step": 3647
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.9444089189300783e-05,
+      "loss": 0.8571,
+      "step": 3648
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.943402777520372e-05,
+      "loss": 0.8523,
+      "step": 3649
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.9423965619541106e-05,
+      "loss": 0.7953,
+      "step": 3650
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.9413902723995786e-05,
+      "loss": 0.9062,
+      "step": 3651
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.9403839090250718e-05,
+      "loss": 0.7663,
+      "step": 3652
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.939377471998899e-05,
+      "loss": 0.6177,
+      "step": 3653
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.9383709614893816e-05,
+      "loss": 0.9606,
+      "step": 3654
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.9373643776648515e-05,
+      "loss": 0.9441,
+      "step": 3655
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.9363577206936537e-05,
+      "loss": 0.6669,
+      "step": 3656
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.9353509907441474e-05,
+      "loss": 0.7862,
+      "step": 3657
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.934344187984701e-05,
+      "loss": 0.8221,
+      "step": 3658
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.9333373125836972e-05,
+      "loss": 0.9023,
+      "step": 3659
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.93233036470953e-05,
+      "loss": 0.6912,
+      "step": 3660
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.9313233445306053e-05,
+      "loss": 0.7796,
+      "step": 3661
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.9303162522153417e-05,
+      "loss": 1.0635,
+      "step": 3662
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.929309087932169e-05,
+      "loss": 0.8811,
+      "step": 3663
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.9283018518495305e-05,
+      "loss": 0.8334,
+      "step": 3664
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.92729454413588e-05,
+      "loss": 0.7729,
+      "step": 3665
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.926287164959684e-05,
+      "loss": 0.5765,
+      "step": 3666
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.9252797144894212e-05,
+      "loss": 0.6004,
+      "step": 3667
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.9242721928935807e-05,
+      "loss": 0.8771,
+      "step": 3668
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.9232646003406666e-05,
+      "loss": 0.8588,
+      "step": 3669
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.9222569369991915e-05,
+      "loss": 0.812,
+      "step": 3670
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.9212492030376814e-05,
+      "loss": 0.8464,
+      "step": 3671
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.920241398624675e-05,
+      "loss": 0.6056,
+      "step": 3672
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.9192335239287205e-05,
+      "loss": 0.9548,
+      "step": 3673
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.9182255791183793e-05,
+      "loss": 0.6484,
+      "step": 3674
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.9172175643622264e-05,
+      "loss": 0.666,
+      "step": 3675
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.9162094798288424e-05,
+      "loss": 0.729,
+      "step": 3676
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.9152013256868278e-05,
+      "loss": 0.7399,
+      "step": 3677
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.9141931021047886e-05,
+      "loss": 0.7973,
+      "step": 3678
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.9131848092513432e-05,
+      "loss": 0.8499,
+      "step": 3679
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.9121764472951252e-05,
+      "loss": 0.7247,
+      "step": 3680
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.911168016404776e-05,
+      "loss": 0.7253,
+      "step": 3681
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.9101595167489492e-05,
+      "loss": 0.6633,
+      "step": 3682
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.9091509484963126e-05,
+      "loss": 0.6266,
+      "step": 3683
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.908142311815542e-05,
+      "loss": 0.6641,
+      "step": 3684
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.9071336068753253e-05,
+      "loss": 0.8661,
+      "step": 3685
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.9061248338443642e-05,
+      "loss": 0.8581,
+      "step": 3686
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.9051159928913696e-05,
+      "loss": 0.6394,
+      "step": 3687
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.9041070841850637e-05,
+      "loss": 0.9711,
+      "step": 3688
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.9030981078941816e-05,
+      "loss": 0.6407,
+      "step": 3689
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.902089064187468e-05,
+      "loss": 0.9687,
+      "step": 3690
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.9010799532336803e-05,
+      "loss": 0.9316,
+      "step": 3691
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.900070775201586e-05,
+      "loss": 0.7947,
+      "step": 3692
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.8990615302599645e-05,
+      "loss": 0.9703,
+      "step": 3693
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.8980522185776065e-05,
+      "loss": 0.5889,
+      "step": 3694
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.8970428403233124e-05,
+      "loss": 0.8777,
+      "step": 3695
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.8960333956658965e-05,
+      "loss": 0.7234,
+      "step": 3696
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.8950238847741812e-05,
+      "loss": 0.7759,
+      "step": 3697
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.8940143078170023e-05,
+      "loss": 0.8831,
+      "step": 3698
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.893004664963205e-05,
+      "loss": 0.7743,
+      "step": 3699
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.8919949563816468e-05,
+      "loss": 0.565,
+      "step": 3700
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.8909851822411954e-05,
+      "loss": 0.7795,
+      "step": 3701
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.8899753427107296e-05,
+      "loss": 0.944,
+      "step": 3702
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.8889654379591384e-05,
+      "loss": 0.7683,
+      "step": 3703
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.8879554681553255e-05,
+      "loss": 0.7611,
+      "step": 3704
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.8869454334681995e-05,
+      "loss": 0.8337,
+      "step": 3705
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.8859353340666827e-05,
+      "loss": 0.788,
+      "step": 3706
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.8849251701197117e-05,
+      "loss": 0.9526,
+      "step": 3707
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.8839149417962276e-05,
+      "loss": 0.7403,
+      "step": 3708
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.8829046492651852e-05,
+      "loss": 0.8682,
+      "step": 3709
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.881894292695552e-05,
+      "loss": 0.7525,
+      "step": 3710
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.8808838722563026e-05,
+      "loss": 0.6809,
+      "step": 3711
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.879873388116425e-05,
+      "loss": 0.9103,
+      "step": 3712
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.8788628404449177e-05,
+      "loss": 0.6729,
+      "step": 3713
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.8778522294107857e-05,
+      "loss": 0.8549,
+      "step": 3714
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.8768415551830515e-05,
+      "loss": 0.7465,
+      "step": 3715
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.8758308179307426e-05,
+      "loss": 1.0045,
+      "step": 3716
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.874820017822899e-05,
+      "loss": 0.6285,
+      "step": 3717
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.873809155028572e-05,
+      "loss": 0.8311,
+      "step": 3718
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.872798229716821e-05,
+      "loss": 0.8892,
+      "step": 3719
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.8717872420567195e-05,
+      "loss": 0.8336,
+      "step": 3720
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.870776192217348e-05,
+      "loss": 0.7981,
+      "step": 3721
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.8697650803677994e-05,
+      "loss": 0.7813,
+      "step": 3722
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.8687539066771762e-05,
+      "loss": 0.8806,
+      "step": 3723
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.867742671314591e-05,
+      "loss": 0.807,
+      "step": 3724
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.866731374449167e-05,
+      "loss": 0.818,
+      "step": 3725
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.865720016250038e-05,
+      "loss": 0.7437,
+      "step": 3726
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.8647085968863486e-05,
+      "loss": 1.0006,
+      "step": 3727
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.8636971165272514e-05,
+      "loss": 0.836,
+      "step": 3728
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.8626855753419112e-05,
+      "loss": 1.0467,
+      "step": 3729
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.861673973499503e-05,
+      "loss": 0.6883,
+      "step": 3730
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.860662311169211e-05,
+      "loss": 0.6314,
+      "step": 3731
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.8596505885202295e-05,
+      "loss": 0.7777,
+      "step": 3732
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.858638805721764e-05,
+      "loss": 0.5525,
+      "step": 3733
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.8576269629430286e-05,
+      "loss": 0.7531,
+      "step": 3734
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.8566150603532477e-05,
+      "loss": 0.7053,
+      "step": 3735
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.8556030981216587e-05,
+      "loss": 0.7516,
+      "step": 3736
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.8545910764175037e-05,
+      "loss": 0.6861,
+      "step": 3737
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.8535789954100377e-05,
+      "loss": 0.8415,
+      "step": 3738
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.852566855268528e-05,
+      "loss": 0.6864,
+      "step": 3739
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.8515546561622462e-05,
+      "loss": 0.822,
+      "step": 3740
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.8505423982604778e-05,
+      "loss": 0.7033,
+      "step": 3741
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.8495300817325182e-05,
+      "loss": 0.8838,
+      "step": 3742
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.8485177067476698e-05,
+      "loss": 0.8662,
+      "step": 3743
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.8475052734752465e-05,
+      "loss": 0.7751,
+      "step": 3744
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.8464927820845744e-05,
+      "loss": 0.719,
+      "step": 3745
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.845480232744983e-05,
+      "loss": 0.7537,
+      "step": 3746
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.8444676256258183e-05,
+      "loss": 0.8767,
+      "step": 3747
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.8434549608964333e-05,
+      "loss": 0.7865,
+      "step": 3748
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.8424422387261873e-05,
+      "loss": 0.6532,
+      "step": 3749
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.8414294592844547e-05,
+      "loss": 0.8165,
+      "step": 3750
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.840416622740617e-05,
+      "loss": 0.5855,
+      "step": 3751
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.8394037292640634e-05,
+      "loss": 1.0582,
+      "step": 3752
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.8383907790241963e-05,
+      "loss": 0.7996,
+      "step": 3753
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.8373777721904256e-05,
+      "loss": 0.923,
+      "step": 3754
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.83636470893217e-05,
+      "loss": 0.9151,
+      "step": 3755
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.8353515894188586e-05,
+      "loss": 0.7866,
+      "step": 3756
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.8343384138199307e-05,
+      "loss": 0.8362,
+      "step": 3757
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.8333251823048333e-05,
+      "loss": 0.7071,
+      "step": 3758
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.8323118950430238e-05,
+      "loss": 0.7438,
+      "step": 3759
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.831298552203968e-05,
+      "loss": 0.784,
+      "step": 3760
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.8302851539571428e-05,
+      "loss": 0.8873,
+      "step": 3761
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.829271700472032e-05,
+      "loss": 0.8064,
+      "step": 3762
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.828258191918131e-05,
+      "loss": 0.8509,
+      "step": 3763
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.8272446284649423e-05,
+      "loss": 0.6949,
+      "step": 3764
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.8262310102819787e-05,
+      "loss": 0.7639,
+      "step": 3765
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.8252173375387624e-05,
+      "loss": 0.886,
+      "step": 3766
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.8242036104048247e-05,
+      "loss": 0.8352,
+      "step": 3767
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.8231898290497045e-05,
+      "loss": 0.7754,
+      "step": 3768
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.8221759936429515e-05,
+      "loss": 0.7336,
+      "step": 3769
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.821162104354123e-05,
+      "loss": 0.7487,
+      "step": 3770
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.820148161352788e-05,
+      "loss": 0.707,
+      "step": 3771
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.819134164808521e-05,
+      "loss": 0.7664,
+      "step": 3772
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.818120114890907e-05,
+      "loss": 0.8143,
+      "step": 3773
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.8171060117695414e-05,
+      "loss": 0.8736,
+      "step": 3774
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.816091855614026e-05,
+      "loss": 0.9005,
+      "step": 3775
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.815077646593972e-05,
+      "loss": 0.5458,
+      "step": 3776
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.8140633848790022e-05,
+      "loss": 0.6168,
+      "step": 3777
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.813049070638743e-05,
+      "loss": 0.7313,
+      "step": 3778
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.8120347040428345e-05,
+      "loss": 0.8248,
+      "step": 3779
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.8110202852609246e-05,
+      "loss": 0.8318,
+      "step": 3780
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.810005814462666e-05,
+      "loss": 0.7949,
+      "step": 3781
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.8089912918177263e-05,
+      "loss": 0.784,
+      "step": 3782
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.807976717495777e-05,
+      "loss": 0.8253,
+      "step": 3783
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.806962091666499e-05,
+      "loss": 0.6337,
+      "step": 3784
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.805947414499584e-05,
+      "loss": 0.742,
+      "step": 3785
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.8049326861647302e-05,
+      "loss": 0.9684,
+      "step": 3786
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.803917906831645e-05,
+      "loss": 0.8163,
+      "step": 3787
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.802903076670045e-05,
+      "loss": 0.8138,
+      "step": 3788
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.8018881958496545e-05,
+      "loss": 0.8533,
+      "step": 3789
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.800873264540206e-05,
+      "loss": 0.8023,
+      "step": 3790
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.799858282911441e-05,
+      "loss": 0.7811,
+      "step": 3791
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.79884325113311e-05,
+      "loss": 0.7803,
+      "step": 3792
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.79782816937497e-05,
+      "loss": 0.7156,
+      "step": 3793
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.7968130378067887e-05,
+      "loss": 0.8145,
+      "step": 3794
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.79579785659834e-05,
+      "loss": 0.746,
+      "step": 3795
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.7947826259194082e-05,
+      "loss": 0.588,
+      "step": 3796
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.793767345939784e-05,
+      "loss": 0.8826,
+      "step": 3797
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.7927520168292675e-05,
+      "loss": 0.7563,
+      "step": 3798
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.7917366387576664e-05,
+      "loss": 0.8604,
+      "step": 3799
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.7907212118947968e-05,
+      "loss": 0.6851,
+      "step": 3800
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.7897057364104834e-05,
+      "loss": 0.8426,
+      "step": 3801
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.788690212474558e-05,
+      "loss": 0.6903,
+      "step": 3802
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.787674640256861e-05,
+      "loss": 0.6461,
+      "step": 3803
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.786659019927242e-05,
+      "loss": 0.7815,
+      "step": 3804
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.785643351655556e-05,
+      "loss": 0.9016,
+      "step": 3805
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.7846276356116696e-05,
+      "loss": 0.6642,
+      "step": 3806
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.7836118719654546e-05,
+      "loss": 0.8258,
+      "step": 3807
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.7825960608867908e-05,
+      "loss": 0.7834,
+      "step": 3808
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.781580202545568e-05,
+      "loss": 0.9356,
+      "step": 3809
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.7805642971116818e-05,
+      "loss": 0.8545,
+      "step": 3810
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.7795483447550365e-05,
+      "loss": 0.6501,
+      "step": 3811
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.778532345645546e-05,
+      "loss": 0.7512,
+      "step": 3812
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.777516299953127e-05,
+      "loss": 0.8703,
+      "step": 3813
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.77650020784771e-05,
+      "loss": 0.8613,
+      "step": 3814
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.7754840694992302e-05,
+      "loss": 0.7684,
+      "step": 3815
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.774467885077629e-05,
+      "loss": 0.8505,
+      "step": 3816
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.77345165475286e-05,
+      "loss": 0.8009,
+      "step": 3817
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.7724353786948804e-05,
+      "loss": 0.6708,
+      "step": 3818
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.771419057073656e-05,
+      "loss": 0.5947,
+      "step": 3819
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 2.770402690059162e-05,
+      "loss": 0.8688,
+      "step": 3820
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 2.7693862778213786e-05,
+      "loss": 0.7908,
+      "step": 3821
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 2.7683698205302967e-05,
+      "loss": 0.6576,
+      "step": 3822
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 2.767353318355911e-05,
+      "loss": 0.7495,
+      "step": 3823
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.7663367714682264e-05,
+      "loss": 0.732,
+      "step": 3824
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.7653201800372546e-05,
+      "loss": 0.6163,
+      "step": 3825
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.764303544233015e-05,
+      "loss": 0.7661,
+      "step": 3826
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.7632868642255333e-05,
+      "loss": 0.891,
+      "step": 3827
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.7622701401848437e-05,
+      "loss": 0.7562,
+      "step": 3828
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.761253372280988e-05,
+      "loss": 0.7961,
+      "step": 3829
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.7602365606840136e-05,
+      "loss": 0.9082,
+      "step": 3830
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.7592197055639766e-05,
+      "loss": 0.6838,
+      "step": 3831
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.7582028070909415e-05,
+      "loss": 0.8275,
+      "step": 3832
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.7571858654349768e-05,
+      "loss": 0.655,
+      "step": 3833
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.7561688807661617e-05,
+      "loss": 0.7473,
+      "step": 3834
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.75515185325458e-05,
+      "loss": 0.8144,
+      "step": 3835
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.754134783070324e-05,
+      "loss": 0.6772,
+      "step": 3836
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.7531176703834925e-05,
+      "loss": 0.7258,
+      "step": 3837
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.752100515364193e-05,
+      "loss": 0.6141,
+      "step": 3838
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.7510833181825374e-05,
+      "loss": 0.7477,
+      "step": 3839
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.7500660790086458e-05,
+      "loss": 0.6827,
+      "step": 3840
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.7490487980126477e-05,
+      "loss": 0.7049,
+      "step": 3841
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.7480314753646752e-05,
+      "loss": 0.7723,
+      "step": 3842
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.74701411123487e-05,
+      "loss": 0.8909,
+      "step": 3843
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.7459967057933822e-05,
+      "loss": 0.764,
+      "step": 3844
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.7449792592103644e-05,
+      "loss": 0.6639,
+      "step": 3845
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.7439617716559808e-05,
+      "loss": 0.9217,
+      "step": 3846
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.7429442433003995e-05,
+      "loss": 0.7572,
+      "step": 3847
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.741926674313795e-05,
+      "loss": 0.7995,
+      "step": 3848
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.740909064866352e-05,
+      "loss": 0.834,
+      "step": 3849
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.7398914151282596e-05,
+      "loss": 0.7443,
+      "step": 3850
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.7388737252697117e-05,
+      "loss": 0.757,
+      "step": 3851
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.737855995460913e-05,
+      "loss": 0.8761,
+      "step": 3852
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.7368382258720726e-05,
+      "loss": 0.801,
+      "step": 3853
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.7358204166734058e-05,
+      "loss": 0.7399,
+      "step": 3854
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.7348025680351363e-05,
+      "loss": 0.6578,
+      "step": 3855
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.733784680127493e-05,
+      "loss": 0.8198,
+      "step": 3856
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.7327667531207113e-05,
+      "loss": 0.7779,
+      "step": 3857
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.7317487871850343e-05,
+      "loss": 0.7642,
+      "step": 3858
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.7307307824907113e-05,
+      "loss": 0.6429,
+      "step": 3859
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.7297127392079964e-05,
+      "loss": 0.5574,
+      "step": 3860
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.7286946575071526e-05,
+      "loss": 0.9101,
+      "step": 3861
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.7276765375584472e-05,
+      "loss": 0.6973,
+      "step": 3862
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.726658379532156e-05,
+      "loss": 0.7654,
+      "step": 3863
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.725640183598559e-05,
+      "loss": 0.8112,
+      "step": 3864
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.7246219499279446e-05,
+      "loss": 0.7122,
+      "step": 3865
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.7236036786906056e-05,
+      "loss": 0.7953,
+      "step": 3866
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.7225853700568422e-05,
+      "loss": 0.676,
+      "step": 3867
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.7215670241969614e-05,
+      "loss": 0.4741,
+      "step": 3868
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.7205486412812746e-05,
+      "loss": 0.9691,
+      "step": 3869
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.7195302214801012e-05,
+      "loss": 0.8613,
+      "step": 3870
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.718511764963766e-05,
+      "loss": 0.8007,
+      "step": 3871
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.7174932719025985e-05,
+      "loss": 0.9833,
+      "step": 3872
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.716474742466938e-05,
+      "loss": 0.908,
+      "step": 3873
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.715456176827127e-05,
+      "loss": 0.6841,
+      "step": 3874
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.7144375751535135e-05,
+      "loss": 0.7736,
+      "step": 3875
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.7134189376164542e-05,
+      "loss": 0.8108,
+      "step": 3876
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.7124002643863105e-05,
+      "loss": 0.8319,
+      "step": 3877
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.7113815556334478e-05,
+      "loss": 0.9638,
+      "step": 3878
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.7103628115282403e-05,
+      "loss": 0.6774,
+      "step": 3879
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.709344032241068e-05,
+      "loss": 0.7392,
+      "step": 3880
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.7083252179423148e-05,
+      "loss": 0.7201,
+      "step": 3881
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.7073063688023713e-05,
+      "loss": 0.7908,
+      "step": 3882
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.7062874849916346e-05,
+      "loss": 0.7391,
+      "step": 3883
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.7052685666805073e-05,
+      "loss": 0.8475,
+      "step": 3884
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.7042496140393976e-05,
+      "loss": 0.7804,
+      "step": 3885
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.7032306272387187e-05,
+      "loss": 0.8153,
+      "step": 3886
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.702211606448891e-05,
+      "loss": 0.6791,
+      "step": 3887
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.7011925518403398e-05,
+      "loss": 0.6599,
+      "step": 3888
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.7001734635834957e-05,
+      "loss": 0.5191,
+      "step": 3889
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.699154341848796e-05,
+      "loss": 0.8444,
+      "step": 3890
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.6981351868066823e-05,
+      "loss": 0.814,
+      "step": 3891
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.6971159986276024e-05,
+      "loss": 0.95,
+      "step": 3892
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.69609677748201e-05,
+      "loss": 0.9507,
+      "step": 3893
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.6950775235403637e-05,
+      "loss": 0.9039,
+      "step": 3894
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.6940582369731276e-05,
+      "loss": 0.8736,
+      "step": 3895
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.6930389179507714e-05,
+      "loss": 0.9532,
+      "step": 3896
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.692019566643772e-05,
+      "loss": 0.6925,
+      "step": 3897
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.6910001832226084e-05,
+      "loss": 0.8265,
+      "step": 3898
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.689980767857766e-05,
+      "loss": 0.4946,
+      "step": 3899
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.6889613207197384e-05,
+      "loss": 0.6397,
+      "step": 3900
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.6879418419790204e-05,
+      "loss": 0.8625,
+      "step": 3901
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.686922331806114e-05,
+      "loss": 0.7974,
+      "step": 3902
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.6859027903715273e-05,
+      "loss": 0.7917,
+      "step": 3903
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.6848832178457716e-05,
+      "loss": 0.7084,
+      "step": 3904
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.6838636143993662e-05,
+      "loss": 0.9137,
+      "step": 3905
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.6828439802028327e-05,
+      "loss": 0.7477,
+      "step": 3906
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.681824315426698e-05,
+      "loss": 0.7459,
+      "step": 3907
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.6808046202414976e-05,
+      "loss": 0.6971,
+      "step": 3908
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.6797848948177684e-05,
+      "loss": 0.9005,
+      "step": 3909
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.6787651393260528e-05,
+      "loss": 0.8192,
+      "step": 3910
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.6777453539369002e-05,
+      "loss": 0.7908,
+      "step": 3911
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.6767255388208634e-05,
+      "loss": 0.7521,
+      "step": 3912
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.6757056941485003e-05,
+      "loss": 0.7673,
+      "step": 3913
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.674685820090375e-05,
+      "loss": 0.8906,
+      "step": 3914
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.6736659168170547e-05,
+      "loss": 0.6928,
+      "step": 3915
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.672645984499112e-05,
+      "loss": 0.805,
+      "step": 3916
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.6716260233071254e-05,
+      "loss": 0.8045,
+      "step": 3917
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.6706060334116777e-05,
+      "loss": 0.7506,
+      "step": 3918
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.6695860149833556e-05,
+      "loss": 0.7257,
+      "step": 3919
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.6685659681927515e-05,
+      "loss": 0.7703,
+      "step": 3920
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.667545893210463e-05,
+      "loss": 0.7737,
+      "step": 3921
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.6665257902070905e-05,
+      "loss": 0.7919,
+      "step": 3922
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.6655056593532416e-05,
+      "loss": 1.0049,
+      "step": 3923
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.6644855008195267e-05,
+      "loss": 0.6867,
+      "step": 3924
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.6634653147765615e-05,
+      "loss": 0.7131,
+      "step": 3925
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.6624451013949665e-05,
+      "loss": 0.7549,
+      "step": 3926
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.6614248608453658e-05,
+      "loss": 0.7771,
+      "step": 3927
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.660404593298389e-05,
+      "loss": 0.8485,
+      "step": 3928
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.6593842989246704e-05,
+      "loss": 0.8379,
+      "step": 3929
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.6583639778948477e-05,
+      "loss": 0.6851,
+      "step": 3930
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.657343630379564e-05,
+      "loss": 0.7695,
+      "step": 3931
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.6563232565494674e-05,
+      "loss": 0.6751,
+      "step": 3932
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.655302856575208e-05,
+      "loss": 0.8187,
+      "step": 3933
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.6542824306274424e-05,
+      "loss": 0.7163,
+      "step": 3934
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.6532619788768315e-05,
+      "loss": 0.6921,
+      "step": 3935
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.6522415014940394e-05,
+      "loss": 0.6777,
+      "step": 3936
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.6512209986497343e-05,
+      "loss": 0.7409,
+      "step": 3937
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.6502004705145922e-05,
+      "loss": 0.5506,
+      "step": 3938
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.649179917259287e-05,
+      "loss": 0.7351,
+      "step": 3939
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.6481593390545023e-05,
+      "loss": 0.8139,
+      "step": 3940
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.6471387360709245e-05,
+      "loss": 0.6793,
+      "step": 3941
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.6461181084792414e-05,
+      "loss": 0.8055,
+      "step": 3942
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.6450974564501486e-05,
+      "loss": 0.7897,
+      "step": 3943
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.644076780154345e-05,
+      "loss": 0.7246,
+      "step": 3944
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.6430560797625303e-05,
+      "loss": 0.6452,
+      "step": 3945
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.6420353554454123e-05,
+      "loss": 0.7963,
+      "step": 3946
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.641014607373702e-05,
+      "loss": 0.6733,
+      "step": 3947
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.6399938357181124e-05,
+      "loss": 0.8912,
+      "step": 3948
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.6389730406493623e-05,
+      "loss": 0.5859,
+      "step": 3949
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.6379522223381735e-05,
+      "loss": 0.8676,
+      "step": 3950
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.636931380955272e-05,
+      "loss": 0.6434,
+      "step": 3951
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.6359105166713875e-05,
+      "loss": 0.6088,
+      "step": 3952
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.634889629657254e-05,
+      "loss": 0.712,
+      "step": 3953
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.6338687200836086e-05,
+      "loss": 0.8437,
+      "step": 3954
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.6328477881211927e-05,
+      "loss": 0.762,
+      "step": 3955
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.6318268339407516e-05,
+      "loss": 0.9381,
+      "step": 3956
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.6308058577130335e-05,
+      "loss": 0.6068,
+      "step": 3957
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.6297848596087906e-05,
+      "loss": 0.7902,
+      "step": 3958
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.6287638397987795e-05,
+      "loss": 0.6962,
+      "step": 3959
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.6277427984537593e-05,
+      "loss": 0.6462,
+      "step": 3960
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.626721735744494e-05,
+      "loss": 0.5469,
+      "step": 3961
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.6257006518417506e-05,
+      "loss": 0.6249,
+      "step": 3962
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.624679546916297e-05,
+      "loss": 0.8365,
+      "step": 3963
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.6236584211389115e-05,
+      "loss": 0.6308,
+      "step": 3964
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.6226372746803674e-05,
+      "loss": 0.7371,
+      "step": 3965
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.621616107711447e-05,
+      "loss": 0.8379,
+      "step": 3966
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.6205949204029363e-05,
+      "loss": 0.7508,
+      "step": 3967
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.6195737129256203e-05,
+      "loss": 0.7445,
+      "step": 3968
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.618552485450291e-05,
+      "loss": 0.7251,
+      "step": 3969
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.6175312381477442e-05,
+      "loss": 0.8073,
+      "step": 3970
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.6165099711887758e-05,
+      "loss": 0.7884,
+      "step": 3971
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.615488684744187e-05,
+      "loss": 0.6973,
+      "step": 3972
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.614467378984783e-05,
+      "loss": 0.7475,
+      "step": 3973
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.61344605408137e-05,
+      "loss": 0.6944,
+      "step": 3974
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.61242471020476e-05,
+      "loss": 0.721,
+      "step": 3975
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.6114033475257672e-05,
+      "loss": 0.7481,
+      "step": 3976
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.6103819662152063e-05,
+      "loss": 0.7958,
+      "step": 3977
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.609360566443899e-05,
+      "loss": 0.8958,
+      "step": 3978
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.6083391483826697e-05,
+      "loss": 0.7362,
+      "step": 3979
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.6073177122023417e-05,
+      "loss": 0.7227,
+      "step": 3980
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.6062962580737465e-05,
+      "loss": 0.7844,
+      "step": 3981
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 2.605274786167716e-05,
+      "loss": 0.7352,
+      "step": 3982
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 2.604253296655086e-05,
+      "loss": 0.7176,
+      "step": 3983
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 2.6032317897066934e-05,
+      "loss": 0.7125,
+      "step": 3984
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 2.60221026549338e-05,
+      "loss": 0.6724,
+      "step": 3985
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.60118872418599e-05,
+      "loss": 0.7762,
+      "step": 3986
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.60016716595537e-05,
+      "loss": 0.6492,
+      "step": 3987
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.59914559097237e-05,
+      "loss": 0.6903,
+      "step": 3988
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.598123999407842e-05,
+      "loss": 0.9344,
+      "step": 3989
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.597102391432642e-05,
+      "loss": 0.9524,
+      "step": 3990
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.5960807672176273e-05,
+      "loss": 0.9369,
+      "step": 3991
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.5950591269336593e-05,
+      "loss": 0.7023,
+      "step": 3992
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.5940374707516015e-05,
+      "loss": 0.7935,
+      "step": 3993
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.5930157988423194e-05,
+      "loss": 0.683,
+      "step": 3994
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.5919941113766816e-05,
+      "loss": 0.9588,
+      "step": 3995
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.59097240852556e-05,
+      "loss": 0.7216,
+      "step": 3996
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.5899506904598287e-05,
+      "loss": 0.6883,
+      "step": 3997
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.588928957350363e-05,
+      "loss": 0.8226,
+      "step": 3998
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.5879072093680436e-05,
+      "loss": 0.69,
+      "step": 3999
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.5868854466837506e-05,
+      "loss": 0.8399,
+      "step": 4000
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.5858636694683675e-05,
+      "loss": 0.8281,
+      "step": 4001
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.584841877892783e-05,
+      "loss": 0.6943,
+      "step": 4002
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.5838200721278834e-05,
+      "loss": 0.6316,
+      "step": 4003
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.5827982523445597e-05,
+      "loss": 0.8204,
+      "step": 4004
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.5817764187137083e-05,
+      "loss": 0.7156,
+      "step": 4005
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.5807545714062215e-05,
+      "loss": 0.6673,
+      "step": 4006
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.5797327105929993e-05,
+      "loss": 0.8136,
+      "step": 4007
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.5787108364449425e-05,
+      "loss": 0.5903,
+      "step": 4008
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.5776889491329515e-05,
+      "loss": 0.6744,
+      "step": 4009
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.5766670488279325e-05,
+      "loss": 0.5706,
+      "step": 4010
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.5756451357007934e-05,
+      "loss": 0.5631,
+      "step": 4011
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.574623209922441e-05,
+      "loss": 0.9825,
+      "step": 4012
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.5736012716637875e-05,
+      "loss": 0.7718,
+      "step": 4013
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.5725793210957467e-05,
+      "loss": 0.6236,
+      "step": 4014
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.5715573583892338e-05,
+      "loss": 0.5563,
+      "step": 4015
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.570535383715165e-05,
+      "loss": 0.6917,
+      "step": 4016
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.5695133972444613e-05,
+      "loss": 0.8337,
+      "step": 4017
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.568491399148043e-05,
+      "loss": 0.8337,
+      "step": 4018
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.567469389596834e-05,
+      "loss": 0.6557,
+      "step": 4019
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.5664473687617587e-05,
+      "loss": 0.7297,
+      "step": 4020
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.5654253368137443e-05,
+      "loss": 0.8049,
+      "step": 4021
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.5644032939237205e-05,
+      "loss": 0.652,
+      "step": 4022
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.563381240262618e-05,
+      "loss": 0.6484,
+      "step": 4023
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.5623591760013687e-05,
+      "loss": 0.5943,
+      "step": 4024
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.5613371013109073e-05,
+      "loss": 0.9256,
+      "step": 4025
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.5603150163621692e-05,
+      "loss": 0.7993,
+      "step": 4026
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.5592929213260937e-05,
+      "loss": 0.7575,
+      "step": 4027
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.5582708163736192e-05,
+      "loss": 0.7333,
+      "step": 4028
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.5572487016756867e-05,
+      "loss": 0.9372,
+      "step": 4029
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.55622657740324e-05,
+      "loss": 0.7559,
+      "step": 4030
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.5552044437272227e-05,
+      "loss": 0.7542,
+      "step": 4031
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.5541823008185807e-05,
+      "loss": 0.8749,
+      "step": 4032
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.5531601488482614e-05,
+      "loss": 1.0312,
+      "step": 4033
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.552137987987215e-05,
+      "loss": 0.482,
+      "step": 4034
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.551115818406391e-05,
+      "loss": 0.7948,
+      "step": 4035
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.5500936402767406e-05,
+      "loss": 0.7566,
+      "step": 4036
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.549071453769219e-05,
+      "loss": 0.5697,
+      "step": 4037
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.54804925905478e-05,
+      "loss": 0.5987,
+      "step": 4038
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.547027056304379e-05,
+      "loss": 0.8836,
+      "step": 4039
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.5460048456889756e-05,
+      "loss": 0.7873,
+      "step": 4040
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.5449826273795265e-05,
+      "loss": 0.845,
+      "step": 4041
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.5439604015469926e-05,
+      "loss": 0.8294,
+      "step": 4042
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.5429381683623367e-05,
+      "loss": 0.7645,
+      "step": 4043
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.5419159279965182e-05,
+      "loss": 0.8041,
+      "step": 4044
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.5408936806205036e-05,
+      "loss": 0.8151,
+      "step": 4045
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.5398714264052577e-05,
+      "loss": 0.7676,
+      "step": 4046
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.5388491655217445e-05,
+      "loss": 0.8707,
+      "step": 4047
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.5378268981409337e-05,
+      "loss": 0.7203,
+      "step": 4048
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.5368046244337923e-05,
+      "loss": 0.5779,
+      "step": 4049
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.5357823445712897e-05,
+      "loss": 0.9175,
+      "step": 4050
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.534760058724397e-05,
+      "loss": 0.9179,
+      "step": 4051
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.533737767064085e-05,
+      "loss": 0.6207,
+      "step": 4052
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.5327154697613264e-05,
+      "loss": 0.7435,
+      "step": 4053
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.531693166987094e-05,
+      "loss": 0.7723,
+      "step": 4054
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.530670858912363e-05,
+      "loss": 0.8498,
+      "step": 4055
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.5296485457081076e-05,
+      "loss": 0.6457,
+      "step": 4056
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.5286262275453042e-05,
+      "loss": 0.7276,
+      "step": 4057
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.5276039045949294e-05,
+      "loss": 1.0237,
+      "step": 4058
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.526581577027961e-05,
+      "loss": 0.8346,
+      "step": 4059
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.5255592450153777e-05,
+      "loss": 0.794,
+      "step": 4060
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.5245369087281574e-05,
+      "loss": 0.7306,
+      "step": 4061
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.5235145683372814e-05,
+      "loss": 0.5338,
+      "step": 4062
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.5224922240137293e-05,
+      "loss": 0.5414,
+      "step": 4063
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.521469875928483e-05,
+      "loss": 0.8355,
+      "step": 4064
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.520447524252523e-05,
+      "loss": 0.8229,
+      "step": 4065
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.5194251691568337e-05,
+      "loss": 0.777,
+      "step": 4066
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.5184028108123964e-05,
+      "loss": 0.8061,
+      "step": 4067
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.5173804493901954e-05,
+      "loss": 0.5623,
+      "step": 4068
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.5163580850612152e-05,
+      "loss": 0.9117,
+      "step": 4069
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.515335717996439e-05,
+      "loss": 0.597,
+      "step": 4070
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.514313348366852e-05,
+      "loss": 0.6179,
+      "step": 4071
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.5132909763434413e-05,
+      "loss": 0.6993,
+      "step": 4072
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.5122686020971903e-05,
+      "loss": 0.6942,
+      "step": 4073
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.511246225799087e-05,
+      "loss": 0.7596,
+      "step": 4074
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.5102238476201183e-05,
+      "loss": 0.8086,
+      "step": 4075
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.5092014677312686e-05,
+      "loss": 0.6901,
+      "step": 4076
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.5081790863035277e-05,
+      "loss": 0.6715,
+      "step": 4077
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.5071567035078818e-05,
+      "loss": 0.6127,
+      "step": 4078
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.5061343195153177e-05,
+      "loss": 0.5873,
+      "step": 4079
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.5051119344968242e-05,
+      "loss": 0.6139,
+      "step": 4080
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.50408954862339e-05,
+      "loss": 0.8265,
+      "step": 4081
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.503067162066002e-05,
+      "loss": 0.825,
+      "step": 4082
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.5020447749956487e-05,
+      "loss": 0.5962,
+      "step": 4083
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.501022387583319e-05,
+      "loss": 0.929,
+      "step": 4084
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6002,
+      "step": 4085
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.498977612416682e-05,
+      "loss": 0.9227,
+      "step": 4086
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.4979552250043526e-05,
+      "loss": 0.9044,
+      "step": 4087
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.4969328379339983e-05,
+      "loss": 0.7411,
+      "step": 4088
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.4959104513766103e-05,
+      "loss": 0.933,
+      "step": 4089
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.494888065503176e-05,
+      "loss": 0.5606,
+      "step": 4090
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.4938656804846826e-05,
+      "loss": 0.8379,
+      "step": 4091
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.492843296492119e-05,
+      "loss": 0.6853,
+      "step": 4092
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4918209136964736e-05,
+      "loss": 0.7338,
+      "step": 4093
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.490798532268731e-05,
+      "loss": 0.8487,
+      "step": 4094
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4897761523798823e-05,
+      "loss": 0.7334,
+      "step": 4095
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.4887537742009132e-05,
+      "loss": 0.5075,
+      "step": 4096
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.48773139790281e-05,
+      "loss": 0.7466,
+      "step": 4097
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.48670902365656e-05,
+      "loss": 0.9114,
+      "step": 4098
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.485686651633148e-05,
+      "loss": 0.7253,
+      "step": 4099
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 2.4846642820035617e-05,
+      "loss": 0.7111,
+      "step": 4100
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.4836419149387857e-05,
+      "loss": 0.7911,
+      "step": 4101
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.4826195506098055e-05,
+      "loss": 0.7495,
+      "step": 4102
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.4815971891876045e-05,
+      "loss": 0.9207,
+      "step": 4103
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.4805748308431666e-05,
+      "loss": 0.6953,
+      "step": 4104
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.4795524757474774e-05,
+      "loss": 0.8273,
+      "step": 4105
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.4785301240715177e-05,
+      "loss": 0.7071,
+      "step": 4106
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.477507775986271e-05,
+      "loss": 0.6332,
+      "step": 4107
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 2.47648543166272e-05,
+      "loss": 0.8775,
+      "step": 4108
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.4754630912718435e-05,
+      "loss": 0.6398,
+      "step": 4109
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.4744407549846232e-05,
+      "loss": 0.8097,
+      "step": 4110
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.4734184229720394e-05,
+      "loss": 0.7025,
+      "step": 4111
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.472396095405071e-05,
+      "loss": 0.9709,
+      "step": 4112
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.4713737724546963e-05,
+      "loss": 0.5883,
+      "step": 4113
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.4703514542918934e-05,
+      "loss": 0.7788,
+      "step": 4114
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.4693291410876372e-05,
+      "loss": 0.8452,
+      "step": 4115
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 2.4683068330129062e-05,
+      "loss": 0.7945,
+      "step": 4116
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.4672845302386742e-05,
+      "loss": 0.7516,
+      "step": 4117
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.4662622329359153e-05,
+      "loss": 0.7472,
+      "step": 4118
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.4652399412756033e-05,
+      "loss": 0.8482,
+      "step": 4119
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.46421765542871e-05,
+      "loss": 0.7705,
+      "step": 4120
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.4631953755662076e-05,
+      "loss": 0.784,
+      "step": 4121
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.4621731018590672e-05,
+      "loss": 0.7062,
+      "step": 4122
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.461150834478256e-05,
+      "loss": 0.959,
+      "step": 4123
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 2.4601285735947436e-05,
+      "loss": 0.7975,
+      "step": 4124
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.459106319379497e-05,
+      "loss": 1.0081,
+      "step": 4125
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.458084072003482e-05,
+      "loss": 0.6473,
+      "step": 4126
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.4570618316376642e-05,
+      "loss": 0.5971,
+      "step": 4127
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.456039598453008e-05,
+      "loss": 0.7441,
+      "step": 4128
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.4550173726204745e-05,
+      "loss": 0.5153,
+      "step": 4129
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.4539951543110256e-05,
+      "loss": 0.6991,
+      "step": 4130
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.452972943695621e-05,
+      "loss": 0.65,
+      "step": 4131
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 2.4519507409452207e-05,
+      "loss": 0.7089,
+      "step": 4132
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.4509285462307816e-05,
+      "loss": 0.6484,
+      "step": 4133
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.4499063597232603e-05,
+      "loss": 0.7995,
+      "step": 4134
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.44888418159361e-05,
+      "loss": 0.6416,
+      "step": 4135
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.4478620120127854e-05,
+      "loss": 0.783,
+      "step": 4136
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.446839851151739e-05,
+      "loss": 0.674,
+      "step": 4137
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.4458176991814195e-05,
+      "loss": 0.8482,
+      "step": 4138
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.444795556272778e-05,
+      "loss": 0.825,
+      "step": 4139
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.443773422596761e-05,
+      "loss": 0.7337,
+      "step": 4140
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.4427512983243142e-05,
+      "loss": 0.6832,
+      "step": 4141
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.4417291836263814e-05,
+      "loss": 0.7223,
+      "step": 4142
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.440707078673907e-05,
+      "loss": 0.8341,
+      "step": 4143
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 2.439684983637831e-05,
+      "loss": 0.7518,
+      "step": 4144
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.4386628986890936e-05,
+      "loss": 0.6185,
+      "step": 4145
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.4376408239986326e-05,
+      "loss": 0.7764,
+      "step": 4146
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.436618759737382e-05,
+      "loss": 0.5388,
+      "step": 4147
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.4355967060762798e-05,
+      "loss": 1.0196,
+      "step": 4148
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.434574663186256e-05,
+      "loss": 0.7593,
+      "step": 4149
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.433552631238242e-05,
+      "loss": 0.8834,
+      "step": 4150
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.4325306104031667e-05,
+      "loss": 0.8701,
+      "step": 4151
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 2.431508600851958e-05,
+      "loss": 0.7456,
+      "step": 4152
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.4304866027555386e-05,
+      "loss": 0.7993,
+      "step": 4153
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.4294646162848354e-05,
+      "loss": 0.6602,
+      "step": 4154
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.4284426416107668e-05,
+      "loss": 0.7164,
+      "step": 4155
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.4274206789042536e-05,
+      "loss": 0.742,
+      "step": 4156
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.426398728336213e-05,
+      "loss": 0.8487,
+      "step": 4157
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.4253767900775593e-05,
+      "loss": 0.7671,
+      "step": 4158
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 2.4243548642992072e-05,
+      "loss": 0.8131,
+      "step": 4159
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.4233329511720677e-05,
+      "loss": 0.6497,
+      "step": 4160
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.422311050867049e-05,
+      "loss": 0.7394,
+      "step": 4161
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.4212891635550588e-05,
+      "loss": 0.8431,
+      "step": 4162
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.4202672894070006e-05,
+      "loss": 0.7905,
+      "step": 4163
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.4192454285937784e-05,
+      "loss": 0.7366,
+      "step": 4164
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.4182235812862926e-05,
+      "loss": 0.6892,
+      "step": 4165
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.4172017476554405e-05,
+      "loss": 0.7085,
+      "step": 4166
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 2.4161799278721182e-05,
+      "loss": 0.6584,
+      "step": 4167
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.4151581221072183e-05,
+      "loss": 0.7341,
+      "step": 4168
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.4141363305316324e-05,
+      "loss": 0.7752,
+      "step": 4169
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.4131145533162503e-05,
+      "loss": 0.8393,
+      "step": 4170
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.412092790631957e-05,
+      "loss": 0.858,
+      "step": 4171
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.4110710426496376e-05,
+      "loss": 0.5035,
+      "step": 4172
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.4100493095401726e-05,
+      "loss": 0.5735,
+      "step": 4173
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.4090275914744402e-05,
+      "loss": 0.7001,
+      "step": 4174
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.408005888623319e-05,
+      "loss": 0.7897,
+      "step": 4175
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.4069842011576815e-05,
+      "loss": 0.7739,
+      "step": 4176
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.405962529248399e-05,
+      "loss": 0.7528,
+      "step": 4177
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.4049408730663416e-05,
+      "loss": 0.7442,
+      "step": 4178
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 2.4039192327823726e-05,
+      "loss": 0.7845,
+      "step": 4179
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.4028976085673585e-05,
+      "loss": 0.5922,
+      "step": 4180
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.4018760005921585e-05,
+      "loss": 0.7076,
+      "step": 4181
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.4008544090276307e-05,
+      "loss": 0.9308,
+      "step": 4182
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.3998328340446306e-05,
+      "loss": 0.763,
+      "step": 4183
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.398811275814011e-05,
+      "loss": 0.7654,
+      "step": 4184
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.39778973450662e-05,
+      "loss": 0.8023,
+      "step": 4185
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.3967682102933072e-05,
+      "loss": 0.7619,
+      "step": 4186
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 2.3957467033449147e-05,
+      "loss": 0.734,
+      "step": 4187
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.3947252138322844e-05,
+      "loss": 0.7417,
+      "step": 4188
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.3937037419262538e-05,
+      "loss": 0.688,
+      "step": 4189
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.392682287797658e-05,
+      "loss": 0.7727,
+      "step": 4190
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.3916608516173312e-05,
+      "loss": 0.6928,
+      "step": 4191
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.3906394335561015e-05,
+      "loss": 0.5381,
+      "step": 4192
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.3896180337847946e-05,
+      "loss": 0.8297,
+      "step": 4193
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.3885966524742344e-05,
+      "loss": 0.7172,
+      "step": 4194
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 2.38757528979524e-05,
+      "loss": 0.8222,
+      "step": 4195
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.3865539459186302e-05,
+      "loss": 0.6541,
+      "step": 4196
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.3855326210152177e-05,
+      "loss": 0.8035,
+      "step": 4197
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.3845113152558137e-05,
+      "loss": 0.6539,
+      "step": 4198
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.3834900288112254e-05,
+      "loss": 0.6146,
+      "step": 4199
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.3824687618522567e-05,
+      "loss": 0.736,
+      "step": 4200
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.381447514549709e-05,
+      "loss": 0.8648,
+      "step": 4201
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.38042628707438e-05,
+      "loss": 0.6216,
+      "step": 4202
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 2.3794050795970643e-05,
+      "loss": 0.7899,
+      "step": 4203
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.3783838922885532e-05,
+      "loss": 0.739,
+      "step": 4204
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.3773627253196335e-05,
+      "loss": 0.9011,
+      "step": 4205
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.3763415788610894e-05,
+      "loss": 0.8193,
+      "step": 4206
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.3753204530837028e-05,
+      "loss": 0.6148,
+      "step": 4207
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.3742993481582504e-05,
+      "loss": 0.7136,
+      "step": 4208
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.3732782642555062e-05,
+      "loss": 0.8377,
+      "step": 4209
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.3722572015462416e-05,
+      "loss": 0.8325,
+      "step": 4210
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.3712361602012218e-05,
+      "loss": 0.7235,
+      "step": 4211
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.37021514039121e-05,
+      "loss": 0.814,
+      "step": 4212
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.3691941422869674e-05,
+      "loss": 0.7602,
+      "step": 4213
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.368173166059249e-05,
+      "loss": 0.649,
+      "step": 4214
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 2.3671522118788075e-05,
+      "loss": 0.5569,
+      "step": 4215
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.3661312799163923e-05,
+      "loss": 0.826,
+      "step": 4216
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.365110370342746e-05,
+      "loss": 0.742,
+      "step": 4217
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.3640894833286127e-05,
+      "loss": 0.6263,
+      "step": 4218
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.3630686190447283e-05,
+      "loss": 0.7094,
+      "step": 4219
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.3620477776618267e-05,
+      "loss": 0.6948,
+      "step": 4220
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.3610269593506383e-05,
+      "loss": 0.5723,
+      "step": 4221
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.3600061642818875e-05,
+      "loss": 0.7207,
+      "step": 4222
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 2.3589853926262977e-05,
+      "loss": 0.8579,
+      "step": 4223
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.357964644554588e-05,
+      "loss": 0.713,
+      "step": 4224
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.3569439202374706e-05,
+      "loss": 0.7672,
+      "step": 4225
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.3559232198456564e-05,
+      "loss": 0.8692,
+      "step": 4226
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.354902543549852e-05,
+      "loss": 0.6436,
+      "step": 4227
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.3538818915207588e-05,
+      "loss": 0.7899,
+      "step": 4228
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.352861263929076e-05,
+      "loss": 0.6118,
+      "step": 4229
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.3518406609454983e-05,
+      "loss": 0.7055,
+      "step": 4230
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 2.3508200827407136e-05,
+      "loss": 0.7739,
+      "step": 4231
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.349799529485409e-05,
+      "loss": 0.6414,
+      "step": 4232
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.3487790013502652e-05,
+      "loss": 0.6899,
+      "step": 4233
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.3477584985059612e-05,
+      "loss": 0.5672,
+      "step": 4234
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.346738021123169e-05,
+      "loss": 0.7133,
+      "step": 4235
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.3457175693725586e-05,
+      "loss": 0.6421,
+      "step": 4236
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.3446971434247932e-05,
+      "loss": 0.6689,
+      "step": 4237
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.3436767434505332e-05,
+      "loss": 0.7386,
+      "step": 4238
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.342656369620436e-05,
+      "loss": 0.852,
+      "step": 4239
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.341636022105153e-05,
+      "loss": 0.7327,
+      "step": 4240
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.3406157010753306e-05,
+      "loss": 0.6215,
+      "step": 4241
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.339595406701612e-05,
+      "loss": 0.8951,
+      "step": 4242
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.3385751391546355e-05,
+      "loss": 0.7263,
+      "step": 4243
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.3375548986050344e-05,
+      "loss": 0.7592,
+      "step": 4244
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.336534685223439e-05,
+      "loss": 0.8019,
+      "step": 4245
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.3355144991804735e-05,
+      "loss": 0.6969,
+      "step": 4246
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.3344943406467586e-05,
+      "loss": 0.7205,
+      "step": 4247
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.33347420979291e-05,
+      "loss": 0.8402,
+      "step": 4248
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.3324541067895368e-05,
+      "loss": 0.763,
+      "step": 4249
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.331434031807249e-05,
+      "loss": 0.7046,
+      "step": 4250
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.3304139850166447e-05,
+      "loss": 0.6212,
+      "step": 4251
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.329393966588323e-05,
+      "loss": 0.794,
+      "step": 4252
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.328373976692875e-05,
+      "loss": 0.7447,
+      "step": 4253
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.327354015500888e-05,
+      "loss": 0.7326,
+      "step": 4254
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.3263340831829455e-05,
+      "loss": 0.6173,
+      "step": 4255
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.3253141799096258e-05,
+      "loss": 0.5115,
+      "step": 4256
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.3242943058515e-05,
+      "loss": 0.8692,
+      "step": 4257
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.323274461179137e-05,
+      "loss": 0.653,
+      "step": 4258
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.3222546460631004e-05,
+      "loss": 0.7366,
+      "step": 4259
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.321234860673947e-05,
+      "loss": 0.7822,
+      "step": 4260
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.3202151051822318e-05,
+      "loss": 0.6718,
+      "step": 4261
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.319195379758503e-05,
+      "loss": 0.7537,
+      "step": 4262
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.318175684573302e-05,
+      "loss": 0.6416,
+      "step": 4263
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.3171560197971685e-05,
+      "loss": 0.4329,
+      "step": 4264
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.316136385600634e-05,
+      "loss": 0.9355,
+      "step": 4265
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.3151167821542283e-05,
+      "loss": 0.8245,
+      "step": 4266
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.314097209628473e-05,
+      "loss": 0.7619,
+      "step": 4267
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.3130776681938867e-05,
+      "loss": 0.9413,
+      "step": 4268
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.3120581580209808e-05,
+      "loss": 0.879,
+      "step": 4269
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.3110386792802628e-05,
+      "loss": 0.6463,
+      "step": 4270
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.310019232142234e-05,
+      "loss": 0.7355,
+      "step": 4271
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.3089998167773922e-05,
+      "loss": 0.773,
+      "step": 4272
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.3079804333562283e-05,
+      "loss": 0.7883,
+      "step": 4273
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.306961082049229e-05,
+      "loss": 0.9345,
+      "step": 4274
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.3059417630268734e-05,
+      "loss": 0.6411,
+      "step": 4275
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.304922476459637e-05,
+      "loss": 0.6932,
+      "step": 4276
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.3039032225179906e-05,
+      "loss": 0.6869,
+      "step": 4277
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.3028840013723982e-05,
+      "loss": 0.7464,
+      "step": 4278
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.3018648131933186e-05,
+      "loss": 0.7043,
+      "step": 4279
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.300845658151205e-05,
+      "loss": 0.8016,
+      "step": 4280
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.2998265364165038e-05,
+      "loss": 0.7348,
+      "step": 4281
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.2988074481596605e-05,
+      "loss": 0.7839,
+      "step": 4282
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.2977883935511094e-05,
+      "loss": 0.635,
+      "step": 4283
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.296769372761282e-05,
+      "loss": 0.6277,
+      "step": 4284
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.295750385960603e-05,
+      "loss": 0.4828,
+      "step": 4285
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.2947314333194936e-05,
+      "loss": 0.8066,
+      "step": 4286
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.2937125150083653e-05,
+      "loss": 0.7783,
+      "step": 4287
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.2926936311976293e-05,
+      "loss": 0.91,
+      "step": 4288
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.291674782057686e-05,
+      "loss": 0.9141,
+      "step": 4289
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.2906559677589325e-05,
+      "loss": 0.8706,
+      "step": 4290
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.28963718847176e-05,
+      "loss": 0.8401,
+      "step": 4291
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.2886184443665525e-05,
+      "loss": 0.922,
+      "step": 4292
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.28759973561369e-05,
+      "loss": 0.6396,
+      "step": 4293
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.2865810623835464e-05,
+      "loss": 0.7966,
+      "step": 4294
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.2855624248464868e-05,
+      "loss": 0.4471,
+      "step": 4295
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.284543823172874e-05,
+      "loss": 0.6062,
+      "step": 4296
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.283525257533062e-05,
+      "loss": 0.8253,
+      "step": 4297
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.2825067280974014e-05,
+      "loss": 0.7704,
+      "step": 4298
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.281488235036235e-05,
+      "loss": 0.758,
+      "step": 4299
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.2804697785198997e-05,
+      "loss": 0.6723,
+      "step": 4300
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.2794513587187263e-05,
+      "loss": 0.8705,
+      "step": 4301
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.2784329758030395e-05,
+      "loss": 0.7042,
+      "step": 4302
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.277414629943158e-05,
+      "loss": 0.7112,
+      "step": 4303
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.2763963213093946e-05,
+      "loss": 0.6549,
+      "step": 4304
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.275378050072056e-05,
+      "loss": 0.8677,
+      "step": 4305
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.2743598164014416e-05,
+      "loss": 0.7893,
+      "step": 4306
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.273341620467845e-05,
+      "loss": 0.7561,
+      "step": 4307
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.2723234624415527e-05,
+      "loss": 0.7148,
+      "step": 4308
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.2713053424928484e-05,
+      "loss": 0.7321,
+      "step": 4309
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.2702872607920038e-05,
+      "loss": 0.859,
+      "step": 4310
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.2692692175092893e-05,
+      "loss": 0.6619,
+      "step": 4311
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.2682512128149662e-05,
+      "loss": 0.7722,
+      "step": 4312
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.2672332468792882e-05,
+      "loss": 0.7684,
+      "step": 4313
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.2662153198725072e-05,
+      "loss": 0.7225,
+      "step": 4314
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.2651974319648643e-05,
+      "loss": 0.6897,
+      "step": 4315
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.2641795833265948e-05,
+      "loss": 0.7193,
+      "step": 4316
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.2631617741279283e-05,
+      "loss": 0.7397,
+      "step": 4317
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.2621440045390882e-05,
+      "loss": 0.754,
+      "step": 4318
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.2611262747302882e-05,
+      "loss": 0.9553,
+      "step": 4319
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.260108584871741e-05,
+      "loss": 0.6382,
+      "step": 4320
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.2590909351336485e-05,
+      "loss": 0.6847,
+      "step": 4321
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.2580733256862053e-05,
+      "loss": 0.7211,
+      "step": 4322
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.2570557566996018e-05,
+      "loss": 0.7389,
+      "step": 4323
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.2560382283440195e-05,
+      "loss": 0.809,
+      "step": 4324
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.2550207407896358e-05,
+      "loss": 0.7924,
+      "step": 4325
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.2540032942066183e-05,
+      "loss": 0.6413,
+      "step": 4326
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.2529858887651305e-05,
+      "loss": 0.7276,
+      "step": 4327
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.2519685246353257e-05,
+      "loss": 0.6259,
+      "step": 4328
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.2509512019873536e-05,
+      "loss": 0.786,
+      "step": 4329
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.249933920991354e-05,
+      "loss": 0.6749,
+      "step": 4330
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.248916681817463e-05,
+      "loss": 0.6521,
+      "step": 4331
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.2478994846358075e-05,
+      "loss": 0.6487,
+      "step": 4332
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.246882329616508e-05,
+      "loss": 0.6969,
+      "step": 4333
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.245865216929677e-05,
+      "loss": 0.5266,
+      "step": 4334
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.2448481467454205e-05,
+      "loss": 0.6864,
+      "step": 4335
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.243831119233839e-05,
+      "loss": 0.7777,
+      "step": 4336
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.2428141345650238e-05,
+      "loss": 0.6453,
+      "step": 4337
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.241797192909059e-05,
+      "loss": 0.7705,
+      "step": 4338
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.240780294436024e-05,
+      "loss": 0.7515,
+      "step": 4339
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.2397634393159867e-05,
+      "loss": 0.677,
+      "step": 4340
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.238746627719013e-05,
+      "loss": 0.6039,
+      "step": 4341
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.2377298598151565e-05,
+      "loss": 0.7605,
+      "step": 4342
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.2367131357744673e-05,
+      "loss": 0.6294,
+      "step": 4343
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.2356964557669854e-05,
+      "loss": 0.8514,
+      "step": 4344
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.234679819962746e-05,
+      "loss": 0.5405,
+      "step": 4345
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.2336632285317735e-05,
+      "loss": 0.8268,
+      "step": 4346
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.2326466816440896e-05,
+      "loss": 0.6146,
+      "step": 4347
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.2316301794697043e-05,
+      "loss": 0.5689,
+      "step": 4348
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.2306137221786217e-05,
+      "loss": 0.6785,
+      "step": 4349
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.2295973099408387e-05,
+      "loss": 0.8125,
+      "step": 4350
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.228580942926344e-05,
+      "loss": 0.727,
+      "step": 4351
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.2275646213051202e-05,
+      "loss": 0.8884,
+      "step": 4352
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.226548345247141e-05,
+      "loss": 0.5642,
+      "step": 4353
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 2.2255321149223713e-05,
+      "loss": 0.7525,
+      "step": 4354
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.224515930500771e-05,
+      "loss": 0.6593,
+      "step": 4355
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.2234997921522903e-05,
+      "loss": 0.5983,
+      "step": 4356
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.222483700046873e-05,
+      "loss": 0.4947,
+      "step": 4357
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.221467654354455e-05,
+      "loss": 0.5772,
+      "step": 4358
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.2204516552449638e-05,
+      "loss": 0.8037,
+      "step": 4359
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.2194357028883188e-05,
+      "loss": 0.5998,
+      "step": 4360
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.218419797454433e-05,
+      "loss": 0.7102,
+      "step": 4361
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.2174039391132095e-05,
+      "loss": 0.8048,
+      "step": 4362
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.216388128034546e-05,
+      "loss": 0.7104,
+      "step": 4363
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.2153723643883307e-05,
+      "loss": 0.7205,
+      "step": 4364
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.2143566483444447e-05,
+      "loss": 0.6937,
+      "step": 4365
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.2133409800727596e-05,
+      "loss": 0.7686,
+      "step": 4366
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.2123253597431396e-05,
+      "loss": 0.763,
+      "step": 4367
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.211309787525443e-05,
+      "loss": 0.6609,
+      "step": 4368
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.2102942635895175e-05,
+      "loss": 0.7047,
+      "step": 4369
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.2092787881052034e-05,
+      "loss": 0.6695,
+      "step": 4370
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.2082633612423345e-05,
+      "loss": 0.6739,
+      "step": 4371
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.2072479831707335e-05,
+      "loss": 0.7225,
+      "step": 4372
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.206232654060216e-05,
+      "loss": 0.7659,
+      "step": 4373
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.2052173740805923e-05,
+      "loss": 0.8638,
+      "step": 4374
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.2042021434016603e-05,
+      "loss": 0.7052,
+      "step": 4375
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.203186962193212e-05,
+      "loss": 0.6871,
+      "step": 4376
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.202171830625031e-05,
+      "loss": 0.7439,
+      "step": 4377
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.2011567488668904e-05,
+      "loss": 0.6994,
+      "step": 4378
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.2001417170885596e-05,
+      "loss": 0.6847,
+      "step": 4379
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.1991267354597947e-05,
+      "loss": 0.6729,
+      "step": 4380
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.1981118041503464e-05,
+      "loss": 0.6338,
+      "step": 4381
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.1970969233299554e-05,
+      "loss": 0.7306,
+      "step": 4382
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.196082093168355e-05,
+      "loss": 0.6046,
+      "step": 4383
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.19506731383527e-05,
+      "loss": 0.6551,
+      "step": 4384
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.1940525855004167e-05,
+      "loss": 0.8873,
+      "step": 4385
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.1930379083335017e-05,
+      "loss": 0.9373,
+      "step": 4386
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.1920232825042242e-05,
+      "loss": 0.9102,
+      "step": 4387
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.1910087081822746e-05,
+      "loss": 0.673,
+      "step": 4388
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 2.1899941855373334e-05,
+      "loss": 0.7639,
+      "step": 4389
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.188979714739076e-05,
+      "loss": 0.6552,
+      "step": 4390
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.1879652959571658e-05,
+      "loss": 0.9245,
+      "step": 4391
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.186950929361258e-05,
+      "loss": 0.6946,
+      "step": 4392
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.1859366151209994e-05,
+      "loss": 0.6491,
+      "step": 4393
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.184922353406028e-05,
+      "loss": 0.7961,
+      "step": 4394
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.1839081443859746e-05,
+      "loss": 0.6562,
+      "step": 4395
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.1828939882304592e-05,
+      "loss": 0.8058,
+      "step": 4396
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.181879885109094e-05,
+      "loss": 0.7979,
+      "step": 4397
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.1808658351914802e-05,
+      "loss": 0.6655,
+      "step": 4398
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.1798518386472125e-05,
+      "loss": 0.5881,
+      "step": 4399
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.178837895645877e-05,
+      "loss": 0.7834,
+      "step": 4400
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.1778240063570494e-05,
+      "loss": 0.6819,
+      "step": 4401
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.176810170950296e-05,
+      "loss": 0.6257,
+      "step": 4402
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.1757963895951766e-05,
+      "loss": 0.7841,
+      "step": 4403
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.174782662461238e-05,
+      "loss": 0.5623,
+      "step": 4404
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.1737689897180215e-05,
+      "loss": 0.6321,
+      "step": 4405
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.1727553715350583e-05,
+      "loss": 0.536,
+      "step": 4406
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.1717418080818696e-05,
+      "loss": 0.5328,
+      "step": 4407
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.1707282995279683e-05,
+      "loss": 0.9435,
+      "step": 4408
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.1697148460428584e-05,
+      "loss": 0.7378,
+      "step": 4409
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.168701447796032e-05,
+      "loss": 0.5885,
+      "step": 4410
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.1676881049569768e-05,
+      "loss": 0.5153,
+      "step": 4411
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.1666748176951673e-05,
+      "loss": 0.6571,
+      "step": 4412
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.16566158618007e-05,
+      "loss": 0.7968,
+      "step": 4413
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.1646484105811417e-05,
+      "loss": 0.8035,
+      "step": 4414
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.1636352910678302e-05,
+      "loss": 0.6199,
+      "step": 4415
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.1626222278095746e-05,
+      "loss": 0.6942,
+      "step": 4416
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.161609220975804e-05,
+      "loss": 0.7709,
+      "step": 4417
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.1605962707359368e-05,
+      "loss": 0.6072,
+      "step": 4418
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.159583377259384e-05,
+      "loss": 0.6148,
+      "step": 4419
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.1585705407155456e-05,
+      "loss": 0.5602,
+      "step": 4420
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.1575577612738126e-05,
+      "loss": 0.8875,
+      "step": 4421
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.1565450391035673e-05,
+      "loss": 0.7716,
+      "step": 4422
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.155532374374182e-05,
+      "loss": 0.7205,
+      "step": 4423
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.154519767255017e-05,
+      "loss": 0.6962,
+      "step": 4424
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 2.1535072179154268e-05,
+      "loss": 0.9021,
+      "step": 4425
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.1524947265247534e-05,
+      "loss": 0.7249,
+      "step": 4426
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.1514822932523308e-05,
+      "loss": 0.7154,
+      "step": 4427
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.1504699182674827e-05,
+      "loss": 0.8363,
+      "step": 4428
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.1494576017395228e-05,
+      "loss": 1.0036,
+      "step": 4429
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.148445343837755e-05,
+      "loss": 0.4468,
+      "step": 4430
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.1474331447314732e-05,
+      "loss": 0.7627,
+      "step": 4431
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.1464210045899622e-05,
+      "loss": 0.7248,
+      "step": 4432
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.145408923582497e-05,
+      "loss": 0.5239,
+      "step": 4433
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.144396901878342e-05,
+      "loss": 0.5588,
+      "step": 4434
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.1433849396467525e-05,
+      "loss": 0.8435,
+      "step": 4435
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.1423730370569726e-05,
+      "loss": 0.7439,
+      "step": 4436
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.141361194278237e-05,
+      "loss": 0.8008,
+      "step": 4437
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.140349411479771e-05,
+      "loss": 0.8005,
+      "step": 4438
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.13933768883079e-05,
+      "loss": 0.7247,
+      "step": 4439
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.1383260265004974e-05,
+      "loss": 0.7721,
+      "step": 4440
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.1373144246580897e-05,
+      "loss": 0.7814,
+      "step": 4441
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.1363028834727488e-05,
+      "loss": 0.7385,
+      "step": 4442
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.1352914031136523e-05,
+      "loss": 0.8478,
+      "step": 4443
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.134279983749962e-05,
+      "loss": 0.6871,
+      "step": 4444
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.1332686255508337e-05,
+      "loss": 0.544,
+      "step": 4445
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.13225732868541e-05,
+      "loss": 0.895,
+      "step": 4446
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.131246093322825e-05,
+      "loss": 0.8969,
+      "step": 4447
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.1302349196322e-05,
+      "loss": 0.5862,
+      "step": 4448
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.1292238077826522e-05,
+      "loss": 0.7049,
+      "step": 4449
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.1282127579432808e-05,
+      "loss": 0.7353,
+      "step": 4450
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.1272017702831794e-05,
+      "loss": 0.8137,
+      "step": 4451
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.126190844971429e-05,
+      "loss": 0.6143,
+      "step": 4452
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.1251799821771012e-05,
+      "loss": 0.6878,
+      "step": 4453
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.1241691820692576e-05,
+      "loss": 0.9958,
+      "step": 4454
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.1231584448169494e-05,
+      "loss": 0.8054,
+      "step": 4455
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.1221477705892146e-05,
+      "loss": 0.7559,
+      "step": 4456
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.1211371595550836e-05,
+      "loss": 0.7093,
+      "step": 4457
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.1201266118835748e-05,
+      "loss": 0.4979,
+      "step": 4458
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.1191161277436973e-05,
+      "loss": 0.5109,
+      "step": 4459
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.1181057073044484e-05,
+      "loss": 0.8047,
+      "step": 4460
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.1170953507348154e-05,
+      "loss": 0.8051,
+      "step": 4461
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.116085058203774e-05,
+      "loss": 0.7282,
+      "step": 4462
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.1150748298802895e-05,
+      "loss": 0.7654,
+      "step": 4463
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.1140646659333168e-05,
+      "loss": 0.5285,
+      "step": 4464
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.1130545665318014e-05,
+      "loss": 0.8782,
+      "step": 4465
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.112044531844675e-05,
+      "loss": 0.5642,
+      "step": 4466
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.111034562040862e-05,
+      "loss": 0.5694,
+      "step": 4467
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.1100246572892717e-05,
+      "loss": 0.6618,
+      "step": 4468
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.1090148177588055e-05,
+      "loss": 0.6569,
+      "step": 4469
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.108005043618354e-05,
+      "loss": 0.7327,
+      "step": 4470
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.1069953350367953e-05,
+      "loss": 0.7575,
+      "step": 4471
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.1059856921829986e-05,
+      "loss": 0.6385,
+      "step": 4472
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.10497611522582e-05,
+      "loss": 0.6331,
+      "step": 4473
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.1039666043341034e-05,
+      "loss": 0.5744,
+      "step": 4474
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.102957159676688e-05,
+      "loss": 0.5462,
+      "step": 4475
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.1019477814223944e-05,
+      "loss": 0.5761,
+      "step": 4476
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.1009384697400358e-05,
+      "loss": 0.7947,
+      "step": 4477
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.0999292247984144e-05,
+      "loss": 0.7814,
+      "step": 4478
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.0989200467663206e-05,
+      "loss": 0.5619,
+      "step": 4479
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.097910935812532e-05,
+      "loss": 0.8935,
+      "step": 4480
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.096901892105819e-05,
+      "loss": 0.5708,
+      "step": 4481
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.095892915814937e-05,
+      "loss": 0.8848,
+      "step": 4482
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.0948840071086313e-05,
+      "loss": 0.8786,
+      "step": 4483
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.0938751661556363e-05,
+      "loss": 0.6921,
+      "step": 4484
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.092866393124675e-05,
+      "loss": 0.9069,
+      "step": 4485
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.0918576881844586e-05,
+      "loss": 0.524,
+      "step": 4486
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.0908490515036883e-05,
+      "loss": 0.8063,
+      "step": 4487
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.089840483251051e-05,
+      "loss": 0.6548,
+      "step": 4488
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.088831983595225e-05,
+      "loss": 0.692,
+      "step": 4489
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.0878235527048757e-05,
+      "loss": 0.8175,
+      "step": 4490
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.0868151907486567e-05,
+      "loss": 0.7007,
+      "step": 4491
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.0858068978952124e-05,
+      "loss": 0.4749,
+      "step": 4492
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.084798674313173e-05,
+      "loss": 0.7074,
+      "step": 4493
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.083790520171158e-05,
+      "loss": 0.8817,
+      "step": 4494
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.0827824356377752e-05,
+      "loss": 0.6779,
+      "step": 4495
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.0817744208816203e-05,
+      "loss": 0.6715,
+      "step": 4496
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.0807664760712798e-05,
+      "loss": 0.7602,
+      "step": 4497
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.0797586013753254e-05,
+      "loss": 0.7129,
+      "step": 4498
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.0787507969623192e-05,
+      "loss": 0.8848,
+      "step": 4499
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.0777430630008094e-05,
+      "loss": 0.6463,
+      "step": 4500
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.076735399659334e-05,
+      "loss": 0.7948,
+      "step": 4501
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.0757278071064196e-05,
+      "loss": 0.6722,
+      "step": 4502
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.0747202855105797e-05,
+      "loss": 0.592,
+      "step": 4503
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.0737128350403166e-05,
+      "loss": 0.8382,
+      "step": 4504
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.072705455864121e-05,
+      "loss": 0.5995,
+      "step": 4505
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.0716981481504708e-05,
+      "loss": 0.7781,
+      "step": 4506
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.0706909120678313e-05,
+      "loss": 0.6559,
+      "step": 4507
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.0696837477846593e-05,
+      "loss": 0.9329,
+      "step": 4508
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.0686766554693952e-05,
+      "loss": 0.5454,
+      "step": 4509
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.0676696352904707e-05,
+      "loss": 0.7497,
+      "step": 4510
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.0666626874163037e-05,
+      "loss": 0.8057,
+      "step": 4511
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.065655812015299e-05,
+      "loss": 0.7616,
+      "step": 4512
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.064649009255853e-05,
+      "loss": 0.7121,
+      "step": 4513
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.0636422793063465e-05,
+      "loss": 0.7014,
+      "step": 4514
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.0626356223351495e-05,
+      "loss": 0.814,
+      "step": 4515
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.061629038510619e-05,
+      "loss": 0.7366,
+      "step": 4516
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.0606225280011004e-05,
+      "loss": 0.7457,
+      "step": 4517
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.0596160909749278e-05,
+      "loss": 0.6751,
+      "step": 4518
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.058609727600422e-05,
+      "loss": 0.9158,
+      "step": 4519
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.05760343804589e-05,
+      "loss": 0.7635,
+      "step": 4520
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.0565972224796286e-05,
+      "loss": 0.9777,
+      "step": 4521
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.0555910810699223e-05,
+      "loss": 0.6112,
+      "step": 4522
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.0545850139850407e-05,
+      "loss": 0.5645,
+      "step": 4523
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.0535790213932442e-05,
+      "loss": 0.7345,
+      "step": 4524
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.05257310346278e-05,
+      "loss": 0.4875,
+      "step": 4525
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.05156726036188e-05,
+      "loss": 0.649,
+      "step": 4526
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.0505614922587667e-05,
+      "loss": 0.6229,
+      "step": 4527
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.0495557993216478e-05,
+      "loss": 0.6718,
+      "step": 4528
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.0485501817187218e-05,
+      "loss": 0.6048,
+      "step": 4529
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.0475446396181717e-05,
+      "loss": 0.7579,
+      "step": 4530
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.0465391731881682e-05,
+      "loss": 0.6014,
+      "step": 4531
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.0455337825968696e-05,
+      "loss": 0.7456,
+      "step": 4532
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.044528468012421e-05,
+      "loss": 0.6368,
+      "step": 4533
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.0435232296029573e-05,
+      "loss": 0.8143,
+      "step": 4534
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.042518067536598e-05,
+      "loss": 0.7751,
+      "step": 4535
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.04151298198145e-05,
+      "loss": 0.6939,
+      "step": 4536
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.0405079731056096e-05,
+      "loss": 0.6678,
+      "step": 4537
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.0395030410771568e-05,
+      "loss": 0.6672,
+      "step": 4538
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.0384981860641606e-05,
+      "loss": 0.8023,
+      "step": 4539
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.0374934082346798e-05,
+      "loss": 0.7095,
+      "step": 4540
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.0364887077567557e-05,
+      "loss": 0.5871,
+      "step": 4541
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.0354840847984192e-05,
+      "loss": 0.728,
+      "step": 4542
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.0344795395276878e-05,
+      "loss": 0.5036,
+      "step": 4543
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.033475072112564e-05,
+      "loss": 0.9811,
+      "step": 4544
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.032470682721042e-05,
+      "loss": 0.7235,
+      "step": 4545
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.0314663715210994e-05,
+      "loss": 0.8497,
+      "step": 4546
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.0304621386807004e-05,
+      "loss": 0.8331,
+      "step": 4547
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.029457984367798e-05,
+      "loss": 0.6998,
+      "step": 4548
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.0284539087503313e-05,
+      "loss": 0.7591,
+      "step": 4549
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.0274499119962238e-05,
+      "loss": 0.6327,
+      "step": 4550
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.0264459942733915e-05,
+      "loss": 0.6858,
+      "step": 4551
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.0254421557497334e-05,
+      "loss": 0.709,
+      "step": 4552
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.0244383965931337e-05,
+      "loss": 0.8145,
+      "step": 4553
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.0234347169714668e-05,
+      "loss": 0.7342,
+      "step": 4554
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.0224311170525912e-05,
+      "loss": 0.7729,
+      "step": 4555
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.0214275970043543e-05,
+      "loss": 0.6173,
+      "step": 4556
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.020424156994589e-05,
+      "loss": 0.7026,
+      "step": 4557
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.019420797191115e-05,
+      "loss": 0.7992,
+      "step": 4558
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.018417517761737e-05,
+      "loss": 0.7574,
+      "step": 4559
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.0174143188742483e-05,
+      "loss": 0.6986,
+      "step": 4560
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.0164112006964293e-05,
+      "loss": 0.6427,
+      "step": 4561
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.0154081633960446e-05,
+      "loss": 0.6726,
+      "step": 4562
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.0144052071408467e-05,
+      "loss": 0.6164,
+      "step": 4563
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.0134023320985748e-05,
+      "loss": 0.7046,
+      "step": 4564
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.012399538436953e-05,
+      "loss": 0.7374,
+      "step": 4565
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.011396826323692e-05,
+      "loss": 0.8067,
+      "step": 4566
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.010394195926492e-05,
+      "loss": 0.8237,
+      "step": 4567
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.0093916474130353e-05,
+      "loss": 0.4731,
+      "step": 4568
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.0083891809509935e-05,
+      "loss": 0.5446,
+      "step": 4569
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.0073867967080227e-05,
+      "loss": 0.6676,
+      "step": 4570
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.0063844948517645e-05,
+      "loss": 0.7662,
+      "step": 4571
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.0053822755498513e-05,
+      "loss": 0.7354,
+      "step": 4572
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.0043801389698963e-05,
+      "loss": 0.7111,
+      "step": 4573
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.0043801389698963e-05,
+      "loss": 0.7102,
+      "step": 4574
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.0033780852795014e-05,
+      "loss": 0.7531,
+      "step": 4575
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.002376114646254e-05,
+      "loss": 0.5529,
+      "step": 4576
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.0013742272377284e-05,
+      "loss": 0.6794,
+      "step": 4577
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.0003724232214843e-05,
+      "loss": 0.9046,
+      "step": 4578
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.9993707027650686e-05,
+      "loss": 0.7234,
+      "step": 4579
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.998369066036012e-05,
+      "loss": 0.7481,
+      "step": 4580
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.9973675132018325e-05,
+      "loss": 0.7548,
+      "step": 4581
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.9963660444300342e-05,
+      "loss": 0.7285,
+      "step": 4582
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.9953646598881066e-05,
+      "loss": 0.7007,
+      "step": 4583
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.9943633597435272e-05,
+      "loss": 0.7042,
+      "step": 4584
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.9933621441637564e-05,
+      "loss": 0.6634,
+      "step": 4585
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.9923610133162413e-05,
+      "loss": 0.7439,
+      "step": 4586
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.991359967368416e-05,
+      "loss": 0.6569,
+      "step": 4587
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.9903590064876987e-05,
+      "loss": 0.4944,
+      "step": 4588
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.989358130841496e-05,
+      "loss": 0.7991,
+      "step": 4589
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.9883573405971968e-05,
+      "loss": 0.6849,
+      "step": 4590
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.9873566359221797e-05,
+      "loss": 0.7797,
+      "step": 4591
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.9863560169838042e-05,
+      "loss": 0.6092,
+      "step": 4592
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.9853554839494188e-05,
+      "loss": 0.7658,
+      "step": 4593
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.984355036986358e-05,
+      "loss": 0.6158,
+      "step": 4594
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.98335467626194e-05,
+      "loss": 0.5727,
+      "step": 4595
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.9823544019434694e-05,
+      "loss": 0.7113,
+      "step": 4596
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.981354214198237e-05,
+      "loss": 0.8315,
+      "step": 4597
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.9803541131935173e-05,
+      "loss": 0.585,
+      "step": 4598
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.9793540990965707e-05,
+      "loss": 0.7532,
+      "step": 4599
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.9783541720746464e-05,
+      "loss": 0.699,
+      "step": 4600
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.9773543322949743e-05,
+      "loss": 0.8774,
+      "step": 4601
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.976354579924773e-05,
+      "loss": 0.8035,
+      "step": 4602
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.9753549151312452e-05,
+      "loss": 0.5839,
+      "step": 4603
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.9743553380815767e-05,
+      "loss": 0.6747,
+      "step": 4604
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.973355848942945e-05,
+      "loss": 0.7915,
+      "step": 4605
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.9723564478825062e-05,
+      "loss": 0.7868,
+      "step": 4606
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.971357135067405e-05,
+      "loss": 0.6883,
+      "step": 4607
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.9703579106647707e-05,
+      "loss": 0.7712,
+      "step": 4608
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.9693587748417186e-05,
+      "loss": 0.7293,
+      "step": 4609
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.968359727765346e-05,
+      "loss": 0.6125,
+      "step": 4610
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.9673607696027407e-05,
+      "loss": 0.5207,
+      "step": 4611
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.966361900520971e-05,
+      "loss": 0.7894,
+      "step": 4612
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.9653631206870924e-05,
+      "loss": 0.7114,
+      "step": 4613
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.964364430268145e-05,
+      "loss": 0.5838,
+      "step": 4614
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.9633658294311535e-05,
+      "loss": 0.6787,
+      "step": 4615
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.962367318343129e-05,
+      "loss": 0.6657,
+      "step": 4616
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.961368897171067e-05,
+      "loss": 0.5513,
+      "step": 4617
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.9603705660819466e-05,
+      "loss": 0.6903,
+      "step": 4618
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.9593723252427336e-05,
+      "loss": 0.8316,
+      "step": 4619
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.9583741748203768e-05,
+      "loss": 0.6853,
+      "step": 4620
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.9573761149818125e-05,
+      "loss": 0.7266,
+      "step": 4621
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.9563781458939604e-05,
+      "loss": 0.8511,
+      "step": 4622
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.9553802677237252e-05,
+      "loss": 0.6108,
+      "step": 4623
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.954382480637995e-05,
+      "loss": 0.7622,
+      "step": 4624
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.9533847848036445e-05,
+      "loss": 0.5756,
+      "step": 4625
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.9523871803875326e-05,
+      "loss": 0.6686,
+      "step": 4626
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.9513896675565034e-05,
+      "loss": 0.7436,
+      "step": 4627
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.950392246477385e-05,
+      "loss": 0.6214,
+      "step": 4628
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.94939491731699e-05,
+      "loss": 0.6522,
+      "step": 4629
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.948397680242116e-05,
+      "loss": 0.5356,
+      "step": 4630
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.9474005354195447e-05,
+      "loss": 0.6746,
+      "step": 4631
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.9464034830160435e-05,
+      "loss": 0.6106,
+      "step": 4632
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.9454065231983637e-05,
+      "loss": 0.6454,
+      "step": 4633
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.944409656133241e-05,
+      "loss": 0.7072,
+      "step": 4634
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.9434128819873955e-05,
+      "loss": 0.8227,
+      "step": 4635
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.9424162009275304e-05,
+      "loss": 0.6851,
+      "step": 4636
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.941419613120338e-05,
+      "loss": 0.5913,
+      "step": 4637
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.9404231187324902e-05,
+      "loss": 0.8682,
+      "step": 4638
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.9394267179306444e-05,
+      "loss": 0.6862,
+      "step": 4639
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.9384304108814438e-05,
+      "loss": 0.7286,
+      "step": 4640
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.9374341977515148e-05,
+      "loss": 0.7591,
+      "step": 4641
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.936438078707467e-05,
+      "loss": 0.6652,
+      "step": 4642
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.9354420539158975e-05,
+      "loss": 0.6875,
+      "step": 4643
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.934446123543385e-05,
+      "loss": 0.807,
+      "step": 4644
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.933450287756493e-05,
+      "loss": 0.7342,
+      "step": 4645
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.932454546721769e-05,
+      "loss": 0.6721,
+      "step": 4646
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.9314589006057447e-05,
+      "loss": 0.5883,
+      "step": 4647
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.930463349574937e-05,
+      "loss": 0.7511,
+      "step": 4648
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.9294678937958468e-05,
+      "loss": 0.7023,
+      "step": 4649
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.9284725334349566e-05,
+      "loss": 0.708,
+      "step": 4650
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.927477268658735e-05,
+      "loss": 0.584,
+      "step": 4651
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.9264820996336345e-05,
+      "loss": 0.479,
+      "step": 4652
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.9254870265260916e-05,
+      "loss": 0.8368,
+      "step": 4653
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.924492049502527e-05,
+      "loss": 0.6157,
+      "step": 4654
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.9234971687293442e-05,
+      "loss": 0.7021,
+      "step": 4655
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.922502384372931e-05,
+      "loss": 0.7462,
+      "step": 4656
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.92150769659966e-05,
+      "loss": 0.6355,
+      "step": 4657
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.920513105575886e-05,
+      "loss": 0.7162,
+      "step": 4658
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.9195186114679496e-05,
+      "loss": 0.6194,
+      "step": 4659
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.9185242144421745e-05,
+      "loss": 0.3954,
+      "step": 4660
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.9175299146648674e-05,
+      "loss": 0.8994,
+      "step": 4661
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.9165357123023184e-05,
+      "loss": 0.8122,
+      "step": 4662
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.9155416075208025e-05,
+      "loss": 0.718,
+      "step": 4663
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.9145476004865787e-05,
+      "loss": 0.9021,
+      "step": 4664
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.913553691365888e-05,
+      "loss": 0.8453,
+      "step": 4665
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.912559880324957e-05,
+      "loss": 0.6114,
+      "step": 4666
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.911566167529995e-05,
+      "loss": 0.7138,
+      "step": 4667
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.9105725531471928e-05,
+      "loss": 0.7498,
+      "step": 4668
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.909579037342728e-05,
+      "loss": 0.7499,
+      "step": 4669
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.9085856202827605e-05,
+      "loss": 0.9045,
+      "step": 4670
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.907592302133433e-05,
+      "loss": 0.5993,
+      "step": 4671
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.906599083060873e-05,
+      "loss": 0.6518,
+      "step": 4672
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.905605963231191e-05,
+      "loss": 0.6526,
+      "step": 4673
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.904612942810478e-05,
+      "loss": 0.7041,
+      "step": 4674
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.903620021964814e-05,
+      "loss": 0.6642,
+      "step": 4675
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.9026272008602576e-05,
+      "loss": 0.7722,
+      "step": 4676
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.9016344796628527e-05,
+      "loss": 0.7031,
+      "step": 4677
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.9006418585386266e-05,
+      "loss": 0.7379,
+      "step": 4678
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.8996493376535884e-05,
+      "loss": 0.6028,
+      "step": 4679
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.898656917173733e-05,
+      "loss": 0.5957,
+      "step": 4680
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.8976645972650363e-05,
+      "loss": 0.4483,
+      "step": 4681
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.8966723780934574e-05,
+      "loss": 0.7748,
+      "step": 4682
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.8956802598249403e-05,
+      "loss": 0.7473,
+      "step": 4683
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.8946882426254105e-05,
+      "loss": 0.8751,
+      "step": 4684
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.8936963266607764e-05,
+      "loss": 0.878,
+      "step": 4685
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.892704512096932e-05,
+      "loss": 0.833,
+      "step": 4686
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.891712799099752e-05,
+      "loss": 0.8262,
+      "step": 4687
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.8907211878350934e-05,
+      "loss": 0.8947,
+      "step": 4688
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.8897296784687984e-05,
+      "loss": 0.6098,
+      "step": 4689
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.8887382711666907e-05,
+      "loss": 0.7577,
+      "step": 4690
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.8877469660945784e-05,
+      "loss": 0.4155,
+      "step": 4691
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.886755763418251e-05,
+      "loss": 0.567,
+      "step": 4692
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.8857646633034823e-05,
+      "loss": 0.7965,
+      "step": 4693
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.8847736659160265e-05,
+      "loss": 0.7375,
+      "step": 4694
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.8837827714216224e-05,
+      "loss": 0.717,
+      "step": 4695
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.8827919799859926e-05,
+      "loss": 0.652,
+      "step": 4696
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.8818012917748402e-05,
+      "loss": 0.8377,
+      "step": 4697
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.8808107069538528e-05,
+      "loss": 0.6705,
+      "step": 4698
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.8798202256886998e-05,
+      "loss": 0.6792,
+      "step": 4699
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.878829848145033e-05,
+      "loss": 0.6242,
+      "step": 4700
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.8778395744884873e-05,
+      "loss": 0.839,
+      "step": 4701
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.8768494048846806e-05,
+      "loss": 0.7586,
+      "step": 4702
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.8758593394992135e-05,
+      "loss": 0.7236,
+      "step": 4703
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.8748693784976677e-05,
+      "loss": 0.6832,
+      "step": 4704
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.87387952204561e-05,
+      "loss": 0.6994,
+      "step": 4705
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.8728897703085855e-05,
+      "loss": 0.8338,
+      "step": 4706
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.8719001234521283e-05,
+      "loss": 0.6229,
+      "step": 4707
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.8709105816417477e-05,
+      "loss": 0.7387,
+      "step": 4708
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.8699211450429406e-05,
+      "loss": 0.729,
+      "step": 4709
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.8689318138211836e-05,
+      "loss": 0.6846,
+      "step": 4710
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.867942588141937e-05,
+      "loss": 0.6488,
+      "step": 4711
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.8669534681706435e-05,
+      "loss": 0.6963,
+      "step": 4712
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.8659644540727277e-05,
+      "loss": 0.7009,
+      "step": 4713
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.8649755460135964e-05,
+      "loss": 0.7238,
+      "step": 4714
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.8649755460135964e-05,
+      "loss": 0.9311,
+      "step": 4715
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.8639867441586377e-05,
+      "loss": 0.6069,
+      "step": 4716
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.8629980486732245e-05,
+      "loss": 0.642,
+      "step": 4717
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.8620094597227093e-05,
+      "loss": 0.6843,
+      "step": 4718
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.8610209774724287e-05,
+      "loss": 0.7097,
+      "step": 4719
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.8600326020877007e-05,
+      "loss": 0.7813,
+      "step": 4720
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.859044333733825e-05,
+      "loss": 0.7538,
+      "step": 4721
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.858056172576083e-05,
+      "loss": 0.6076,
+      "step": 4722
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.8570681187797402e-05,
+      "loss": 0.6971,
+      "step": 4723
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.856080172510042e-05,
+      "loss": 0.5881,
+      "step": 4724
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.8550923339322176e-05,
+      "loss": 0.7519,
+      "step": 4725
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.8541046032114773e-05,
+      "loss": 0.6339,
+      "step": 4726
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.853116980513012e-05,
+      "loss": 0.6175,
+      "step": 4727
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.8521294660019974e-05,
+      "loss": 0.6147,
+      "step": 4728
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.8511420598435884e-05,
+      "loss": 0.6538,
+      "step": 4729
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.8501547622029237e-05,
+      "loss": 0.4959,
+      "step": 4730
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.849167573245123e-05,
+      "loss": 0.6464,
+      "step": 4731
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.848180493135289e-05,
+      "loss": 0.7454,
+      "step": 4732
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.847193522038503e-05,
+      "loss": 0.6232,
+      "step": 4733
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.846206660119831e-05,
+      "loss": 0.7308,
+      "step": 4734
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.8452199075443212e-05,
+      "loss": 0.7179,
+      "step": 4735
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.8442332644770006e-05,
+      "loss": 0.6505,
+      "step": 4736
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.8432467310828806e-05,
+      "loss": 0.5822,
+      "step": 4737
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.8422603075269533e-05,
+      "loss": 0.7374,
+      "step": 4738
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.8412739939741907e-05,
+      "loss": 0.5917,
+      "step": 4739
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.840287790589551e-05,
+      "loss": 0.7978,
+      "step": 4740
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.8393016975379684e-05,
+      "loss": 0.5185,
+      "step": 4741
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.8383157149843624e-05,
+      "loss": 0.808,
+      "step": 4742
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.8373298430936327e-05,
+      "loss": 0.5905,
+      "step": 4743
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.836344082030661e-05,
+      "loss": 0.5374,
+      "step": 4744
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.8353584319603084e-05,
+      "loss": 0.6561,
+      "step": 4745
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.8343728930474226e-05,
+      "loss": 0.7826,
+      "step": 4746
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.833387465456826e-05,
+      "loss": 0.7025,
+      "step": 4747
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.832402149353327e-05,
+      "loss": 0.8561,
+      "step": 4748
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.8314169449017143e-05,
+      "loss": 0.546,
+      "step": 4749
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.8304318522667568e-05,
+      "loss": 0.7197,
+      "step": 4750
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.8294468716132067e-05,
+      "loss": 0.6324,
+      "step": 4751
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.8284620031057963e-05,
+      "loss": 0.5596,
+      "step": 4752
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.827477246909238e-05,
+      "loss": 0.4525,
+      "step": 4753
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.8264926031882272e-05,
+      "loss": 0.548,
+      "step": 4754
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.825508072107439e-05,
+      "loss": 0.779,
+      "step": 4755
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.8245236538315325e-05,
+      "loss": 0.563,
+      "step": 4756
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.8235393485251446e-05,
+      "loss": 0.6793,
+      "step": 4757
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.8225551563528956e-05,
+      "loss": 0.7843,
+      "step": 4758
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.8215710774793847e-05,
+      "loss": 0.6972,
+      "step": 4759
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.820587112069194e-05,
+      "loss": 0.696,
+      "step": 4760
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.8196032602868857e-05,
+      "loss": 0.6564,
+      "step": 4761
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.818619522297004e-05,
+      "loss": 0.7411,
+      "step": 4762
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.8176358982640733e-05,
+      "loss": 0.7336,
+      "step": 4763
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.8166523883525992e-05,
+      "loss": 0.6396,
+      "step": 4764
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.8156689927270672e-05,
+      "loss": 0.6636,
+      "step": 4765
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.8146857115519437e-05,
+      "loss": 0.6318,
+      "step": 4766
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.8137025449916804e-05,
+      "loss": 0.6489,
+      "step": 4767
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.812719493210703e-05,
+      "loss": 0.6863,
+      "step": 4768
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.8117365563734224e-05,
+      "loss": 0.7352,
+      "step": 4769
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.810753734644229e-05,
+      "loss": 0.8269,
+      "step": 4770
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.8097710281874928e-05,
+      "loss": 0.6719,
+      "step": 4771
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.808788437167568e-05,
+      "loss": 0.6544,
+      "step": 4772
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.8078059617487862e-05,
+      "loss": 0.7207,
+      "step": 4773
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.806823602095461e-05,
+      "loss": 0.6784,
+      "step": 4774
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.8058413583718856e-05,
+      "loss": 0.6515,
+      "step": 4775
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.8048592307423358e-05,
+      "loss": 0.6483,
+      "step": 4776
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.8038772193710646e-05,
+      "loss": 0.5977,
+      "step": 4777
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.80289532442231e-05,
+      "loss": 0.7026,
+      "step": 4778
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.8019135460602884e-05,
+      "loss": 0.5733,
+      "step": 4779
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.800931884449195e-05,
+      "loss": 0.6218,
+      "step": 4780
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.799950339753208e-05,
+      "loss": 0.8655,
+      "step": 4781
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.798968912136484e-05,
+      "loss": 0.9146,
+      "step": 4782
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.797987601763162e-05,
+      "loss": 0.8918,
+      "step": 4783
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.797006408797361e-05,
+      "loss": 0.6632,
+      "step": 4784
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.7960253334031793e-05,
+      "loss": 0.7229,
+      "step": 4785
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.795044375744696e-05,
+      "loss": 0.6344,
+      "step": 4786
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.79406353598597e-05,
+      "loss": 0.8942,
+      "step": 4787
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.7930828142910417e-05,
+      "loss": 0.6879,
+      "step": 4788
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.7921022108239315e-05,
+      "loss": 0.6176,
+      "step": 4789
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.7911217257486397e-05,
+      "loss": 0.7756,
+      "step": 4790
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.7901413592291464e-05,
+      "loss": 0.6305,
+      "step": 4791
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.789161111429412e-05,
+      "loss": 0.7893,
+      "step": 4792
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.7881809825133768e-05,
+      "loss": 0.7694,
+      "step": 4793
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.787200972644963e-05,
+      "loss": 0.6408,
+      "step": 4794
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.786221081988071e-05,
+      "loss": 0.5705,
+      "step": 4795
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.7852413107065823e-05,
+      "loss": 0.7556,
+      "step": 4796
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.7842616589643576e-05,
+      "loss": 0.6536,
+      "step": 4797
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.7832821269252364e-05,
+      "loss": 0.6018,
+      "step": 4798
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.782302714753043e-05,
+      "loss": 0.7428,
+      "step": 4799
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.7813234226115764e-05,
+      "loss": 0.5275,
+      "step": 4800
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.780344250664618e-05,
+      "loss": 0.6104,
+      "step": 4801
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.7793651990759285e-05,
+      "loss": 0.514,
+      "step": 4802
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.7783862680092488e-05,
+      "loss": 0.5134,
+      "step": 4803
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.7774074576282984e-05,
+      "loss": 0.9238,
+      "step": 4804
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.77642876809678e-05,
+      "loss": 0.7127,
+      "step": 4805
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.7754501995783717e-05,
+      "loss": 0.5572,
+      "step": 4806
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.7744717522367346e-05,
+      "loss": 0.4875,
+      "step": 4807
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.773493426235508e-05,
+      "loss": 0.6367,
+      "step": 4808
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.77251522173831e-05,
+      "loss": 0.7758,
+      "step": 4809
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.771537138908742e-05,
+      "loss": 0.7792,
+      "step": 4810
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.770559177910382e-05,
+      "loss": 0.593,
+      "step": 4811
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.7695813389067874e-05,
+      "loss": 0.6719,
+      "step": 4812
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.7686036220614964e-05,
+      "loss": 0.7476,
+      "step": 4813
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.7676260275380263e-05,
+      "loss": 0.5739,
+      "step": 4814
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.7666485554998752e-05,
+      "loss": 0.5779,
+      "step": 4815
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.7656712061105187e-05,
+      "loss": 0.5378,
+      "step": 4816
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.7646939795334138e-05,
+      "loss": 0.8699,
+      "step": 4817
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.7637168759319948e-05,
+      "loss": 0.7429,
+      "step": 4818
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.762739895469678e-05,
+      "loss": 0.6915,
+      "step": 4819
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.7617630383098556e-05,
+      "loss": 0.6533,
+      "step": 4820
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.7607863046159033e-05,
+      "loss": 0.8734,
+      "step": 4821
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.7598096945511738e-05,
+      "loss": 0.6973,
+      "step": 4822
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.7588332082789993e-05,
+      "loss": 0.681,
+      "step": 4823
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.7578568459626915e-05,
+      "loss": 0.8101,
+      "step": 4824
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.7568806077655405e-05,
+      "loss": 0.9693,
+      "step": 4825
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.7559044938508175e-05,
+      "loss": 0.4308,
+      "step": 4826
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.754928504381772e-05,
+      "loss": 0.7298,
+      "step": 4827
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.7539526395216322e-05,
+      "loss": 0.7019,
+      "step": 4828
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.7529768994336065e-05,
+      "loss": 0.5003,
+      "step": 4829
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.75200128428088e-05,
+      "loss": 0.5288,
+      "step": 4830
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.7510257942266213e-05,
+      "loss": 0.8239,
+      "step": 4831
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.7500504294339735e-05,
+      "loss": 0.7154,
+      "step": 4832
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.7490751900660614e-05,
+      "loss": 0.7669,
+      "step": 4833
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.7481000762859882e-05,
+      "loss": 0.7669,
+      "step": 4834
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.7471250882568364e-05,
+      "loss": 0.6966,
+      "step": 4835
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.7461502261416656e-05,
+      "loss": 0.7332,
+      "step": 4836
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.745175490103518e-05,
+      "loss": 0.7455,
+      "step": 4837
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.744200880305411e-05,
+      "loss": 0.7093,
+      "step": 4838
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.7432263969103434e-05,
+      "loss": 0.8167,
+      "step": 4839
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.7422520400812914e-05,
+      "loss": 0.6381,
+      "step": 4840
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.74127780998121e-05,
+      "loss": 0.5218,
+      "step": 4841
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.7403037067730347e-05,
+      "loss": 0.8648,
+      "step": 4842
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.7393297306196786e-05,
+      "loss": 0.8694,
+      "step": 4843
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.7383558816840328e-05,
+      "loss": 0.5384,
+      "step": 4844
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.7373821601289686e-05,
+      "loss": 0.6708,
+      "step": 4845
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.7364085661173347e-05,
+      "loss": 0.6937,
+      "step": 4846
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.7354350998119584e-05,
+      "loss": 0.7831,
+      "step": 4847
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.7344617613756482e-05,
+      "loss": 0.5777,
+      "step": 4848
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.733488550971189e-05,
+      "loss": 0.6599,
+      "step": 4849
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.732515468761343e-05,
+      "loss": 0.9631,
+      "step": 4850
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.7315425149088544e-05,
+      "loss": 0.7663,
+      "step": 4851
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.730569689576442e-05,
+      "loss": 0.7246,
+      "step": 4852
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.729596992926808e-05,
+      "loss": 0.6775,
+      "step": 4853
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.728624425122628e-05,
+      "loss": 0.4698,
+      "step": 4854
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.72765198632656e-05,
+      "loss": 0.4628,
+      "step": 4855
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.7266796767012377e-05,
+      "loss": 0.7669,
+      "step": 4856
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.7257074964092738e-05,
+      "loss": 0.7622,
+      "step": 4857
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.724735445613261e-05,
+      "loss": 0.7047,
+      "step": 4858
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.723763524475769e-05,
+      "loss": 0.7363,
+      "step": 4859
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.722791733159345e-05,
+      "loss": 0.4983,
+      "step": 4860
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.7218200718265176e-05,
+      "loss": 0.8441,
+      "step": 4861
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.7208485406397896e-05,
+      "loss": 0.5354,
+      "step": 4862
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.7198771397616438e-05,
+      "loss": 0.5419,
+      "step": 4863
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.7189058693545427e-05,
+      "loss": 0.6359,
+      "step": 4864
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.717934729580925e-05,
+      "loss": 0.618,
+      "step": 4865
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.7169637206032084e-05,
+      "loss": 0.7077,
+      "step": 4866
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.7159928425837894e-05,
+      "loss": 0.7258,
+      "step": 4867
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.7150220956850388e-05,
+      "loss": 0.6019,
+      "step": 4868
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.7140514800693124e-05,
+      "loss": 0.6049,
+      "step": 4869
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.7130809958989375e-05,
+      "loss": 0.5374,
+      "step": 4870
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.7121106433362223e-05,
+      "loss": 0.5084,
+      "step": 4871
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.7111404225434534e-05,
+      "loss": 0.5392,
+      "step": 4872
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.7101703336828934e-05,
+      "loss": 0.7733,
+      "step": 4873
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.709200376916785e-05,
+      "loss": 0.7517,
+      "step": 4874
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.708230552407349e-05,
+      "loss": 0.5425,
+      "step": 4875
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.7072608603167806e-05,
+      "loss": 0.8682,
+      "step": 4876
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.706291300807256e-05,
+      "loss": 0.5358,
+      "step": 4877
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.7053218740409294e-05,
+      "loss": 0.8528,
+      "step": 4878
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.70435258017993e-05,
+      "loss": 0.8425,
+      "step": 4879
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.7033834193863686e-05,
+      "loss": 0.6666,
+      "step": 4880
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.702414391822331e-05,
+      "loss": 0.8798,
+      "step": 4881
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.701445497649881e-05,
+      "loss": 0.5022,
+      "step": 4882
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.7004767370310614e-05,
+      "loss": 0.7715,
+      "step": 4883
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.6995081101278903e-05,
+      "loss": 0.6228,
+      "step": 4884
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.6985396171023664e-05,
+      "loss": 0.6613,
+      "step": 4885
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.6975712581164643e-05,
+      "loss": 0.7875,
+      "step": 4886
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.6966030333321368e-05,
+      "loss": 0.6687,
+      "step": 4887
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.6956349429113128e-05,
+      "loss": 0.4356,
+      "step": 4888
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.6946669870159e-05,
+      "loss": 0.6803,
+      "step": 4889
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.6936991658077843e-05,
+      "loss": 0.858,
+      "step": 4890
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.6927314794488276e-05,
+      "loss": 0.6432,
+      "step": 4891
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.69176392810087e-05,
+      "loss": 0.635,
+      "step": 4892
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.6907965119257295e-05,
+      "loss": 0.7269,
+      "step": 4893
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.6898292310852e-05,
+      "loss": 0.6749,
+      "step": 4894
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.688862085741053e-05,
+      "loss": 0.8505,
+      "step": 4895
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.68789507605504e-05,
+      "loss": 0.6279,
+      "step": 4896
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.686928202188886e-05,
+      "loss": 0.7545,
+      "step": 4897
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.685961464304296e-05,
+      "loss": 0.6425,
+      "step": 4898
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.6849948625629516e-05,
+      "loss": 0.5578,
+      "step": 4899
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.6840283971265093e-05,
+      "loss": 0.8122,
+      "step": 4900
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.683062068156608e-05,
+      "loss": 0.5744,
+      "step": 4901
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.682095875814858e-05,
+      "loss": 0.7483,
+      "step": 4902
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.6811298202628513e-05,
+      "loss": 0.6272,
+      "step": 4903
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.6801639016621533e-05,
+      "loss": 0.9074,
+      "step": 4904
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.67919812017431e-05,
+      "loss": 0.5173,
+      "step": 4905
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.678232475960841e-05,
+      "loss": 0.7249,
+      "step": 4906
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.6772669691832467e-05,
+      "loss": 0.7696,
+      "step": 4907
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.6763016000030014e-05,
+      "loss": 0.7237,
+      "step": 4908
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.6753363685815577e-05,
+      "loss": 0.6859,
+      "step": 4909
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.6743712750803448e-05,
+      "loss": 0.6648,
+      "step": 4910
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.6734063196607685e-05,
+      "loss": 0.7771,
+      "step": 4911
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.6724415024842138e-05,
+      "loss": 0.7084,
+      "step": 4912
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.67147682371204e-05,
+      "loss": 0.7134,
+      "step": 4913
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.670512283505583e-05,
+      "loss": 0.6336,
+      "step": 4914
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.6695478820261573e-05,
+      "loss": 0.8984,
+      "step": 4915
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.6685836194350535e-05,
+      "loss": 0.7368,
+      "step": 4916
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.667619495893539e-05,
+      "loss": 0.9449,
+      "step": 4917
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.666655511562858e-05,
+      "loss": 0.5743,
+      "step": 4918
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.665691666604232e-05,
+      "loss": 0.537,
+      "step": 4919
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.664727961178857e-05,
+      "loss": 0.6968,
+      "step": 4920
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.6637643954479076e-05,
+      "loss": 0.4623,
+      "step": 4921
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.6628009695725346e-05,
+      "loss": 0.6237,
+      "step": 4922
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.6618376837138665e-05,
+      "loss": 0.597,
+      "step": 4923
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.660874538033006e-05,
+      "loss": 0.647,
+      "step": 4924
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.6599115326910346e-05,
+      "loss": 0.5828,
+      "step": 4925
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.6589486678490087e-05,
+      "loss": 0.7312,
+      "step": 4926
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.6579859436679618e-05,
+      "loss": 0.5732,
+      "step": 4927
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.657023360308904e-05,
+      "loss": 0.7204,
+      "step": 4928
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.656060917932823e-05,
+      "loss": 0.6059,
+      "step": 4929
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.6550986167006802e-05,
+      "loss": 0.7876,
+      "step": 4930
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.6541364567734162e-05,
+      "loss": 0.7521,
+      "step": 4931
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.6531744383119446e-05,
+      "loss": 0.6688,
+      "step": 4932
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.65221256147716e-05,
+      "loss": 0.628,
+      "step": 4933
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.6512508264299298e-05,
+      "loss": 0.6318,
+      "step": 4934
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.6502892333310977e-05,
+      "loss": 0.7764,
+      "step": 4935
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.6493277823414857e-05,
+      "loss": 0.6877,
+      "step": 4936
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.648366473621891e-05,
+      "loss": 0.5569,
+      "step": 4937
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.647405307333085e-05,
+      "loss": 0.6956,
+      "step": 4938
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.64644428363582e-05,
+      "loss": 0.4621,
+      "step": 4939
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.6454834026908204e-05,
+      "loss": 0.9506,
+      "step": 4940
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.6445226646587873e-05,
+      "loss": 0.692,
+      "step": 4941
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.6435620697004e-05,
+      "loss": 0.8277,
+      "step": 4942
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.6426016179763104e-05,
+      "loss": 0.7964,
+      "step": 4943
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.641641309647151e-05,
+      "loss": 0.6764,
+      "step": 4944
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.6406811448735264e-05,
+      "loss": 0.7195,
+      "step": 4945
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.6397211238160192e-05,
+      "loss": 0.5965,
+      "step": 4946
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.6387612466351867e-05,
+      "loss": 0.6551,
+      "step": 4947
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.6378015134915627e-05,
+      "loss": 0.6812,
+      "step": 4948
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.6368419245456575e-05,
+      "loss": 0.7849,
+      "step": 4949
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.635882479957957e-05,
+      "loss": 0.7008,
+      "step": 4950
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.6349231798889232e-05,
+      "loss": 0.7397,
+      "step": 4951
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.6339640244989922e-05,
+      "loss": 0.5922,
+      "step": 4952
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.6330050139485782e-05,
+      "loss": 0.6747,
+      "step": 4953
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.6320461483980688e-05,
+      "loss": 0.7663,
+      "step": 4954
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.6310874280078307e-05,
+      "loss": 0.7218,
+      "step": 4955
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.630128852938203e-05,
+      "loss": 0.6753,
+      "step": 4956
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.629170423349503e-05,
+      "loss": 0.5996,
+      "step": 4957
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.6282121394020217e-05,
+      "loss": 0.6401,
+      "step": 4958
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.6272540012560255e-05,
+      "loss": 0.58,
+      "step": 4959
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.6262960090717595e-05,
+      "loss": 0.6674,
+      "step": 4960
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.625338163009441e-05,
+      "loss": 0.7145,
+      "step": 4961
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.6243804632292646e-05,
+      "loss": 0.7887,
+      "step": 4962
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.623422909891401e-05,
+      "loss": 0.7964,
+      "step": 4963
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.6224655031559944e-05,
+      "loss": 0.4385,
+      "step": 4964
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.6215082431831648e-05,
+      "loss": 0.5144,
+      "step": 4965
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.62055113013301e-05,
+      "loss": 0.6466,
+      "step": 4966
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.619594164165601e-05,
+      "loss": 0.741,
+      "step": 4967
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.6186373454409846e-05,
+      "loss": 0.7115,
+      "step": 4968
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.617680674119184e-05,
+      "loss": 0.6799,
+      "step": 4969
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.6167241503601954e-05,
+      "loss": 0.7435,
+      "step": 4970
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.6157677743239942e-05,
+      "loss": 0.7247,
+      "step": 4971
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.6148115461705266e-05,
+      "loss": 0.5284,
+      "step": 4972
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.6138554660597173e-05,
+      "loss": 0.648,
+      "step": 4973
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.6128995341514645e-05,
+      "loss": 0.8833,
+      "step": 4974
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.6119437506056427e-05,
+      "loss": 0.696,
+      "step": 4975
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.6109881155821012e-05,
+      "loss": 0.7199,
+      "step": 4976
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.6100326292406656e-05,
+      "loss": 0.715,
+      "step": 4977
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.6090772917411335e-05,
+      "loss": 0.6968,
+      "step": 4978
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.6081221032432805e-05,
+      "loss": 0.6723,
+      "step": 4979
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.6071670639068566e-05,
+      "loss": 0.6687,
+      "step": 4980
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.6062121738915853e-05,
+      "loss": 0.6228,
+      "step": 4981
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.605257433357168e-05,
+      "loss": 0.7157,
+      "step": 4982
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.6043028424632794e-05,
+      "loss": 0.6313,
+      "step": 4983
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.6033484013695687e-05,
+      "loss": 0.4696,
+      "step": 4984
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.6023941102356603e-05,
+      "loss": 0.7713,
+      "step": 4985
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.6014399692211544e-05,
+      "loss": 0.6658,
+      "step": 4986
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.6004859784856264e-05,
+      "loss": 0.7412,
+      "step": 4987
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.5995321381886248e-05,
+      "loss": 0.585,
+      "step": 4988
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.598578448489674e-05,
+      "loss": 0.7338,
+      "step": 4989
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.5976249095482733e-05,
+      "loss": 0.5917,
+      "step": 4990
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.5966715215238948e-05,
+      "loss": 0.5429,
+      "step": 4991
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.5957182845759905e-05,
+      "loss": 0.6724,
+      "step": 4992
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.5947651988639816e-05,
+      "loss": 0.797,
+      "step": 4993
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.5938122645472668e-05,
+      "loss": 0.557,
+      "step": 4994
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.5928594817852187e-05,
+      "loss": 0.7323,
+      "step": 4995
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.5919068507371845e-05,
+      "loss": 0.6754,
+      "step": 4996
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.5909543715624854e-05,
+      "loss": 0.8579,
+      "step": 4997
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.5909543715624854e-05,
+      "loss": 0.7792,
+      "step": 4998
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.5900020444204205e-05,
+      "loss": 0.5702,
+      "step": 4999
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.5890498694702594e-05,
+      "loss": 0.6509,
+      "step": 5000
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.5880978468712477e-05,
+      "loss": 0.7789,
+      "step": 5001
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.5871459767826066e-05,
+      "loss": 0.76,
+      "step": 5002
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.586194259363529e-05,
+      "loss": 0.665,
+      "step": 5003
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.585242694773186e-05,
+      "loss": 0.7538,
+      "step": 5004
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.5842912831707217e-05,
+      "loss": 0.6911,
+      "step": 5005
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.5833400247152528e-05,
+      "loss": 0.5991,
+      "step": 5006
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.582388919565872e-05,
+      "loss": 0.4856,
+      "step": 5007
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.581437967881647e-05,
+      "loss": 0.7511,
+      "step": 5008
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.5804871698216166e-05,
+      "loss": 0.6849,
+      "step": 5009
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.5795365255447992e-05,
+      "loss": 0.5642,
+      "step": 5010
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.578586035210184e-05,
+      "loss": 0.6491,
+      "step": 5011
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.5776356989767336e-05,
+      "loss": 0.6383,
+      "step": 5012
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.5766855170033875e-05,
+      "loss": 0.5016,
+      "step": 5013
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.575735489449057e-05,
+      "loss": 0.6729,
+      "step": 5014
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.57478561647263e-05,
+      "loss": 0.803,
+      "step": 5015
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.5738358982329666e-05,
+      "loss": 0.6559,
+      "step": 5016
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.5728863348889022e-05,
+      "loss": 0.7,
+      "step": 5017
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.571936926599245e-05,
+      "loss": 0.8199,
+      "step": 5018
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.5709876735227784e-05,
+      "loss": 0.5746,
+      "step": 5019
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.5700385758182595e-05,
+      "loss": 0.7403,
+      "step": 5020
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.5690896336444195e-05,
+      "loss": 0.5409,
+      "step": 5021
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.5681408471599638e-05,
+      "loss": 0.6381,
+      "step": 5022
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.5671922165235713e-05,
+      "loss": 0.7262,
+      "step": 5023
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.5662437418938944e-05,
+      "loss": 0.6033,
+      "step": 5024
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.5652954234295596e-05,
+      "loss": 0.6198,
+      "step": 5025
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.5643472612891696e-05,
+      "loss": 0.5081,
+      "step": 5026
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.5633992556312976e-05,
+      "loss": 0.6489,
+      "step": 5027
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.5624514066144926e-05,
+      "loss": 0.5775,
+      "step": 5028
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.5615037143972772e-05,
+      "loss": 0.6074,
+      "step": 5029
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.560556179138145e-05,
+      "loss": 0.6826,
+      "step": 5030
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.5596088009955695e-05,
+      "loss": 0.7997,
+      "step": 5031
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.5586615801279922e-05,
+      "loss": 0.6614,
+      "step": 5032
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.5577145166938305e-05,
+      "loss": 0.5698,
+      "step": 5033
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.5567676108514754e-05,
+      "loss": 0.8376,
+      "step": 5034
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.5558208627592906e-05,
+      "loss": 0.6555,
+      "step": 5035
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.554874272575616e-05,
+      "loss": 0.7055,
+      "step": 5036
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.5539278404587627e-05,
+      "loss": 0.7321,
+      "step": 5037
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.5529815665670155e-05,
+      "loss": 0.6369,
+      "step": 5038
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.5520354510586333e-05,
+      "loss": 0.6472,
+      "step": 5039
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.5510894940918487e-05,
+      "loss": 0.7765,
+      "step": 5040
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.5501436958248667e-05,
+      "loss": 0.7062,
+      "step": 5041
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.5491980564158677e-05,
+      "loss": 0.6451,
+      "step": 5042
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.548252576023005e-05,
+      "loss": 0.566,
+      "step": 5043
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.547307254804403e-05,
+      "loss": 0.7403,
+      "step": 5044
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.5463620929181628e-05,
+      "loss": 0.6687,
+      "step": 5045
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.545417090522355e-05,
+      "loss": 0.6805,
+      "step": 5046
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.5444722477750284e-05,
+      "loss": 0.5585,
+      "step": 5047
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.5435275648342012e-05,
+      "loss": 0.4389,
+      "step": 5048
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.5425830418578674e-05,
+      "loss": 0.8079,
+      "step": 5049
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.541638679003991e-05,
+      "loss": 0.5892,
+      "step": 5050
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.5406944764305117e-05,
+      "loss": 0.6662,
+      "step": 5051
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.539750434295343e-05,
+      "loss": 0.7268,
+      "step": 5052
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.53880655275637e-05,
+      "loss": 0.611,
+      "step": 5053
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.5378628319714512e-05,
+      "loss": 0.6953,
+      "step": 5054
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.5369192720984196e-05,
+      "loss": 0.5835,
+      "step": 5055
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.5359758732950786e-05,
+      "loss": 0.3708,
+      "step": 5056
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.5350326357192058e-05,
+      "loss": 0.8775,
+      "step": 5057
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.5340895595285543e-05,
+      "loss": 0.7789,
+      "step": 5058
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.5331466448808465e-05,
+      "loss": 0.6913,
+      "step": 5059
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.53220389193378e-05,
+      "loss": 0.8666,
+      "step": 5060
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.531261300845026e-05,
+      "loss": 0.8224,
+      "step": 5061
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.530318871772224e-05,
+      "loss": 0.5821,
+      "step": 5062
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.529376604872994e-05,
+      "loss": 0.6799,
+      "step": 5063
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.528434500304922e-05,
+      "loss": 0.729,
+      "step": 5064
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.5274925582255703e-05,
+      "loss": 0.7239,
+      "step": 5065
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.5265507787924732e-05,
+      "loss": 0.8826,
+      "step": 5066
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.5256091621631385e-05,
+      "loss": 0.5732,
+      "step": 5067
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.524667708495044e-05,
+      "loss": 0.6163,
+      "step": 5068
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.5237264179456456e-05,
+      "loss": 0.6201,
+      "step": 5069
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.5227852906723663e-05,
+      "loss": 0.6836,
+      "step": 5070
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.5218443268326052e-05,
+      "loss": 0.6445,
+      "step": 5071
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.5209035265837327e-05,
+      "loss": 0.7328,
+      "step": 5072
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.519962890083092e-05,
+      "loss": 0.6763,
+      "step": 5073
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.5190224174879997e-05,
+      "loss": 0.7072,
+      "step": 5074
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.518082108955745e-05,
+      "loss": 0.5679,
+      "step": 5075
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.5171419646435874e-05,
+      "loss": 0.5697,
+      "step": 5076
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.5162019847087617e-05,
+      "loss": 0.419,
+      "step": 5077
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.5152621693084735e-05,
+      "loss": 0.7439,
+      "step": 5078
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.5143225185999022e-05,
+      "loss": 0.7244,
+      "step": 5079
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.5133830327401988e-05,
+      "loss": 0.8464,
+      "step": 5080
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.512443711886487e-05,
+      "loss": 0.8493,
+      "step": 5081
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.5115045561958618e-05,
+      "loss": 0.8069,
+      "step": 5082
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.5105655658253925e-05,
+      "loss": 0.7972,
+      "step": 5083
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.5096267409321194e-05,
+      "loss": 0.8753,
+      "step": 5084
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.5086880816730559e-05,
+      "loss": 0.5878,
+      "step": 5085
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.5077495882051873e-05,
+      "loss": 0.7186,
+      "step": 5086
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.5068112606854717e-05,
+      "loss": 0.3954,
+      "step": 5087
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.5058730992708375e-05,
+      "loss": 0.5408,
+      "step": 5088
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.5049351041181876e-05,
+      "loss": 0.7752,
+      "step": 5089
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.5039972753843965e-05,
+      "loss": 0.7062,
+      "step": 5090
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.5030596132263106e-05,
+      "loss": 0.7073,
+      "step": 5091
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.5021221178007482e-05,
+      "loss": 0.6204,
+      "step": 5092
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.5011847892645007e-05,
+      "loss": 0.8152,
+      "step": 5093
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.5002476277743288e-05,
+      "loss": 0.646,
+      "step": 5094
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.4993106334869702e-05,
+      "loss": 0.6543,
+      "step": 5095
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.4983738065591302e-05,
+      "loss": 0.5914,
+      "step": 5096
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.497437147147488e-05,
+      "loss": 0.8157,
+      "step": 5097
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.4965006554086941e-05,
+      "loss": 0.7243,
+      "step": 5098
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.4955643314993722e-05,
+      "loss": 0.6883,
+      "step": 5099
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.4946281755761152e-05,
+      "loss": 0.655,
+      "step": 5100
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.4936921877954923e-05,
+      "loss": 0.672,
+      "step": 5101
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.4927563683140405e-05,
+      "loss": 0.8,
+      "step": 5102
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.4918207172882709e-05,
+      "loss": 0.5944,
+      "step": 5103
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.4908852348746649e-05,
+      "loss": 0.7095,
+      "step": 5104
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.489949921229677e-05,
+      "loss": 0.7095,
+      "step": 5105
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.4890147765097331e-05,
+      "loss": 0.6444,
+      "step": 5106
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.4880798008712316e-05,
+      "loss": 0.623,
+      "step": 5107
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.48714499447054e-05,
+      "loss": 0.662,
+      "step": 5108
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.4862103574640002e-05,
+      "loss": 0.6726,
+      "step": 5109
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.4852758900079242e-05,
+      "loss": 0.701,
+      "step": 5110
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.4843415922585974e-05,
+      "loss": 0.9642,
+      "step": 5111
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.483407464372275e-05,
+      "loss": 0.5773,
+      "step": 5112
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.4824735065051856e-05,
+      "loss": 0.6308,
+      "step": 5113
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.4815397188135266e-05,
+      "loss": 0.6541,
+      "step": 5114
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.4806061014534694e-05,
+      "loss": 0.6802,
+      "step": 5115
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.4796726545811554e-05,
+      "loss": 0.7511,
+      "step": 5116
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.4787393783526993e-05,
+      "loss": 0.7295,
+      "step": 5117
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.477806272924186e-05,
+      "loss": 0.5661,
+      "step": 5118
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.4768733384516717e-05,
+      "loss": 0.6654,
+      "step": 5119
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.4759405750911842e-05,
+      "loss": 0.5589,
+      "step": 5120
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.4750079829987223e-05,
+      "loss": 0.7318,
+      "step": 5121
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.4740755623302577e-05,
+      "loss": 0.6067,
+      "step": 5122
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.4731433132417316e-05,
+      "loss": 0.591,
+      "step": 5123
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.4722112358890577e-05,
+      "loss": 0.5814,
+      "step": 5124
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.4712793304281214e-05,
+      "loss": 0.6135,
+      "step": 5125
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.4703475970147766e-05,
+      "loss": 0.4763,
+      "step": 5126
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.4694160358048509e-05,
+      "loss": 0.6038,
+      "step": 5127
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.4684846469541433e-05,
+      "loss": 0.7226,
+      "step": 5128
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.467553430618423e-05,
+      "loss": 0.5822,
+      "step": 5129
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.4666223869534301e-05,
+      "loss": 0.7077,
+      "step": 5130
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.4656915161148772e-05,
+      "loss": 0.689,
+      "step": 5131
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.4647608182584449e-05,
+      "loss": 0.6106,
+      "step": 5132
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.4638302935397902e-05,
+      "loss": 0.5537,
+      "step": 5133
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.4628999421145356e-05,
+      "loss": 0.717,
+      "step": 5134
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.4619697641382779e-05,
+      "loss": 0.5608,
+      "step": 5135
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.4610397597665838e-05,
+      "loss": 0.768,
+      "step": 5136
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.4601099291549908e-05,
+      "loss": 0.4785,
+      "step": 5137
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.459180272459009e-05,
+      "loss": 0.7722,
+      "step": 5138
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.4582507898341169e-05,
+      "loss": 0.5624,
+      "step": 5139
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.4573214814357653e-05,
+      "loss": 0.5,
+      "step": 5140
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.4563923474193758e-05,
+      "loss": 0.6282,
+      "step": 5141
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.4554633879403407e-05,
+      "loss": 0.7608,
+      "step": 5142
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.4545346031540231e-05,
+      "loss": 0.6742,
+      "step": 5143
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.4536059932157567e-05,
+      "loss": 0.8205,
+      "step": 5144
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.4526775582808466e-05,
+      "loss": 0.5233,
+      "step": 5145
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.4517492985045678e-05,
+      "loss": 0.6919,
+      "step": 5146
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.4508212140421668e-05,
+      "loss": 0.6123,
+      "step": 5147
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.4498933050488586e-05,
+      "loss": 0.5193,
+      "step": 5148
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.4489655716798345e-05,
+      "loss": 0.4205,
+      "step": 5149
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.4480380140902489e-05,
+      "loss": 0.5311,
+      "step": 5150
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.4471106324352318e-05,
+      "loss": 0.744,
+      "step": 5151
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.4461834268698823e-05,
+      "loss": 0.5362,
+      "step": 5152
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.4452563975492695e-05,
+      "loss": 0.6531,
+      "step": 5153
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.4443295446284349e-05,
+      "loss": 0.7624,
+      "step": 5154
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.44340286826239e-05,
+      "loss": 0.6718,
+      "step": 5155
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.442476368606114e-05,
+      "loss": 0.6781,
+      "step": 5156
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.4415500458145597e-05,
+      "loss": 0.6442,
+      "step": 5157
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.4406239000426486e-05,
+      "loss": 0.7287,
+      "step": 5158
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.439697931445273e-05,
+      "loss": 0.7008,
+      "step": 5159
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.4387721401772974e-05,
+      "loss": 0.6092,
+      "step": 5160
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.437846526393555e-05,
+      "loss": 0.6402,
+      "step": 5161
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.4369210902488476e-05,
+      "loss": 0.6039,
+      "step": 5162
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.4359958318979499e-05,
+      "loss": 0.6328,
+      "step": 5163
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.4350707514956049e-05,
+      "loss": 0.6612,
+      "step": 5164
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.4341458491965292e-05,
+      "loss": 0.7079,
+      "step": 5165
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.4332211251554065e-05,
+      "loss": 0.8094,
+      "step": 5166
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.4322965795268922e-05,
+      "loss": 0.6505,
+      "step": 5167
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.4313722124656092e-05,
+      "loss": 0.6419,
+      "step": 5168
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.4304480241261528e-05,
+      "loss": 0.6887,
+      "step": 5169
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.4295240146630901e-05,
+      "loss": 0.6526,
+      "step": 5170
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.4286001842309554e-05,
+      "loss": 0.6405,
+      "step": 5171
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.4276765329842537e-05,
+      "loss": 0.6234,
+      "step": 5172
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.4267530610774615e-05,
+      "loss": 0.5837,
+      "step": 5173
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.4258297686650224e-05,
+      "loss": 0.6872,
+      "step": 5174
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.4249066559013516e-05,
+      "loss": 0.562,
+      "step": 5175
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.4239837229408362e-05,
+      "loss": 0.6046,
+      "step": 5176
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.4230609699378306e-05,
+      "loss": 0.85,
+      "step": 5177
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.4221383970466596e-05,
+      "loss": 0.9017,
+      "step": 5178
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.4212160044216197e-05,
+      "loss": 0.8649,
+      "step": 5179
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.4202937922169724e-05,
+      "loss": 0.6473,
+      "step": 5180
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.4193717605869553e-05,
+      "loss": 0.699,
+      "step": 5181
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.418449909685772e-05,
+      "loss": 0.6079,
+      "step": 5182
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.4175282396675968e-05,
+      "loss": 0.877,
+      "step": 5183
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.4166067506865738e-05,
+      "loss": 0.6444,
+      "step": 5184
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.415685442896817e-05,
+      "loss": 0.5953,
+      "step": 5185
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.4147643164524077e-05,
+      "loss": 0.7541,
+      "step": 5186
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.4138433715074018e-05,
+      "loss": 0.6155,
+      "step": 5187
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.4129226082158206e-05,
+      "loss": 0.7541,
+      "step": 5188
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.412002026731657e-05,
+      "loss": 0.7334,
+      "step": 5189
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.4110816272088724e-05,
+      "loss": 0.6239,
+      "step": 5190
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.4101614098013983e-05,
+      "loss": 0.5463,
+      "step": 5191
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.409241374663136e-05,
+      "loss": 0.7347,
+      "step": 5192
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.4083215219479562e-05,
+      "loss": 0.6213,
+      "step": 5193
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.407401851809699e-05,
+      "loss": 0.5777,
+      "step": 5194
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.4064823644021736e-05,
+      "loss": 0.7259,
+      "step": 5195
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.4055630598791592e-05,
+      "loss": 0.5024,
+      "step": 5196
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.4046439383944038e-05,
+      "loss": 0.5816,
+      "step": 5197
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.4037250001016255e-05,
+      "loss": 0.4976,
+      "step": 5198
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.4028062451545116e-05,
+      "loss": 0.4928,
+      "step": 5199
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.4018876737067178e-05,
+      "loss": 0.9009,
+      "step": 5200
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.4009692859118705e-05,
+      "loss": 0.6804,
+      "step": 5201
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.4000510819235652e-05,
+      "loss": 0.5359,
+      "step": 5202
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.3991330618953652e-05,
+      "loss": 0.4609,
+      "step": 5203
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.3982152259808046e-05,
+      "loss": 0.6108,
+      "step": 5204
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.3972975743333858e-05,
+      "loss": 0.7479,
+      "step": 5205
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.3963801071065813e-05,
+      "loss": 0.7533,
+      "step": 5206
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.3954628244538307e-05,
+      "loss": 0.5603,
+      "step": 5207
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.3945457265285476e-05,
+      "loss": 0.6539,
+      "step": 5208
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.393628813484108e-05,
+      "loss": 0.7211,
+      "step": 5209
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.3927120854738612e-05,
+      "loss": 0.5488,
+      "step": 5210
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.3917955426511253e-05,
+      "loss": 0.5532,
+      "step": 5211
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.3908791851691856e-05,
+      "loss": 0.5256,
+      "step": 5212
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.389963013181299e-05,
+      "loss": 0.848,
+      "step": 5213
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.3890470268406907e-05,
+      "loss": 0.7242,
+      "step": 5214
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.3881312263005519e-05,
+      "loss": 0.6644,
+      "step": 5215
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.3872156117140456e-05,
+      "loss": 0.6287,
+      "step": 5216
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.3863001832343033e-05,
+      "loss": 0.8508,
+      "step": 5217
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.3853849410144243e-05,
+      "loss": 0.6739,
+      "step": 5218
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.3844698852074795e-05,
+      "loss": 0.6497,
+      "step": 5219
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.3835550159665065e-05,
+      "loss": 0.7804,
+      "step": 5220
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.3826403334445101e-05,
+      "loss": 0.9533,
+      "step": 5221
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.3817258377944664e-05,
+      "loss": 0.3892,
+      "step": 5222
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.380811529169319e-05,
+      "loss": 0.7077,
+      "step": 5223
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.3798974077219823e-05,
+      "loss": 0.6756,
+      "step": 5224
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.3789834736053372e-05,
+      "loss": 0.466,
+      "step": 5225
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.3780697269722348e-05,
+      "loss": 0.5097,
+      "step": 5226
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.3771561679754913e-05,
+      "loss": 0.8027,
+      "step": 5227
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.3762427967678953e-05,
+      "loss": 0.6843,
+      "step": 5228
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.3753296135022039e-05,
+      "loss": 0.7488,
+      "step": 5229
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.3744166183311413e-05,
+      "loss": 0.7486,
+      "step": 5230
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.3735038114074003e-05,
+      "loss": 0.6742,
+      "step": 5231
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.3725911928836438e-05,
+      "loss": 0.7104,
+      "step": 5232
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.3716787629125e-05,
+      "loss": 0.715,
+      "step": 5233
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.3707665216465677e-05,
+      "loss": 0.6833,
+      "step": 5234
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.3698544692384155e-05,
+      "loss": 0.7914,
+      "step": 5235
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.3689426058405782e-05,
+      "loss": 0.6247,
+      "step": 5236
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.3680309316055595e-05,
+      "loss": 0.4831,
+      "step": 5237
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.3671194466858334e-05,
+      "loss": 0.8362,
+      "step": 5238
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.3662081512338365e-05,
+      "loss": 0.8508,
+      "step": 5239
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.365297045401981e-05,
+      "loss": 0.5095,
+      "step": 5240
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.3643861293426433e-05,
+      "loss": 0.6468,
+      "step": 5241
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.3634754032081687e-05,
+      "loss": 0.6543,
+      "step": 5242
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.3625648671508712e-05,
+      "loss": 0.7486,
+      "step": 5243
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.361654521323033e-05,
+      "loss": 0.5569,
+      "step": 5244
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.360744365876902e-05,
+      "loss": 0.6213,
+      "step": 5245
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.3598344009646987e-05,
+      "loss": 0.9584,
+      "step": 5246
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.3589246267386092e-05,
+      "loss": 0.7271,
+      "step": 5247
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.3580150433507873e-05,
+      "loss": 0.7042,
+      "step": 5248
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.3571056509533565e-05,
+      "loss": 0.6501,
+      "step": 5249
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.356196449698406e-05,
+      "loss": 0.4399,
+      "step": 5250
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.3552874397379958e-05,
+      "loss": 0.4505,
+      "step": 5251
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.3543786212241519e-05,
+      "loss": 0.7431,
+      "step": 5252
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.353469994308869e-05,
+      "loss": 0.741,
+      "step": 5253
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.3525615591441098e-05,
+      "loss": 0.6688,
+      "step": 5254
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.3516533158818051e-05,
+      "loss": 0.7148,
+      "step": 5255
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.350745264673853e-05,
+      "loss": 0.4687,
+      "step": 5256
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.3498374056721197e-05,
+      "loss": 0.82,
+      "step": 5257
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.34892973902844e-05,
+      "loss": 0.5031,
+      "step": 5258
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.348022264894615e-05,
+      "loss": 0.507,
+      "step": 5259
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.3471149834224154e-05,
+      "loss": 0.6097,
+      "step": 5260
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.3462078947635781e-05,
+      "loss": 0.5974,
+      "step": 5261
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.345300999069809e-05,
+      "loss": 0.683,
+      "step": 5262
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.3443942964927806e-05,
+      "loss": 0.7012,
+      "step": 5263
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.3434877871841339e-05,
+      "loss": 0.5781,
+      "step": 5264
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.3425814712954777e-05,
+      "loss": 0.5793,
+      "step": 5265
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.341675348978388e-05,
+      "loss": 0.5159,
+      "step": 5266
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.3407694203844078e-05,
+      "loss": 0.4805,
+      "step": 5267
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.3398636856650492e-05,
+      "loss": 0.5154,
+      "step": 5268
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.338958144971791e-05,
+      "loss": 0.7495,
+      "step": 5269
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.3380527984560792e-05,
+      "loss": 0.7173,
+      "step": 5270
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.3371476462693272e-05,
+      "loss": 0.5246,
+      "step": 5271
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.3362426885629195e-05,
+      "loss": 0.8357,
+      "step": 5272
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.335337925488202e-05,
+      "loss": 0.5067,
+      "step": 5273
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.3344333571964917e-05,
+      "loss": 0.8346,
+      "step": 5274
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.333528983839073e-05,
+      "loss": 0.8266,
+      "step": 5275
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.332624805567197e-05,
+      "loss": 0.6384,
+      "step": 5276
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.331720822532081e-05,
+      "loss": 0.8616,
+      "step": 5277
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.3308170348849142e-05,
+      "loss": 0.4657,
+      "step": 5278
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.3299134427768465e-05,
+      "loss": 0.7681,
+      "step": 5279
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.3290100463589999e-05,
+      "loss": 0.5896,
+      "step": 5280
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.328106845782462e-05,
+      "loss": 0.64,
+      "step": 5281
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.3272038411982873e-05,
+      "loss": 0.7654,
+      "step": 5282
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.3263010327574992e-05,
+      "loss": 0.6499,
+      "step": 5283
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.325398420611088e-05,
+      "loss": 0.3956,
+      "step": 5284
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.3244960049100086e-05,
+      "loss": 0.6525,
+      "step": 5285
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.3235937858051847e-05,
+      "loss": 0.8354,
+      "step": 5286
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.3226917634475073e-05,
+      "loss": 0.6151,
+      "step": 5287
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.3217899379878358e-05,
+      "loss": 0.6084,
+      "step": 5288
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.3208883095769945e-05,
+      "loss": 0.7076,
+      "step": 5289
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.3199868783657766e-05,
+      "loss": 0.645,
+      "step": 5290
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.3190856445049394e-05,
+      "loss": 0.8336,
+      "step": 5291
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.3181846081452093e-05,
+      "loss": 0.5917,
+      "step": 5292
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.3172837694372797e-05,
+      "loss": 0.7315,
+      "step": 5293
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.3163831285318118e-05,
+      "loss": 0.6057,
+      "step": 5294
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.3154826855794317e-05,
+      "loss": 0.5441,
+      "step": 5295
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.3145824407307342e-05,
+      "loss": 0.7886,
+      "step": 5296
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.3136823941362784e-05,
+      "loss": 0.5499,
+      "step": 5297
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.3127825459465914e-05,
+      "loss": 0.7249,
+      "step": 5298
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.3118828963121705e-05,
+      "loss": 0.5919,
+      "step": 5299
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.3109834453834748e-05,
+      "loss": 0.8694,
+      "step": 5300
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.3100841933109332e-05,
+      "loss": 0.495,
+      "step": 5301
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.3091851402449407e-05,
+      "loss": 0.6939,
+      "step": 5302
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.3082862863358571e-05,
+      "loss": 0.7404,
+      "step": 5303
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.3073876317340106e-05,
+      "loss": 0.7087,
+      "step": 5304
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.306489176589698e-05,
+      "loss": 0.6787,
+      "step": 5305
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.3055909210531797e-05,
+      "loss": 0.6413,
+      "step": 5306
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.3046928652746832e-05,
+      "loss": 0.7534,
+      "step": 5307
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.303795009404405e-05,
+      "loss": 0.6802,
+      "step": 5308
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.3028973535925026e-05,
+      "loss": 0.7036,
+      "step": 5309
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.3019998979891073e-05,
+      "loss": 0.6026,
+      "step": 5310
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.3011026427443124e-05,
+      "loss": 0.8617,
+      "step": 5311
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.3002055880081778e-05,
+      "loss": 0.7149,
+      "step": 5312
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.2993087339307316e-05,
+      "loss": 0.9209,
+      "step": 5313
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.2984120806619673e-05,
+      "loss": 0.5541,
+      "step": 5314
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.2975156283518447e-05,
+      "loss": 0.5218,
+      "step": 5315
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.2966193771502905e-05,
+      "loss": 0.6798,
+      "step": 5316
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.2957233272071977e-05,
+      "loss": 0.4338,
+      "step": 5317
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.2948274786724252e-05,
+      "loss": 0.5896,
+      "step": 5318
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.2939318316957983e-05,
+      "loss": 0.5848,
+      "step": 5319
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.2930363864271092e-05,
+      "loss": 0.6265,
+      "step": 5320
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.2921411430161157e-05,
+      "loss": 0.5533,
+      "step": 5321
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.291246101612542e-05,
+      "loss": 0.7096,
+      "step": 5322
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.2903512623660793e-05,
+      "loss": 0.5449,
+      "step": 5323
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.2894566254263835e-05,
+      "loss": 0.6947,
+      "step": 5324
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.2885621909430778e-05,
+      "loss": 0.5889,
+      "step": 5325
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.2876679590657515e-05,
+      "loss": 0.7622,
+      "step": 5326
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.2867739299439588e-05,
+      "loss": 0.7268,
+      "step": 5327
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.2858801037272217e-05,
+      "loss": 0.6482,
+      "step": 5328
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.2849864805650275e-05,
+      "loss": 0.5974,
+      "step": 5329
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.2840930606068289e-05,
+      "loss": 0.6087,
+      "step": 5330
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.2831998440020456e-05,
+      "loss": 0.7589,
+      "step": 5331
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.2823068309000633e-05,
+      "loss": 0.6601,
+      "step": 5332
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.2814140214502326e-05,
+      "loss": 0.5314,
+      "step": 5333
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.2805214158018712e-05,
+      "loss": 0.6693,
+      "step": 5334
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.279629014104262e-05,
+      "loss": 0.4356,
+      "step": 5335
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.278736816506654e-05,
+      "loss": 0.935,
+      "step": 5336
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.2778448231582624e-05,
+      "loss": 0.6755,
+      "step": 5337
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.2769530342082674e-05,
+      "loss": 0.8072,
+      "step": 5338
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.2760614498058166e-05,
+      "loss": 0.7667,
+      "step": 5339
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.2751700701000207e-05,
+      "loss": 0.6537,
+      "step": 5340
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.2742788952399587e-05,
+      "loss": 0.7016,
+      "step": 5341
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.2733879253746759e-05,
+      "loss": 0.5686,
+      "step": 5342
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.2724971606531794e-05,
+      "loss": 0.6328,
+      "step": 5343
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.271606601224446e-05,
+      "loss": 0.6596,
+      "step": 5344
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.2707162472374156e-05,
+      "loss": 0.7689,
+      "step": 5345
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.2698260988409947e-05,
+      "loss": 0.6845,
+      "step": 5346
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.2689361561840568e-05,
+      "loss": 0.7177,
+      "step": 5347
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.2680464194154402e-05,
+      "loss": 0.5632,
+      "step": 5348
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.2671568886839457e-05,
+      "loss": 0.6622,
+      "step": 5349
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.2662675641383437e-05,
+      "loss": 0.7429,
+      "step": 5350
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.2653784459273682e-05,
+      "loss": 0.7067,
+      "step": 5351
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.2644895341997187e-05,
+      "loss": 0.6493,
+      "step": 5352
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.2636008291040618e-05,
+      "loss": 0.5804,
+      "step": 5353
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.2627123307890284e-05,
+      "loss": 0.6201,
+      "step": 5354
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.2618240394032133e-05,
+      "loss": 0.5604,
+      "step": 5355
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.2609359550951782e-05,
+      "loss": 0.639,
+      "step": 5356
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.2600480780134501e-05,
+      "loss": 0.6812,
+      "step": 5357
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.2591604083065229e-05,
+      "loss": 0.7553,
+      "step": 5358
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.2582729461228527e-05,
+      "loss": 0.7775,
+      "step": 5359
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.257385691610864e-05,
+      "loss": 0.4282,
+      "step": 5360
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.2564986449189432e-05,
+      "loss": 0.4901,
+      "step": 5361
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.2556118061954439e-05,
+      "loss": 0.6033,
+      "step": 5362
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.2547251755886846e-05,
+      "loss": 0.7177,
+      "step": 5363
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.2538387532469506e-05,
+      "loss": 0.6886,
+      "step": 5364
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.2529525393184904e-05,
+      "loss": 0.6575,
+      "step": 5365
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.2520665339515186e-05,
+      "loss": 0.7258,
+      "step": 5366
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.251180737294213e-05,
+      "loss": 0.6986,
+      "step": 5367
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.2502951494947173e-05,
+      "loss": 0.485,
+      "step": 5368
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.249409770701144e-05,
+      "loss": 0.6127,
+      "step": 5369
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.2485246010615658e-05,
+      "loss": 0.8784,
+      "step": 5370
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.2476396407240223e-05,
+      "loss": 0.6769,
+      "step": 5371
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.2467548898365192e-05,
+      "loss": 0.699,
+      "step": 5372
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.2458703485470227e-05,
+      "loss": 0.6923,
+      "step": 5373
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.2449860170034705e-05,
+      "loss": 0.6659,
+      "step": 5374
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.2441018953537609e-05,
+      "loss": 0.6422,
+      "step": 5375
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.243217983745758e-05,
+      "loss": 0.6537,
+      "step": 5376
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.2423342823272912e-05,
+      "loss": 0.6058,
+      "step": 5377
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.241450791246155e-05,
+      "loss": 0.6837,
+      "step": 5378
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.2405675106501058e-05,
+      "loss": 0.5945,
+      "step": 5379
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.2396844406868697e-05,
+      "loss": 0.4389,
+      "step": 5380
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.238801581504134e-05,
+      "loss": 0.7506,
+      "step": 5381
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.2379189332495523e-05,
+      "loss": 0.6402,
+      "step": 5382
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.2370364960707417e-05,
+      "loss": 0.7176,
+      "step": 5383
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.2361542701152854e-05,
+      "loss": 0.5568,
+      "step": 5384
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.2352722555307304e-05,
+      "loss": 0.6962,
+      "step": 5385
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.2343904524645883e-05,
+      "loss": 0.5741,
+      "step": 5386
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.2335088610643355e-05,
+      "loss": 0.527,
+      "step": 5387
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.232627481477414e-05,
+      "loss": 0.6551,
+      "step": 5388
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.2317463138512281e-05,
+      "loss": 0.7759,
+      "step": 5389
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.230865358333149e-05,
+      "loss": 0.5321,
+      "step": 5390
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.2299846150705111e-05,
+      "loss": 0.6946,
+      "step": 5391
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.2291040842106134e-05,
+      "loss": 0.6463,
+      "step": 5392
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.2282237659007198e-05,
+      "loss": 0.8317,
+      "step": 5393
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.2273436602880586e-05,
+      "loss": 0.8193,
+      "step": 5394
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.2264637675198223e-05,
+      "loss": 0.5479,
+      "step": 5395
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.2255840877431679e-05,
+      "loss": 0.6179,
+      "step": 5396
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.2247046211052166e-05,
+      "loss": 0.7521,
+      "step": 5397
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.2238253677530542e-05,
+      "loss": 0.7217,
+      "step": 5398
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.2229463278337308e-05,
+      "loss": 0.6349,
+      "step": 5399
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.2220675014942606e-05,
+      "loss": 0.7284,
+      "step": 5400
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.2211888888816225e-05,
+      "loss": 0.6738,
+      "step": 5401
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.2203104901427592e-05,
+      "loss": 0.5773,
+      "step": 5402
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.2194323054245776e-05,
+      "loss": 0.4647,
+      "step": 5403
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.2185543348739495e-05,
+      "loss": 0.7269,
+      "step": 5404
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.2176765786377092e-05,
+      "loss": 0.647,
+      "step": 5405
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.2167990368626594e-05,
+      "loss": 0.5582,
+      "step": 5406
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.2159217096955608e-05,
+      "loss": 0.6065,
+      "step": 5407
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.2150445972831422e-05,
+      "loss": 0.6139,
+      "step": 5408
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.2141676997720958e-05,
+      "loss": 0.4724,
+      "step": 5409
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.2132910173090776e-05,
+      "loss": 0.6641,
+      "step": 5410
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.2124145500407066e-05,
+      "loss": 0.7794,
+      "step": 5411
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.21153829811357e-05,
+      "loss": 0.6183,
+      "step": 5412
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.2106622616742128e-05,
+      "loss": 0.6796,
+      "step": 5413
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.209786440869148e-05,
+      "loss": 0.8005,
+      "step": 5414
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.2089108358448517e-05,
+      "loss": 0.5496,
+      "step": 5415
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.2080354467477629e-05,
+      "loss": 0.7138,
+      "step": 5416
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.2071602737242868e-05,
+      "loss": 0.5225,
+      "step": 5417
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.2062853169207915e-05,
+      "loss": 0.6173,
+      "step": 5418
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.2054105764836066e-05,
+      "loss": 0.6874,
+      "step": 5419
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.2045360525590282e-05,
+      "loss": 0.5713,
+      "step": 5420
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.2036617452933153e-05,
+      "loss": 0.5955,
+      "step": 5421
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.2027876548326897e-05,
+      "loss": 0.4923,
+      "step": 5422
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.2019137813233403e-05,
+      "loss": 0.6221,
+      "step": 5423
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.2010401249114167e-05,
+      "loss": 0.5354,
+      "step": 5424
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.2001666857430318e-05,
+      "loss": 0.585,
+      "step": 5425
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.1992934639642635e-05,
+      "loss": 0.6667,
+      "step": 5426
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.1984204597211529e-05,
+      "loss": 0.7757,
+      "step": 5427
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.1975476731597063e-05,
+      "loss": 0.6365,
+      "step": 5428
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.1966751044258911e-05,
+      "loss": 0.5437,
+      "step": 5429
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.1958027536656408e-05,
+      "loss": 0.8334,
+      "step": 5430
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.194930621024849e-05,
+      "loss": 0.633,
+      "step": 5431
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.1940587066493745e-05,
+      "loss": 0.6899,
+      "step": 5432
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.1931870106850424e-05,
+      "loss": 0.7021,
+      "step": 5433
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.1923155332776373e-05,
+      "loss": 0.6177,
+      "step": 5434
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.1914442745729093e-05,
+      "loss": 0.6219,
+      "step": 5435
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.1905732347165718e-05,
+      "loss": 0.7527,
+      "step": 5436
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.1897024138543e-05,
+      "loss": 0.6921,
+      "step": 5437
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.188831812131733e-05,
+      "loss": 0.62,
+      "step": 5438
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.1879614296944761e-05,
+      "loss": 0.5381,
+      "step": 5439
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.1870912666880945e-05,
+      "loss": 0.7187,
+      "step": 5440
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.1862213232581185e-05,
+      "loss": 0.6512,
+      "step": 5441
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.1853515995500417e-05,
+      "loss": 0.6476,
+      "step": 5442
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.1844820957093175e-05,
+      "loss": 0.5345,
+      "step": 5443
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.183612811881368e-05,
+      "loss": 0.4167,
+      "step": 5444
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.1827437482115759e-05,
+      "loss": 0.7785,
+      "step": 5445
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.181874904845286e-05,
+      "loss": 0.5518,
+      "step": 5446
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.1810062819278078e-05,
+      "loss": 0.6439,
+      "step": 5447
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.1801378796044135e-05,
+      "loss": 0.6962,
+      "step": 5448
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.1792696980203383e-05,
+      "loss": 0.5916,
+      "step": 5449
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.1784017373207803e-05,
+      "loss": 0.6508,
+      "step": 5450
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.1775339976509017e-05,
+      "loss": 0.5614,
+      "step": 5451
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.1766664791558263e-05,
+      "loss": 0.3337,
+      "step": 5452
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.1757991819806415e-05,
+      "loss": 0.8533,
+      "step": 5453
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.174932106270398e-05,
+      "loss": 0.7469,
+      "step": 5454
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.1740652521701092e-05,
+      "loss": 0.672,
+      "step": 5455
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.1731986198247513e-05,
+      "loss": 0.845,
+      "step": 5456
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.1723322093792638e-05,
+      "loss": 0.8072,
+      "step": 5457
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.1714660209785486e-05,
+      "loss": 0.5631,
+      "step": 5458
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.1706000547674708e-05,
+      "loss": 0.6683,
+      "step": 5459
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.1697343108908584e-05,
+      "loss": 0.7033,
+      "step": 5460
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.1688687894935018e-05,
+      "loss": 0.7012,
+      "step": 5461
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.1680034907201546e-05,
+      "loss": 0.8653,
+      "step": 5462
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.1671384147155331e-05,
+      "loss": 0.5454,
+      "step": 5463
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.166273561624316e-05,
+      "loss": 0.5962,
+      "step": 5464
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.1654089315911452e-05,
+      "loss": 0.6128,
+      "step": 5465
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.1645445247606252e-05,
+      "loss": 0.6624,
+      "step": 5466
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.1636803412773226e-05,
+      "loss": 0.6183,
+      "step": 5467
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.1628163812857674e-05,
+      "loss": 0.7114,
+      "step": 5468
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.1619526449304521e-05,
+      "loss": 0.6464,
+      "step": 5469
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.1610891323558303e-05,
+      "loss": 0.6809,
+      "step": 5470
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.1602258437063227e-05,
+      "loss": 0.5452,
+      "step": 5471
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.1593627791263062e-05,
+      "loss": 0.5456,
+      "step": 5472
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.1584999387601242e-05,
+      "loss": 0.3959,
+      "step": 5473
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.1576373227520823e-05,
+      "loss": 0.7168,
+      "step": 5474
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.1567749312464468e-05,
+      "loss": 0.7045,
+      "step": 5475
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.1559127643874493e-05,
+      "loss": 0.821,
+      "step": 5476
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.1550508223192827e-05,
+      "loss": 0.8321,
+      "step": 5477
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.1541891051860997e-05,
+      "loss": 0.784,
+      "step": 5478
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.1533276131320186e-05,
+      "loss": 0.78,
+      "step": 5479
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.152466346301119e-05,
+      "loss": 0.8538,
+      "step": 5480
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.1516053048374414e-05,
+      "loss": 0.5632,
+      "step": 5481
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.1507444888849925e-05,
+      "loss": 0.7091,
+      "step": 5482
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.1498838985877387e-05,
+      "loss": 0.3796,
+      "step": 5483
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.1490235340896063e-05,
+      "loss": 0.5169,
+      "step": 5484
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.148163395534488e-05,
+      "loss": 0.7531,
+      "step": 5485
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.1473034830662355e-05,
+      "loss": 0.69,
+      "step": 5486
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.1464437968286665e-05,
+      "loss": 0.6774,
+      "step": 5487
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.1455843369655572e-05,
+      "loss": 0.6034,
+      "step": 5488
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.1447251036206488e-05,
+      "loss": 0.7963,
+      "step": 5489
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.1438660969376406e-05,
+      "loss": 0.6187,
+      "step": 5490
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.1430073170601968e-05,
+      "loss": 0.6362,
+      "step": 5491
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.1421487641319456e-05,
+      "loss": 0.578,
+      "step": 5492
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.1412904382964734e-05,
+      "loss": 0.7965,
+      "step": 5493
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.1404323396973307e-05,
+      "loss": 0.7183,
+      "step": 5494
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.1395744684780302e-05,
+      "loss": 0.6844,
+      "step": 5495
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.1387168247820445e-05,
+      "loss": 0.6351,
+      "step": 5496
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.137859408752809e-05,
+      "loss": 0.6548,
+      "step": 5497
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.1370022205337239e-05,
+      "loss": 0.775,
+      "step": 5498
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.1361452602681477e-05,
+      "loss": 0.5706,
+      "step": 5499
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.1352885280994022e-05,
+      "loss": 0.675,
+      "step": 5500
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.1344320241707718e-05,
+      "loss": 0.6826,
+      "step": 5501
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.1335757486254991e-05,
+      "loss": 0.6324,
+      "step": 5502
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.132719701606794e-05,
+      "loss": 0.5978,
+      "step": 5503
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.1318638832578246e-05,
+      "loss": 0.6436,
+      "step": 5504
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.1310082937217214e-05,
+      "loss": 0.6552,
+      "step": 5505
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.1301529331415769e-05,
+      "loss": 0.6826,
+      "step": 5506
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.1292978016604453e-05,
+      "loss": 0.9355,
+      "step": 5507
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.1284428994213422e-05,
+      "loss": 0.56,
+      "step": 5508
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.1275882265672455e-05,
+      "loss": 0.6075,
+      "step": 5509
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.1267337832410935e-05,
+      "loss": 0.65,
+      "step": 5510
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.1258795695857877e-05,
+      "loss": 0.6615,
+      "step": 5511
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.12502558574419e-05,
+      "loss": 0.7258,
+      "step": 5512
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.1241718318591243e-05,
+      "loss": 0.7143,
+      "step": 5513
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.1233183080733764e-05,
+      "loss": 0.5529,
+      "step": 5514
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.1224650145296924e-05,
+      "loss": 0.6401,
+      "step": 5515
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.1216119513707815e-05,
+      "loss": 0.5391,
+      "step": 5516
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.1207591187393135e-05,
+      "loss": 0.7081,
+      "step": 5517
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.119906516777919e-05,
+      "loss": 0.5757,
+      "step": 5518
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.1190541456291917e-05,
+      "loss": 0.5697,
+      "step": 5519
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.1182020054356854e-05,
+      "loss": 0.5554,
+      "step": 5520
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.1173500963399152e-05,
+      "loss": 0.6002,
+      "step": 5521
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.1164984184843586e-05,
+      "loss": 0.4565,
+      "step": 5522
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.1156469720114534e-05,
+      "loss": 0.577,
+      "step": 5523
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.1147957570635995e-05,
+      "loss": 0.7204,
+      "step": 5524
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.1139447737831573e-05,
+      "loss": 0.5615,
+      "step": 5525
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.1130940223124489e-05,
+      "loss": 0.6794,
+      "step": 5526
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.1122435027937577e-05,
+      "loss": 0.661,
+      "step": 5527
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.1113932153693284e-05,
+      "loss": 0.5797,
+      "step": 5528
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.1105431601813662e-05,
+      "loss": 0.5255,
+      "step": 5529
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.1096933373720381e-05,
+      "loss": 0.7032,
+      "step": 5530
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.108843747083472e-05,
+      "loss": 0.5442,
+      "step": 5531
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.1079943894577572e-05,
+      "loss": 0.7477,
+      "step": 5532
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.1071452646369437e-05,
+      "loss": 0.4467,
+      "step": 5533
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.1062963727630419e-05,
+      "loss": 0.7476,
+      "step": 5534
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.1054477139780265e-05,
+      "loss": 0.5466,
+      "step": 5535
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.1045992884238282e-05,
+      "loss": 0.4847,
+      "step": 5536
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.1037510962423425e-05,
+      "loss": 0.6048,
+      "step": 5537
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.1029031375754243e-05,
+      "loss": 0.743,
+      "step": 5538
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.10205541256489e-05,
+      "loss": 0.6563,
+      "step": 5539
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.1012079213525153e-05,
+      "loss": 0.7944,
+      "step": 5540
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.1003606640800419e-05,
+      "loss": 0.5007,
+      "step": 5541
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.099513640889165e-05,
+      "loss": 0.6747,
+      "step": 5542
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.0986668519215457e-05,
+      "loss": 0.5943,
+      "step": 5543
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.0978202973188045e-05,
+      "loss": 0.4965,
+      "step": 5544
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.0969739772225218e-05,
+      "loss": 0.3978,
+      "step": 5545
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.0961278917742419e-05,
+      "loss": 0.5212,
+      "step": 5546
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.0952820411154674e-05,
+      "loss": 0.7165,
+      "step": 5547
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.0944364253876599e-05,
+      "loss": 0.5206,
+      "step": 5548
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.093591044732245e-05,
+      "loss": 0.6285,
+      "step": 5549
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.092745899290607e-05,
+      "loss": 0.7367,
+      "step": 5550
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.091900989204093e-05,
+      "loss": 0.6553,
+      "step": 5551
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.0910563146140084e-05,
+      "loss": 0.651,
+      "step": 5552
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.0902118756616215e-05,
+      "loss": 0.6254,
+      "step": 5553
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.0893676724881577e-05,
+      "loss": 0.7091,
+      "step": 5554
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.088523705234806e-05,
+      "loss": 0.6821,
+      "step": 5555
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.087679974042714e-05,
+      "loss": 0.597,
+      "step": 5556
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.0868364790529933e-05,
+      "loss": 0.6137,
+      "step": 5557
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0859932204067122e-05,
+      "loss": 0.6091,
+      "step": 5558
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0851501982449016e-05,
+      "loss": 0.6119,
+      "step": 5559
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0843074127085507e-05,
+      "loss": 0.6387,
+      "step": 5560
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.0834648639386103e-05,
+      "loss": 0.6756,
+      "step": 5561
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0826225520759933e-05,
+      "loss": 0.7889,
+      "step": 5562
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0817804772615713e-05,
+      "loss": 0.6369,
+      "step": 5563
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.080938639636176e-05,
+      "loss": 0.6111,
+      "step": 5564
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0800970393406008e-05,
+      "loss": 0.6714,
+      "step": 5565
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.0792556765155956e-05,
+      "loss": 0.622,
+      "step": 5566
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.0784145513018768e-05,
+      "loss": 0.6123,
+      "step": 5567
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.0775736638401163e-05,
+      "loss": 0.5984,
+      "step": 5568
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.0767330142709478e-05,
+      "loss": 0.5786,
+      "step": 5569
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.075892602734965e-05,
+      "loss": 0.6721,
+      "step": 5570
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.0750524293727232e-05,
+      "loss": 0.5371,
+      "step": 5571
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.0742124943247334e-05,
+      "loss": 0.5753,
+      "step": 5572
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.0733727977314725e-05,
+      "loss": 0.8292,
+      "step": 5573
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.0725333397333746e-05,
+      "loss": 0.8788,
+      "step": 5574
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.0716941204708335e-05,
+      "loss": 0.842,
+      "step": 5575
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.0708551400842042e-05,
+      "loss": 0.6302,
+      "step": 5576
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.0700163987138013e-05,
+      "loss": 0.6927,
+      "step": 5577
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.0691778964998993e-05,
+      "loss": 0.5921,
+      "step": 5578
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.0683396335827331e-05,
+      "loss": 0.8606,
+      "step": 5579
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.067501610102497e-05,
+      "loss": 0.6231,
+      "step": 5580
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.0666638261993458e-05,
+      "loss": 0.5831,
+      "step": 5581
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.065826282013394e-05,
+      "loss": 0.7434,
+      "step": 5582
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.0649889776847161e-05,
+      "loss": 0.5954,
+      "step": 5583
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.0641519133533465e-05,
+      "loss": 0.7423,
+      "step": 5584
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.063315089159279e-05,
+      "loss": 0.711,
+      "step": 5585
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.0624785052424676e-05,
+      "loss": 0.5981,
+      "step": 5586
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.0616421617428266e-05,
+      "loss": 0.5148,
+      "step": 5587
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.0608060588002291e-05,
+      "loss": 0.7185,
+      "step": 5588
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.0599701965545088e-05,
+      "loss": 0.5995,
+      "step": 5589
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.0591345751454593e-05,
+      "loss": 0.556,
+      "step": 5590
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.0582991947128324e-05,
+      "loss": 0.7054,
+      "step": 5591
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.0574640553963413e-05,
+      "loss": 0.4782,
+      "step": 5592
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.0566291573356585e-05,
+      "loss": 0.5657,
+      "step": 5593
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.0557945006704151e-05,
+      "loss": 0.4779,
+      "step": 5594
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.054960085540203e-05,
+      "loss": 0.4714,
+      "step": 5595
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.0541259120845736e-05,
+      "loss": 0.8788,
+      "step": 5596
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.0532919804430372e-05,
+      "loss": 0.6607,
+      "step": 5597
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.0524582907550642e-05,
+      "loss": 0.515,
+      "step": 5598
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.0516248431600845e-05,
+      "loss": 0.4383,
+      "step": 5599
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.0507916377974868e-05,
+      "loss": 0.593,
+      "step": 5600
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.049958674806621e-05,
+      "loss": 0.7292,
+      "step": 5601
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.0491259543267942e-05,
+      "loss": 0.7293,
+      "step": 5602
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.0482934764972748e-05,
+      "loss": 0.5426,
+      "step": 5603
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.0474612414572887e-05,
+      "loss": 0.6259,
+      "step": 5604
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.0466292493460253e-05,
+      "loss": 0.6976,
+      "step": 5605
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.0457975003026276e-05,
+      "loss": 0.5121,
+      "step": 5606
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.044965994466202e-05,
+      "loss": 0.5262,
+      "step": 5607
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.0441347319758127e-05,
+      "loss": 0.4956,
+      "step": 5608
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.0433037129704828e-05,
+      "loss": 0.8247,
+      "step": 5609
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.0424729375891973e-05,
+      "loss": 0.713,
+      "step": 5610
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.0416424059708985e-05,
+      "loss": 0.6473,
+      "step": 5611
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.0408121182544864e-05,
+      "loss": 0.6098,
+      "step": 5612
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.0399820745788225e-05,
+      "loss": 0.8381,
+      "step": 5613
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.0391522750827268e-05,
+      "loss": 0.6492,
+      "step": 5614
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.0383227199049774e-05,
+      "loss": 0.6367,
+      "step": 5615
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.0374934091843153e-05,
+      "loss": 0.7762,
+      "step": 5616
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.036664343059437e-05,
+      "loss": 0.9352,
+      "step": 5617
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.0358355216689975e-05,
+      "loss": 0.3704,
+      "step": 5618
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.0350069451516132e-05,
+      "loss": 0.6854,
+      "step": 5619
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.0341786136458581e-05,
+      "loss": 0.6463,
+      "step": 5620
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.0333505272902674e-05,
+      "loss": 0.4521,
+      "step": 5621
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.0325226862233331e-05,
+      "loss": 0.4871,
+      "step": 5622
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.0316950905835077e-05,
+      "loss": 0.7722,
+      "step": 5623
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.0308677405091998e-05,
+      "loss": 0.6699,
+      "step": 5624
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.030040636138779e-05,
+      "loss": 0.7263,
+      "step": 5625
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.0292137776105756e-05,
+      "loss": 0.7262,
+      "step": 5626
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.028387165062876e-05,
+      "loss": 0.6523,
+      "step": 5627
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.0275607986339263e-05,
+      "loss": 0.6952,
+      "step": 5628
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.0267346784619324e-05,
+      "loss": 0.6961,
+      "step": 5629
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.0259088046850568e-05,
+      "loss": 0.6652,
+      "step": 5630
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.0250831774414212e-05,
+      "loss": 0.7669,
+      "step": 5631
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.0242577968691096e-05,
+      "loss": 0.6129,
+      "step": 5632
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.0234326631061612e-05,
+      "loss": 0.4664,
+      "step": 5633
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.0226077762905742e-05,
+      "loss": 0.8174,
+      "step": 5634
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.0217831365603075e-05,
+      "loss": 0.8255,
+      "step": 5635
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.0209587440532748e-05,
+      "loss": 0.5025,
+      "step": 5636
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.0201345989073532e-05,
+      "loss": 0.6202,
+      "step": 5637
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.0193107012603759e-05,
+      "loss": 0.6304,
+      "step": 5638
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.0184870512501346e-05,
+      "loss": 0.7114,
+      "step": 5639
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.0176636490143801e-05,
+      "loss": 0.5478,
+      "step": 5640
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.0168404946908225e-05,
+      "loss": 0.5982,
+      "step": 5641
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.0160175884171273e-05,
+      "loss": 0.9346,
+      "step": 5642
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.015194930330923e-05,
+      "loss": 0.7065,
+      "step": 5643
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.0143725205697937e-05,
+      "loss": 0.6837,
+      "step": 5644
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.0135503592712827e-05,
+      "loss": 0.6279,
+      "step": 5645
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.012728446572892e-05,
+      "loss": 0.4094,
+      "step": 5646
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.0119067826120814e-05,
+      "loss": 0.4261,
+      "step": 5647
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.0110853675262697e-05,
+      "loss": 0.7176,
+      "step": 5648
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.0102642014528335e-05,
+      "loss": 0.727,
+      "step": 5649
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.0094432845291082e-05,
+      "loss": 0.6505,
+      "step": 5650
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.0086226168923878e-05,
+      "loss": 0.6954,
+      "step": 5651
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.0078021986799238e-05,
+      "loss": 0.4495,
+      "step": 5652
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.0069820300289264e-05,
+      "loss": 0.7986,
+      "step": 5653
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.0061621110765646e-05,
+      "loss": 0.4831,
+      "step": 5654
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.005342441959964e-05,
+      "loss": 0.4903,
+      "step": 5655
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.004523022816211e-05,
+      "loss": 0.599,
+      "step": 5656
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.0037038537823476e-05,
+      "loss": 0.5773,
+      "step": 5657
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.0028849349953756e-05,
+      "loss": 0.6622,
+      "step": 5658
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.002066266592254e-05,
+      "loss": 0.6829,
+      "step": 5659
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.0012478487099008e-05,
+      "loss": 0.5502,
+      "step": 5660
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.0004296814851915e-05,
+      "loss": 0.5468,
+      "step": 5661
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 9.996117650549602e-06,
+      "loss": 0.4895,
+      "step": 5662
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 9.987940995559972e-06,
+      "loss": 0.4575,
+      "step": 5663
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 9.979766851250552e-06,
+      "loss": 0.4954,
+      "step": 5664
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 9.971595218988397e-06,
+      "loss": 0.7278,
+      "step": 5665
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 9.963426100140168e-06,
+      "loss": 0.692,
+      "step": 5666
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 9.955259496072112e-06,
+      "loss": 0.5059,
+      "step": 5667
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 9.947095408150028e-06,
+      "loss": 0.8233,
+      "step": 5668
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 9.938933837739337e-06,
+      "loss": 0.4828,
+      "step": 5669
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 9.930774786205011e-06,
+      "loss": 0.8204,
+      "step": 5670
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 9.922618254911587e-06,
+      "loss": 0.8073,
+      "step": 5671
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 9.914464245223207e-06,
+      "loss": 0.6133,
+      "step": 5672
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 9.906312758503583e-06,
+      "loss": 0.841,
+      "step": 5673
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 9.898163796115994e-06,
+      "loss": 0.4484,
+      "step": 5674
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 9.890017359423325e-06,
+      "loss": 0.7491,
+      "step": 5675
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 9.881873449788023e-06,
+      "loss": 0.5595,
+      "step": 5676
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 9.873732068572087e-06,
+      "loss": 0.6138,
+      "step": 5677
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 9.865593217137128e-06,
+      "loss": 0.7448,
+      "step": 5678
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 9.857456896844314e-06,
+      "loss": 0.6369,
+      "step": 5679
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 9.849323109054415e-06,
+      "loss": 0.3782,
+      "step": 5680
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 9.841191855127748e-06,
+      "loss": 0.6213,
+      "step": 5681
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 9.833063136424229e-06,
+      "loss": 0.8098,
+      "step": 5682
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 9.824936954303324e-06,
+      "loss": 0.5918,
+      "step": 5683
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 9.816813310124088e-06,
+      "loss": 0.5854,
+      "step": 5684
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 9.808692205245168e-06,
+      "loss": 0.6877,
+      "step": 5685
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 9.800573641024768e-06,
+      "loss": 0.6264,
+      "step": 5686
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 9.792457618820669e-06,
+      "loss": 0.8096,
+      "step": 5687
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 9.784344139990239e-06,
+      "loss": 0.5736,
+      "step": 5688
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 9.776233205890392e-06,
+      "loss": 0.7086,
+      "step": 5689
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 9.768124817877632e-06,
+      "loss": 0.5912,
+      "step": 5690
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 9.760018977308064e-06,
+      "loss": 0.52,
+      "step": 5691
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 9.751915685537327e-06,
+      "loss": 0.7659,
+      "step": 5692
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 9.743814943920657e-06,
+      "loss": 0.532,
+      "step": 5693
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 9.73571675381286e-06,
+      "loss": 0.7005,
+      "step": 5694
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 9.727621116568287e-06,
+      "loss": 0.572,
+      "step": 5695
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 9.71952803354091e-06,
+      "loss": 0.8621,
+      "step": 5696
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 9.711437506084245e-06,
+      "loss": 0.4733,
+      "step": 5697
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 9.703349535551387e-06,
+      "loss": 0.6774,
+      "step": 5698
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 9.695264123295e-06,
+      "loss": 0.7279,
+      "step": 5699
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 9.687181270667331e-06,
+      "loss": 0.6756,
+      "step": 5700
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 9.679100979020164e-06,
+      "loss": 0.6539,
+      "step": 5701
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 9.67102324970491e-06,
+      "loss": 0.6201,
+      "step": 5702
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 9.662948084072513e-06,
+      "loss": 0.7302,
+      "step": 5703
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 9.654875483473497e-06,
+      "loss": 0.659,
+      "step": 5704
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 9.646805449257957e-06,
+      "loss": 0.6843,
+      "step": 5705
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 9.638737982775558e-06,
+      "loss": 0.5871,
+      "step": 5706
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 9.630673085375544e-06,
+      "loss": 0.8524,
+      "step": 5707
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 9.622610758406717e-06,
+      "loss": 0.7064,
+      "step": 5708
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 9.614551003217453e-06,
+      "loss": 0.902,
+      "step": 5709
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 9.606493821155707e-06,
+      "loss": 0.5322,
+      "step": 5710
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 9.598439213568988e-06,
+      "loss": 0.4984,
+      "step": 5711
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 9.590387181804389e-06,
+      "loss": 0.6554,
+      "step": 5712
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 9.582337727208559e-06,
+      "loss": 0.4111,
+      "step": 5713
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 9.574290851127728e-06,
+      "loss": 0.5698,
+      "step": 5714
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 9.566246554907693e-06,
+      "loss": 0.5617,
+      "step": 5715
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 9.558204839893805e-06,
+      "loss": 0.6085,
+      "step": 5716
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 9.550165707431006e-06,
+      "loss": 0.527,
+      "step": 5717
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 9.542129158863788e-06,
+      "loss": 0.6901,
+      "step": 5718
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 9.534095195536216e-06,
+      "loss": 0.526,
+      "step": 5719
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 9.526063818791931e-06,
+      "loss": 0.6726,
+      "step": 5720
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 9.518035029974126e-06,
+      "loss": 0.5701,
+      "step": 5721
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 9.510008830425576e-06,
+      "loss": 0.7365,
+      "step": 5722
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 9.501985221488613e-06,
+      "loss": 0.703,
+      "step": 5723
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 9.49396420450514e-06,
+      "loss": 0.618,
+      "step": 5724
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 9.48594578081663e-06,
+      "loss": 0.5779,
+      "step": 5725
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 9.477929951764111e-06,
+      "loss": 0.5898,
+      "step": 5726
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 9.46991671868818e-06,
+      "loss": 0.7169,
+      "step": 5727
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 9.461906082929026e-06,
+      "loss": 0.6456,
+      "step": 5728
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 9.453898045826362e-06,
+      "loss": 0.5151,
+      "step": 5729
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 9.44589260871949e-06,
+      "loss": 0.6476,
+      "step": 5730
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 9.437889772947273e-06,
+      "loss": 0.4133,
+      "step": 5731
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 9.429889539848139e-06,
+      "loss": 0.9183,
+      "step": 5732
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 9.421891910760075e-06,
+      "loss": 0.6645,
+      "step": 5733
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 9.413896887020662e-06,
+      "loss": 0.7946,
+      "step": 5734
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 9.405904469966997e-06,
+      "loss": 0.7552,
+      "step": 5735
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 9.397914660935772e-06,
+      "loss": 0.6442,
+      "step": 5736
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 9.38992746126324e-06,
+      "loss": 0.6686,
+      "step": 5737
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 9.381942872285202e-06,
+      "loss": 0.5412,
+      "step": 5738
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 9.373960895337054e-06,
+      "loss": 0.6162,
+      "step": 5739
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 9.365981531753731e-06,
+      "loss": 0.6491,
+      "step": 5740
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 9.358004782869726e-06,
+      "loss": 0.7425,
+      "step": 5741
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 9.350030650019103e-06,
+      "loss": 0.6656,
+      "step": 5742
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 9.342059134535491e-06,
+      "loss": 0.7015,
+      "step": 5743
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 9.334090237752094e-06,
+      "loss": 0.557,
+      "step": 5744
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 9.32612396100165e-06,
+      "loss": 0.6318,
+      "step": 5745
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 9.318160305616488e-06,
+      "loss": 0.7336,
+      "step": 5746
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 9.310199272928463e-06,
+      "loss": 0.6765,
+      "step": 5747
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 9.30224086426902e-06,
+      "loss": 0.6213,
+      "step": 5748
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 9.29428508096915e-06,
+      "loss": 0.563,
+      "step": 5749
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 9.286331924359431e-06,
+      "loss": 0.601,
+      "step": 5750
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 9.27838139576997e-06,
+      "loss": 0.548,
+      "step": 5751
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 9.270433496530453e-06,
+      "loss": 0.6188,
+      "step": 5752
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 9.262488227970112e-06,
+      "loss": 0.6722,
+      "step": 5753
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 9.25454559141774e-06,
+      "loss": 0.7365,
+      "step": 5754
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 9.246605588201714e-06,
+      "loss": 0.7575,
+      "step": 5755
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 9.23866821964995e-06,
+      "loss": 0.4109,
+      "step": 5756
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 9.230733487089927e-06,
+      "loss": 0.4762,
+      "step": 5757
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 9.222801391848687e-06,
+      "loss": 0.5784,
+      "step": 5758
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 9.214871935252814e-06,
+      "loss": 0.7036,
+      "step": 5759
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 9.20694511862846e-06,
+      "loss": 0.6645,
+      "step": 5760
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 9.199020943301361e-06,
+      "loss": 0.6356,
+      "step": 5761
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 9.19109941059678e-06,
+      "loss": 0.6928,
+      "step": 5762
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 9.18318052183954e-06,
+      "loss": 0.6752,
+      "step": 5763
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 9.17526427835405e-06,
+      "loss": 0.4754,
+      "step": 5764
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 9.167350681464223e-06,
+      "loss": 0.5951,
+      "step": 5765
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 9.15943973249359e-06,
+      "loss": 0.8455,
+      "step": 5766
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 9.151531432765203e-06,
+      "loss": 0.6396,
+      "step": 5767
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 9.143625783601679e-06,
+      "loss": 0.6903,
+      "step": 5768
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 9.135722786325191e-06,
+      "loss": 0.6723,
+      "step": 5769
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 9.127822442257472e-06,
+      "loss": 0.6529,
+      "step": 5770
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 9.119924752719807e-06,
+      "loss": 0.6272,
+      "step": 5771
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 9.112029719033045e-06,
+      "loss": 0.6254,
+      "step": 5772
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 9.104137342517573e-06,
+      "loss": 0.5955,
+      "step": 5773
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 9.096247624493356e-06,
+      "loss": 0.6646,
+      "step": 5774
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 9.088360566279899e-06,
+      "loss": 0.5753,
+      "step": 5775
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 9.080476169196268e-06,
+      "loss": 0.4146,
+      "step": 5776
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 9.072594434561082e-06,
+      "loss": 0.7158,
+      "step": 5777
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 9.064715363692516e-06,
+      "loss": 0.618,
+      "step": 5778
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 9.056838957908301e-06,
+      "loss": 0.6873,
+      "step": 5779
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 9.048965218525719e-06,
+      "loss": 0.5405,
+      "step": 5780
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 9.041094146861606e-06,
+      "loss": 0.6847,
+      "step": 5781
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 9.033225744232355e-06,
+      "loss": 0.5546,
+      "step": 5782
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 9.025360011953909e-06,
+      "loss": 0.4969,
+      "step": 5783
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 9.017496951341772e-06,
+      "loss": 0.6358,
+      "step": 5784
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 9.009636563710988e-06,
+      "loss": 0.7545,
+      "step": 5785
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 9.001778850376167e-06,
+      "loss": 0.5163,
+      "step": 5786
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 8.993923812651461e-06,
+      "loss": 0.6814,
+      "step": 5787
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 8.986071451850586e-06,
+      "loss": 0.623,
+      "step": 5788
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 8.978221769286799e-06,
+      "loss": 0.8089,
+      "step": 5789
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 8.970374766272915e-06,
+      "loss": 0.7876,
+      "step": 5790
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 8.9625304441213e-06,
+      "loss": 0.5381,
+      "step": 5791
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 8.954688804143876e-06,
+      "loss": 0.61,
+      "step": 5792
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 8.946849847652106e-06,
+      "loss": 0.7334,
+      "step": 5793
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 8.939013575957011e-06,
+      "loss": 0.7095,
+      "step": 5794
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 8.931179990369168e-06,
+      "loss": 0.6257,
+      "step": 5795
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 8.923349092198691e-06,
+      "loss": 0.7073,
+      "step": 5796
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 8.915520882755248e-06,
+      "loss": 0.652,
+      "step": 5797
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 8.90769536334809e-06,
+      "loss": 0.555,
+      "step": 5798
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 8.899872535285958e-06,
+      "loss": 0.4508,
+      "step": 5799
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 8.892052399877188e-06,
+      "loss": 0.7051,
+      "step": 5800
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 8.884234958429652e-06,
+      "loss": 0.6295,
+      "step": 5801
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 8.876420212250761e-06,
+      "loss": 0.5437,
+      "step": 5802
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 8.868608162647507e-06,
+      "loss": 0.5877,
+      "step": 5803
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 8.860798810926407e-06,
+      "loss": 0.5985,
+      "step": 5804
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 8.852992158393514e-06,
+      "loss": 0.4578,
+      "step": 5805
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 8.845188206354455e-06,
+      "loss": 0.6391,
+      "step": 5806
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 8.837386956114393e-06,
+      "loss": 0.759,
+      "step": 5807
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 8.829588408978037e-06,
+      "loss": 0.5942,
+      "step": 5808
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 8.821792566249664e-06,
+      "loss": 0.6526,
+      "step": 5809
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 8.813999429233085e-06,
+      "loss": 0.7863,
+      "step": 5810
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 8.806208999231641e-06,
+      "loss": 0.5261,
+      "step": 5811
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 8.79842127754824e-06,
+      "loss": 0.715,
+      "step": 5812
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 8.790636265485334e-06,
+      "loss": 0.5038,
+      "step": 5813
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 8.782853964344925e-06,
+      "loss": 0.5845,
+      "step": 5814
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 8.775074375428562e-06,
+      "loss": 0.6739,
+      "step": 5815
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 8.767297500037339e-06,
+      "loss": 0.5567,
+      "step": 5816
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 8.759523339471878e-06,
+      "loss": 0.5767,
+      "step": 5817
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 8.75175189503237e-06,
+      "loss": 0.4693,
+      "step": 5818
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 8.743983168018536e-06,
+      "loss": 0.607,
+      "step": 5819
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 8.736217159729668e-06,
+      "loss": 0.5199,
+      "step": 5820
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 8.728453871464579e-06,
+      "loss": 0.5609,
+      "step": 5821
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 8.720693304521638e-06,
+      "loss": 0.6412,
+      "step": 5822
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 8.712935460198743e-06,
+      "loss": 0.7573,
+      "step": 5823
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 8.705180339793348e-06,
+      "loss": 0.615,
+      "step": 5824
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 8.697427944602465e-06,
+      "loss": 0.5342,
+      "step": 5825
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 8.689678275922633e-06,
+      "loss": 0.8229,
+      "step": 5826
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 8.681931335049939e-06,
+      "loss": 0.6123,
+      "step": 5827
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 8.67418712328002e-06,
+      "loss": 0.6616,
+      "step": 5828
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 8.666445641908027e-06,
+      "loss": 0.667,
+      "step": 5829
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 8.658706892228705e-06,
+      "loss": 0.5817,
+      "step": 5830
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 8.650970875536304e-06,
+      "loss": 0.5968,
+      "step": 5831
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 8.643237593124632e-06,
+      "loss": 0.7257,
+      "step": 5832
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 8.635507046287028e-06,
+      "loss": 0.6689,
+      "step": 5833
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 8.6277792363164e-06,
+      "loss": 0.5987,
+      "step": 5834
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 8.620054164505145e-06,
+      "loss": 0.5248,
+      "step": 5835
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 8.612331832145268e-06,
+      "loss": 0.7045,
+      "step": 5836
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 8.604612240528275e-06,
+      "loss": 0.6161,
+      "step": 5837
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 8.59689539094522e-06,
+      "loss": 0.6374,
+      "step": 5838
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 8.589181284686707e-06,
+      "loss": 0.5285,
+      "step": 5839
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 8.581469923042871e-06,
+      "loss": 0.3869,
+      "step": 5840
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 8.573761307303393e-06,
+      "loss": 0.7623,
+      "step": 5841
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 8.566055438757498e-06,
+      "loss": 0.5375,
+      "step": 5842
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 8.558352318693947e-06,
+      "loss": 0.606,
+      "step": 5843
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 8.550651948401036e-06,
+      "loss": 0.6872,
+      "step": 5844
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 8.542954329166616e-06,
+      "loss": 0.5619,
+      "step": 5845
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 8.535259462278067e-06,
+      "loss": 0.6369,
+      "step": 5846
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 8.527567349022309e-06,
+      "loss": 0.5341,
+      "step": 5847
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 8.519877990685803e-06,
+      "loss": 0.3197,
+      "step": 5848
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 8.512191388554552e-06,
+      "loss": 0.8312,
+      "step": 5849
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 8.504507543914098e-06,
+      "loss": 0.7387,
+      "step": 5850
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 8.496826458049515e-06,
+      "loss": 0.6516,
+      "step": 5851
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 8.489148132245422e-06,
+      "loss": 0.8249,
+      "step": 5852
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 8.481472567785975e-06,
+      "loss": 0.7943,
+      "step": 5853
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 8.473799765954868e-06,
+      "loss": 0.5383,
+      "step": 5854
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 8.46612972803533e-06,
+      "loss": 0.6587,
+      "step": 5855
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 8.458462455310135e-06,
+      "loss": 0.687,
+      "step": 5856
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 8.45079794906159e-06,
+      "loss": 0.6758,
+      "step": 5857
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 8.443136210571534e-06,
+      "loss": 0.8395,
+      "step": 5858
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 8.435477241121353e-06,
+      "loss": 0.5271,
+      "step": 5859
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 8.427821041991965e-06,
+      "loss": 0.5882,
+      "step": 5860
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 8.420167614463817e-06,
+      "loss": 0.5872,
+      "step": 5861
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 8.412516959816924e-06,
+      "loss": 0.6397,
+      "step": 5862
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 8.40486907933079e-06,
+      "loss": 0.5999,
+      "step": 5863
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 8.397223974284485e-06,
+      "loss": 0.7017,
+      "step": 5864
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 8.389581645956612e-06,
+      "loss": 0.6238,
+      "step": 5865
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 8.381942095625305e-06,
+      "loss": 0.6674,
+      "step": 5866
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 8.374305324568227e-06,
+      "loss": 0.5336,
+      "step": 5867
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 8.366671334062606e-06,
+      "loss": 0.5299,
+      "step": 5868
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 8.359040125385159e-06,
+      "loss": 0.3766,
+      "step": 5869
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 8.351411699812173e-06,
+      "loss": 0.7094,
+      "step": 5870
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 8.343786058619455e-06,
+      "loss": 0.6866,
+      "step": 5871
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 8.336163203082343e-06,
+      "loss": 0.808,
+      "step": 5872
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 8.328543134475728e-06,
+      "loss": 0.8167,
+      "step": 5873
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 8.32092585407403e-06,
+      "loss": 0.7674,
+      "step": 5874
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 8.313311363151172e-06,
+      "loss": 0.7665,
+      "step": 5875
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 8.305699662980643e-06,
+      "loss": 0.8449,
+      "step": 5876
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 8.298090754835458e-06,
+      "loss": 0.5564,
+      "step": 5877
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 8.290484639988155e-06,
+      "loss": 0.6795,
+      "step": 5878
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 8.282881319710827e-06,
+      "loss": 0.3593,
+      "step": 5879
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 8.275280795275086e-06,
+      "loss": 0.4894,
+      "step": 5880
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 8.267683067952062e-06,
+      "loss": 0.7392,
+      "step": 5881
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 8.260088139012435e-06,
+      "loss": 0.6742,
+      "step": 5882
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 8.252496009726407e-06,
+      "loss": 0.657,
+      "step": 5883
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 8.244906681363734e-06,
+      "loss": 0.5808,
+      "step": 5884
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 8.237320155193681e-06,
+      "loss": 0.7927,
+      "step": 5885
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 8.229736432485059e-06,
+      "loss": 0.5958,
+      "step": 5886
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 8.222155514506177e-06,
+      "loss": 0.6188,
+      "step": 5887
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 8.214577402524911e-06,
+      "loss": 0.5662,
+      "step": 5888
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 8.20700209780867e-06,
+      "loss": 0.7781,
+      "step": 5889
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 8.199429601624364e-06,
+      "loss": 0.6939,
+      "step": 5890
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 8.191859915238459e-06,
+      "loss": 0.6589,
+      "step": 5891
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 8.184293039916946e-06,
+      "loss": 0.6217,
+      "step": 5892
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 8.176728976925323e-06,
+      "loss": 0.6335,
+      "step": 5893
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 8.16916772752864e-06,
+      "loss": 0.7717,
+      "step": 5894
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 8.161609292991484e-06,
+      "loss": 0.5462,
+      "step": 5895
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 8.154053674577954e-06,
+      "loss": 0.6694,
+      "step": 5896
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 8.146500873551684e-06,
+      "loss": 0.6708,
+      "step": 5897
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 8.138950891175845e-06,
+      "loss": 0.6181,
+      "step": 5898
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 8.1314037287131e-06,
+      "loss": 0.58,
+      "step": 5899
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 8.123859387425695e-06,
+      "loss": 0.639,
+      "step": 5900
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 8.116317868575369e-06,
+      "loss": 0.6451,
+      "step": 5901
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 8.108779173423398e-06,
+      "loss": 0.6678,
+      "step": 5902
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 8.101243303230586e-06,
+      "loss": 0.907,
+      "step": 5903
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 8.093710259257262e-06,
+      "loss": 0.5348,
+      "step": 5904
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 8.086180042763283e-06,
+      "loss": 0.591,
+      "step": 5905
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 8.07865265500804e-06,
+      "loss": 0.634,
+      "step": 5906
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 8.071128097250439e-06,
+      "loss": 0.6447,
+      "step": 5907
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 8.063606370748921e-06,
+      "loss": 0.7084,
+      "step": 5908
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 8.056087476761451e-06,
+      "loss": 0.7016,
+      "step": 5909
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 8.048571416545523e-06,
+      "loss": 0.5331,
+      "step": 5910
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 8.04105819135815e-06,
+      "loss": 0.6258,
+      "step": 5911
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 8.033547802455882e-06,
+      "loss": 0.5219,
+      "step": 5912
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 8.026040251094779e-06,
+      "loss": 0.6977,
+      "step": 5913
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 8.018535538530448e-06,
+      "loss": 0.5681,
+      "step": 5914
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 8.011033666017997e-06,
+      "loss": 0.5562,
+      "step": 5915
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 8.003534634812076e-06,
+      "loss": 0.5329,
+      "step": 5916
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 7.996038446166856e-06,
+      "loss": 0.5848,
+      "step": 5917
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 7.988545101336028e-06,
+      "loss": 0.4493,
+      "step": 5918
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 7.981054601572816e-06,
+      "loss": 0.5664,
+      "step": 5919
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 7.973566948129946e-06,
+      "loss": 0.6972,
+      "step": 5920
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 7.966082142259718e-06,
+      "loss": 0.5407,
+      "step": 5921
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 7.958600185213893e-06,
+      "loss": 0.6643,
+      "step": 5922
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 7.951121078243794e-06,
+      "loss": 0.6397,
+      "step": 5923
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 7.94364482260026e-06,
+      "loss": 0.5738,
+      "step": 5924
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 7.936171419533653e-06,
+      "loss": 0.5025,
+      "step": 5925
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 7.928700870293843e-06,
+      "loss": 0.6793,
+      "step": 5926
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 7.921233176130264e-06,
+      "loss": 0.5265,
+      "step": 5927
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 7.913768338291821e-06,
+      "loss": 0.733,
+      "step": 5928
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 7.906306358026972e-06,
+      "loss": 0.4299,
+      "step": 5929
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 7.89884723658369e-06,
+      "loss": 0.7359,
+      "step": 5930
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 7.891390975209464e-06,
+      "loss": 0.5283,
+      "step": 5931
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 7.883937575151323e-06,
+      "loss": 0.464,
+      "step": 5932
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 7.876487037655808e-06,
+      "loss": 0.5819,
+      "step": 5933
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 7.86903936396896e-06,
+      "loss": 0.7297,
+      "step": 5934
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 7.86159455533637e-06,
+      "loss": 0.6413,
+      "step": 5935
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 7.85415261300314e-06,
+      "loss": 0.7831,
+      "step": 5936
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 7.846713538213882e-06,
+      "loss": 0.4898,
+      "step": 5937
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 7.839277332212753e-06,
+      "loss": 0.6625,
+      "step": 5938
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 7.831843996243416e-06,
+      "loss": 0.5727,
+      "step": 5939
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 7.824413531549037e-06,
+      "loss": 0.4775,
+      "step": 5940
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 7.816985939372331e-06,
+      "loss": 0.3746,
+      "step": 5941
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 7.809561220955507e-06,
+      "loss": 0.4982,
+      "step": 5942
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 7.802139377540326e-06,
+      "loss": 0.7111,
+      "step": 5943
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 7.794720410368039e-06,
+      "loss": 0.5029,
+      "step": 5944
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 7.787304320679432e-06,
+      "loss": 0.6181,
+      "step": 5945
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 7.779891109714785e-06,
+      "loss": 0.7273,
+      "step": 5946
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 7.77248077871392e-06,
+      "loss": 0.6436,
+      "step": 5947
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 7.765073328916186e-06,
+      "loss": 0.6433,
+      "step": 5948
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 7.757668761560427e-06,
+      "loss": 0.6078,
+      "step": 5949
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 7.750267077885016e-06,
+      "loss": 0.6934,
+      "step": 5950
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 7.742868279127848e-06,
+      "loss": 0.6716,
+      "step": 5951
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 7.735472366526317e-06,
+      "loss": 0.5701,
+      "step": 5952
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 7.728079341317343e-06,
+      "loss": 0.5857,
+      "step": 5953
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 7.720689204737383e-06,
+      "loss": 0.5849,
+      "step": 5954
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 7.713301958022387e-06,
+      "loss": 0.5972,
+      "step": 5955
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 7.70591760240783e-06,
+      "loss": 0.6168,
+      "step": 5956
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 7.698536139128712e-06,
+      "loss": 0.663,
+      "step": 5957
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 7.691157569419512e-06,
+      "loss": 0.7723,
+      "step": 5958
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 7.683781894514286e-06,
+      "loss": 0.6201,
+      "step": 5959
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 7.676409115646555e-06,
+      "loss": 0.5913,
+      "step": 5960
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 7.669039234049377e-06,
+      "loss": 0.645,
+      "step": 5961
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 7.661672250955323e-06,
+      "loss": 0.603,
+      "step": 5962
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 7.65430816759648e-06,
+      "loss": 0.5946,
+      "step": 5963
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 7.646946985204448e-06,
+      "loss": 0.5808,
+      "step": 5964
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 7.63958870501034e-06,
+      "loss": 0.5494,
+      "step": 5965
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 7.632233328244784e-06,
+      "loss": 0.655,
+      "step": 5966
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 7.624880856137931e-06,
+      "loss": 0.5223,
+      "step": 5967
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 7.617531289919433e-06,
+      "loss": 0.559,
+      "step": 5968
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 7.610184630818465e-06,
+      "loss": 0.8036,
+      "step": 5969
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 7.602840880063717e-06,
+      "loss": 0.8554,
+      "step": 5970
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 7.595500038883382e-06,
+      "loss": 0.8297,
+      "step": 5971
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 7.588162108505179e-06,
+      "loss": 0.6155,
+      "step": 5972
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 7.5808270901563285e-06,
+      "loss": 0.6677,
+      "step": 5973
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 7.573494985063579e-06,
+      "loss": 0.5813,
+      "step": 5974
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 7.566165794453173e-06,
+      "loss": 0.8508,
+      "step": 5975
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 7.558839519550881e-06,
+      "loss": 0.622,
+      "step": 5976
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 7.551516161581979e-06,
+      "loss": 0.5708,
+      "step": 5977
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 7.544195721771258e-06,
+      "loss": 0.722,
+      "step": 5978
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 7.536878201343014e-06,
+      "loss": 0.5825,
+      "step": 5979
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 7.529563601521067e-06,
+      "loss": 0.7239,
+      "step": 5980
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 7.522251923528737e-06,
+      "loss": 0.6978,
+      "step": 5981
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 7.514943168588859e-06,
+      "loss": 0.5918,
+      "step": 5982
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 7.5076373379237844e-06,
+      "loss": 0.4998,
+      "step": 5983
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 7.500334432755368e-06,
+      "loss": 0.7112,
+      "step": 5984
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 7.493034454304981e-06,
+      "loss": 0.5882,
+      "step": 5985
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 7.4857374037935e-06,
+      "loss": 0.539,
+      "step": 5986
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 7.478443282441316e-06,
+      "loss": 0.6955,
+      "step": 5987
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 7.4711520914683284e-06,
+      "loss": 0.463,
+      "step": 5988
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 7.463863832093948e-06,
+      "loss": 0.5558,
+      "step": 5989
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 7.456578505537084e-06,
+      "loss": 0.4562,
+      "step": 5990
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 7.449296113016188e-06,
+      "loss": 0.4627,
+      "step": 5991
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 7.442016655749179e-06,
+      "loss": 0.8627,
+      "step": 5992
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 7.434740134953508e-06,
+      "loss": 0.6332,
+      "step": 5993
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 7.4274665518461325e-06,
+      "loss": 0.5028,
+      "step": 5994
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 7.420195907643518e-06,
+      "loss": 0.4263,
+      "step": 5995
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 7.412928203561628e-06,
+      "loss": 0.5861,
+      "step": 5996
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 7.4056634408159685e-06,
+      "loss": 0.7105,
+      "step": 5997
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 7.398401620621509e-06,
+      "loss": 0.7175,
+      "step": 5998
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 7.39114274419275e-06,
+      "loss": 0.5235,
+      "step": 5999
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 7.3838868127436955e-06,
+      "loss": 0.6071,
+      "step": 6000
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 7.376633827487855e-06,
+      "loss": 0.6827,
+      "step": 6001
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 7.369383789638265e-06,
+      "loss": 0.5048,
+      "step": 6002
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 7.362136700407446e-06,
+      "loss": 0.5088,
+      "step": 6003
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 7.3548925610074235e-06,
+      "loss": 0.479,
+      "step": 6004
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 7.347651372649739e-06,
+      "loss": 0.8066,
+      "step": 6005
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 7.3404131365454344e-06,
+      "loss": 0.7,
+      "step": 6006
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 7.333177853905082e-06,
+      "loss": 0.6297,
+      "step": 6007
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 7.325945525938726e-06,
+      "loss": 0.5857,
+      "step": 6008
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 7.3187161538559475e-06,
+      "loss": 0.8261,
+      "step": 6009
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 7.311489738865796e-06,
+      "loss": 0.6426,
+      "step": 6010
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 7.304266282176858e-06,
+      "loss": 0.6175,
+      "step": 6011
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 7.297045784997206e-06,
+      "loss": 0.7585,
+      "step": 6012
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 7.289828248534441e-06,
+      "loss": 0.9113,
+      "step": 6013
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 7.28261367399565e-06,
+      "loss": 0.3447,
+      "step": 6014
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 7.275402062587433e-06,
+      "loss": 0.6701,
+      "step": 6015
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 7.268193415515876e-06,
+      "loss": 0.633,
+      "step": 6016
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 7.260987733986585e-06,
+      "loss": 0.4457,
+      "step": 6017
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 7.253785019204681e-06,
+      "loss": 0.4654,
+      "step": 6018
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 7.246585272374773e-06,
+      "loss": 0.7623,
+      "step": 6019
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 7.2393884947009745e-06,
+      "loss": 0.662,
+      "step": 6020
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 7.232194687386912e-06,
+      "loss": 0.7051,
+      "step": 6021
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 7.225003851635686e-06,
+      "loss": 0.7163,
+      "step": 6022
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 7.2178159886499484e-06,
+      "loss": 0.6449,
+      "step": 6023
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 7.210631099631815e-06,
+      "loss": 0.6819,
+      "step": 6024
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 7.2034491857829205e-06,
+      "loss": 0.6784,
+      "step": 6025
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 7.196270248304396e-06,
+      "loss": 0.6519,
+      "step": 6026
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 7.189094288396886e-06,
+      "loss": 0.7506,
+      "step": 6027
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 7.1819213072605085e-06,
+      "loss": 0.5865,
+      "step": 6028
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 7.174751306094921e-06,
+      "loss": 0.45,
+      "step": 6029
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 7.16758428609926e-06,
+      "loss": 0.7976,
+      "step": 6030
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 7.160420248472169e-06,
+      "loss": 0.8116,
+      "step": 6031
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 7.153259194411787e-06,
+      "loss": 0.4867,
+      "step": 6032
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 7.146101125115765e-06,
+      "loss": 0.6152,
+      "step": 6033
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 7.138946041781247e-06,
+      "loss": 0.6138,
+      "step": 6034
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 7.1317939456048785e-06,
+      "loss": 0.7127,
+      "step": 6035
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 7.124644837782807e-06,
+      "loss": 0.5336,
+      "step": 6036
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 7.117498719510676e-06,
+      "loss": 0.5902,
+      "step": 6037
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 7.110355591983639e-06,
+      "loss": 0.9201,
+      "step": 6038
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 7.103215456396336e-06,
+      "loss": 0.6918,
+      "step": 6039
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 7.09607831394292e-06,
+      "loss": 0.6714,
+      "step": 6040
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 7.088944165817033e-06,
+      "loss": 0.6136,
+      "step": 6041
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 7.0818130132118186e-06,
+      "loss": 0.4044,
+      "step": 6042
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 7.074684857319927e-06,
+      "loss": 0.406,
+      "step": 6043
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 7.067559699333495e-06,
+      "loss": 0.7049,
+      "step": 6044
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 7.060437540444165e-06,
+      "loss": 0.7172,
+      "step": 6045
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 7.05331838184308e-06,
+      "loss": 0.6254,
+      "step": 6046
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 7.046202224720877e-06,
+      "loss": 0.6773,
+      "step": 6047
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 7.03908907026769e-06,
+      "loss": 0.4331,
+      "step": 6048
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 7.031978919673154e-06,
+      "loss": 0.7762,
+      "step": 6049
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 7.024871774126404e-06,
+      "loss": 0.4648,
+      "step": 6050
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 7.017767634816064e-06,
+      "loss": 0.4736,
+      "step": 6051
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 7.0106665029302636e-06,
+      "loss": 0.5787,
+      "step": 6052
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 7.003568379656625e-06,
+      "loss": 0.555,
+      "step": 6053
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 6.996473266182269e-06,
+      "loss": 0.6459,
+      "step": 6054
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 6.989381163693811e-06,
+      "loss": 0.6596,
+      "step": 6055
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 6.982292073377366e-06,
+      "loss": 0.5509,
+      "step": 6056
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 6.975205996418543e-06,
+      "loss": 0.538,
+      "step": 6057
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 6.968122934002447e-06,
+      "loss": 0.4876,
+      "step": 6058
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 6.961042887313676e-06,
+      "loss": 0.4422,
+      "step": 6059
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 6.953965857536327e-06,
+      "loss": 0.4802,
+      "step": 6060
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 6.946891845854006e-06,
+      "loss": 0.7101,
+      "step": 6061
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 6.939820853449786e-06,
+      "loss": 0.6751,
+      "step": 6062
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 6.932752881506252e-06,
+      "loss": 0.4818,
+      "step": 6063
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 6.925687931205485e-06,
+      "loss": 0.8154,
+      "step": 6064
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 6.918626003729045e-06,
+      "loss": 0.4761,
+      "step": 6065
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 6.91156710025802e-06,
+      "loss": 0.8006,
+      "step": 6066
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 6.904511221972964e-06,
+      "loss": 0.7922,
+      "step": 6067
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 6.897458370053925e-06,
+      "loss": 0.5987,
+      "step": 6068
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 6.890408545680452e-06,
+      "loss": 0.8352,
+      "step": 6069
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 6.883361750031589e-06,
+      "loss": 0.4255,
+      "step": 6070
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 6.876317984285866e-06,
+      "loss": 0.7374,
+      "step": 6071
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 6.869277249621325e-06,
+      "loss": 0.549,
+      "step": 6072
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 6.862239547215491e-06,
+      "loss": 0.6042,
+      "step": 6073
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 6.8552048782453636e-06,
+      "loss": 0.7274,
+      "step": 6074
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 6.848173243887457e-06,
+      "loss": 0.6162,
+      "step": 6075
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 6.841144645317762e-06,
+      "loss": 0.3607,
+      "step": 6076
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 6.834119083711793e-06,
+      "loss": 0.6075,
+      "step": 6077
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 6.827096560244519e-06,
+      "loss": 0.7994,
+      "step": 6078
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 6.820077076090428e-06,
+      "loss": 0.58,
+      "step": 6079
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 6.813060632423474e-06,
+      "loss": 0.5643,
+      "step": 6080
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 6.806047230417112e-06,
+      "loss": 0.677,
+      "step": 6081
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 6.799036871244313e-06,
+      "loss": 0.5981,
+      "step": 6082
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 6.7920295560775116e-06,
+      "loss": 0.7957,
+      "step": 6083
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 6.785025286088639e-06,
+      "loss": 0.5513,
+      "step": 6084
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 6.778024062449126e-06,
+      "loss": 0.68,
+      "step": 6085
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 6.771025886329871e-06,
+      "loss": 0.5799,
+      "step": 6086
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 6.764030758901283e-06,
+      "loss": 0.5037,
+      "step": 6087
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 6.757038681333263e-06,
+      "loss": 0.753,
+      "step": 6088
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 6.7500496547951984e-06,
+      "loss": 0.5208,
+      "step": 6089
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 6.743063680455958e-06,
+      "loss": 0.6894,
+      "step": 6090
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 6.736080759483915e-06,
+      "loss": 0.5532,
+      "step": 6091
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 6.729100893046897e-06,
+      "loss": 0.8407,
+      "step": 6092
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 6.7221240823122726e-06,
+      "loss": 0.458,
+      "step": 6093
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 6.715150328446865e-06,
+      "loss": 0.6538,
+      "step": 6094
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 6.70817963261699e-06,
+      "loss": 0.7157,
+      "step": 6095
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 6.701211995988463e-06,
+      "loss": 0.6611,
+      "step": 6096
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 6.694247419726582e-06,
+      "loss": 0.6348,
+      "step": 6097
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 6.687285904996116e-06,
+      "loss": 0.6043,
+      "step": 6098
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 6.680327452961355e-06,
+      "loss": 0.7072,
+      "step": 6099
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 6.673372064786054e-06,
+      "loss": 0.6492,
+      "step": 6100
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 6.666419741633464e-06,
+      "loss": 0.6677,
+      "step": 6101
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 6.659470484666319e-06,
+      "loss": 0.5701,
+      "step": 6102
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 6.652524295046842e-06,
+      "loss": 0.8267,
+      "step": 6103
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 6.645581173936744e-06,
+      "loss": 0.6994,
+      "step": 6104
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 6.638641122497219e-06,
+      "loss": 0.8915,
+      "step": 6105
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 6.631704141888956e-06,
+      "loss": 0.5135,
+      "step": 6106
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 6.624770233272118e-06,
+      "loss": 0.4891,
+      "step": 6107
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 6.6178393978063695e-06,
+      "loss": 0.6438,
+      "step": 6108
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 6.610911636650843e-06,
+      "loss": 0.3984,
+      "step": 6109
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 6.603986950964178e-06,
+      "loss": 0.5546,
+      "step": 6110
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 6.5970653419044765e-06,
+      "loss": 0.5374,
+      "step": 6111
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 6.590146810629347e-06,
+      "loss": 0.5893,
+      "step": 6112
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 6.583231358295866e-06,
+      "loss": 0.5109,
+      "step": 6113
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 6.576318986060612e-06,
+      "loss": 0.6715,
+      "step": 6114
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 6.5694096950796315e-06,
+      "loss": 0.505,
+      "step": 6115
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 6.562503486508464e-06,
+      "loss": 0.6629,
+      "step": 6116
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 6.55560036150214e-06,
+      "loss": 0.5587,
+      "step": 6117
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 6.548700321215159e-06,
+      "loss": 0.722,
+      "step": 6118
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 6.54180336680151e-06,
+      "loss": 0.6956,
+      "step": 6119
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 6.5349094994146935e-06,
+      "loss": 0.6124,
+      "step": 6120
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 6.52801872020764e-06,
+      "loss": 0.5591,
+      "step": 6121
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 6.521131030332805e-06,
+      "loss": 0.5475,
+      "step": 6122
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 6.514246430942111e-06,
+      "loss": 0.6996,
+      "step": 6123
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 6.507364923186965e-06,
+      "loss": 0.6298,
+      "step": 6124
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 6.500486508218271e-06,
+      "loss": 0.4937,
+      "step": 6125
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 6.493611187186405e-06,
+      "loss": 0.6311,
+      "step": 6126
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 6.486738961241209e-06,
+      "loss": 0.4047,
+      "step": 6127
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 6.4798698315320315e-06,
+      "loss": 0.9144,
+      "step": 6128
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 6.473003799207694e-06,
+      "loss": 0.6545,
+      "step": 6129
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 6.466140865416495e-06,
+      "loss": 0.7765,
+      "step": 6130
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 6.459281031306233e-06,
+      "loss": 0.7364,
+      "step": 6131
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 6.452424298024179e-06,
+      "loss": 0.6263,
+      "step": 6132
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 6.445570666717063e-06,
+      "loss": 0.6586,
+      "step": 6133
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 6.438720138531126e-06,
+      "loss": 0.5215,
+      "step": 6134
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 6.431872714612072e-06,
+      "loss": 0.6045,
+      "step": 6135
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 6.425028396105104e-06,
+      "loss": 0.625,
+      "step": 6136
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 6.418187184154892e-06,
+      "loss": 0.7283,
+      "step": 6137
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 6.411349079905593e-06,
+      "loss": 0.6443,
+      "step": 6138
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 6.404514084500829e-06,
+      "loss": 0.6848,
+      "step": 6139
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 6.39768219908371e-06,
+      "loss": 0.5404,
+      "step": 6140
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 6.390853424796847e-06,
+      "loss": 0.6217,
+      "step": 6141
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 6.384027762782305e-06,
+      "loss": 0.717,
+      "step": 6142
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 6.377205214181634e-06,
+      "loss": 0.6626,
+      "step": 6143
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 6.370385780135879e-06,
+      "loss": 0.6134,
+      "step": 6144
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 6.363569461785529e-06,
+      "loss": 0.5498,
+      "step": 6145
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 6.35675626027058e-06,
+      "loss": 0.5838,
+      "step": 6146
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 6.3499461767305115e-06,
+      "loss": 0.5248,
+      "step": 6147
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 6.343139212304267e-06,
+      "loss": 0.6092,
+      "step": 6148
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 6.336335368130267e-06,
+      "loss": 0.6562,
+      "step": 6149
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 6.3295346453464275e-06,
+      "loss": 0.7244,
+      "step": 6150
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 6.3227370450901015e-06,
+      "loss": 0.7459,
+      "step": 6151
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 6.315942568498178e-06,
+      "loss": 0.4148,
+      "step": 6152
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 6.309151216706985e-06,
+      "loss": 0.4621,
+      "step": 6153
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 6.302362990852331e-06,
+      "loss": 0.5702,
+      "step": 6154
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 6.295577892069515e-06,
+      "loss": 0.6874,
+      "step": 6155
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 6.288795921493309e-06,
+      "loss": 0.6523,
+      "step": 6156
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 6.282017080257938e-06,
+      "loss": 0.6269,
+      "step": 6157
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 6.275241369497142e-06,
+      "loss": 0.6754,
+      "step": 6158
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 6.268468790344117e-06,
+      "loss": 0.6677,
+      "step": 6159
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 6.2616993439315376e-06,
+      "loss": 0.4628,
+      "step": 6160
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 6.254933031391552e-06,
+      "loss": 0.5853,
+      "step": 6161
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 6.24816985385579e-06,
+      "loss": 0.839,
+      "step": 6162
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 6.2414098124553485e-06,
+      "loss": 0.6351,
+      "step": 6163
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 6.2346529083208125e-06,
+      "loss": 0.6731,
+      "step": 6164
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 6.22789914258223e-06,
+      "loss": 0.6466,
+      "step": 6165
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 6.22114851636913e-06,
+      "loss": 0.6475,
+      "step": 6166
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 6.214401030810518e-06,
+      "loss": 0.6164,
+      "step": 6167
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 6.207656687034868e-06,
+      "loss": 0.6277,
+      "step": 6168
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 6.200915486170139e-06,
+      "loss": 0.5865,
+      "step": 6169
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 6.194177429343753e-06,
+      "loss": 0.6407,
+      "step": 6170
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 6.187442517682612e-06,
+      "loss": 0.5595,
+      "step": 6171
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 6.180710752313093e-06,
+      "loss": 0.3948,
+      "step": 6172
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 6.173982134361042e-06,
+      "loss": 0.6988,
+      "step": 6173
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 6.167256664951782e-06,
+      "loss": 0.6068,
+      "step": 6174
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 6.160534345210109e-06,
+      "loss": 0.6684,
+      "step": 6175
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 6.153815176260294e-06,
+      "loss": 0.5172,
+      "step": 6176
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 6.14709915922608e-06,
+      "loss": 0.6618,
+      "step": 6177
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 6.14038629523068e-06,
+      "loss": 0.5354,
+      "step": 6178
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 6.13367658539678e-06,
+      "loss": 0.4829,
+      "step": 6179
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 6.126970030846541e-06,
+      "loss": 0.6169,
+      "step": 6180
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 6.120266632701599e-06,
+      "loss": 0.7387,
+      "step": 6181
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 6.113566392083056e-06,
+      "loss": 0.4952,
+      "step": 6182
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 6.106869310111482e-06,
+      "loss": 0.6761,
+      "step": 6183
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 6.100175387906945e-06,
+      "loss": 0.6037,
+      "step": 6184
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 6.093484626588946e-06,
+      "loss": 0.7901,
+      "step": 6185
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 6.086797027276484e-06,
+      "loss": 0.7826,
+      "step": 6186
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 6.080112591088016e-06,
+      "loss": 0.5154,
+      "step": 6187
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 6.0734313191414816e-06,
+      "loss": 0.5963,
+      "step": 6188
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 6.0667532125542715e-06,
+      "loss": 0.7253,
+      "step": 6189
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 6.060078272443289e-06,
+      "loss": 0.6895,
+      "step": 6190
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 6.053406499924854e-06,
+      "loss": 0.6174,
+      "step": 6191
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 6.04673789611479e-06,
+      "loss": 0.6978,
+      "step": 6192
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 6.040072462128379e-06,
+      "loss": 0.6519,
+      "step": 6193
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 6.033410199080375e-06,
+      "loss": 0.5424,
+      "step": 6194
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 6.026751108085013e-06,
+      "loss": 0.4314,
+      "step": 6195
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 6.020095190255989e-06,
+      "loss": 0.6931,
+      "step": 6196
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 6.013442446706455e-06,
+      "loss": 0.6188,
+      "step": 6197
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 6.006792878549047e-06,
+      "loss": 0.5372,
+      "step": 6198
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 6.000146486895861e-06,
+      "loss": 0.5807,
+      "step": 6199
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 5.9935032728584834e-06,
+      "loss": 0.5837,
+      "step": 6200
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 5.986863237547944e-06,
+      "loss": 0.4463,
+      "step": 6201
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 5.9802263820747576e-06,
+      "loss": 0.6242,
+      "step": 6202
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 5.973592707548889e-06,
+      "loss": 0.7543,
+      "step": 6203
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 5.966962215079786e-06,
+      "loss": 0.5912,
+      "step": 6204
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 5.960334905776355e-06,
+      "loss": 0.6463,
+      "step": 6205
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 5.953710780746987e-06,
+      "loss": 0.7708,
+      "step": 6206
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 5.947089841099523e-06,
+      "loss": 0.511,
+      "step": 6207
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 5.940472087941284e-06,
+      "loss": 0.6907,
+      "step": 6208
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 5.933857522379038e-06,
+      "loss": 0.4909,
+      "step": 6209
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 5.92724614551903e-06,
+      "loss": 0.5809,
+      "step": 6210
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 5.92063795846699e-06,
+      "loss": 0.6589,
+      "step": 6211
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 5.914032962328095e-06,
+      "loss": 0.5439,
+      "step": 6212
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 5.907431158206989e-06,
+      "loss": 0.5649,
+      "step": 6213
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 5.900832547207791e-06,
+      "loss": 0.4661,
+      "step": 6214
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 5.894237130434071e-06,
+      "loss": 0.5831,
+      "step": 6215
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 5.887644908988874e-06,
+      "loss": 0.508,
+      "step": 6216
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 5.881055883974723e-06,
+      "loss": 0.5446,
+      "step": 6217
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 5.8744700564935865e-06,
+      "loss": 0.634,
+      "step": 6218
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 5.867887427646907e-06,
+      "loss": 0.7509,
+      "step": 6219
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 5.861307998535601e-06,
+      "loss": 0.5997,
+      "step": 6220
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 5.854731770260016e-06,
+      "loss": 0.5253,
+      "step": 6221
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 5.84815874392001e-06,
+      "loss": 0.8058,
+      "step": 6222
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 5.841588920614874e-06,
+      "loss": 0.6083,
+      "step": 6223
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 5.83502230144338e-06,
+      "loss": 0.6505,
+      "step": 6224
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 5.828458887503749e-06,
+      "loss": 0.6674,
+      "step": 6225
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 5.821898679893681e-06,
+      "loss": 0.5581,
+      "step": 6226
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 5.815341679710326e-06,
+      "loss": 0.5877,
+      "step": 6227
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 5.808787888050312e-06,
+      "loss": 0.7262,
+      "step": 6228
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 5.802237306009717e-06,
+      "loss": 0.6614,
+      "step": 6229
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 5.79568993468409e-06,
+      "loss": 0.5862,
+      "step": 6230
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 5.789145775168442e-06,
+      "loss": 0.5044,
+      "step": 6231
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 5.782604828557245e-06,
+      "loss": 0.7076,
+      "step": 6232
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 5.776067095944434e-06,
+      "loss": 0.6115,
+      "step": 6233
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 5.769532578423406e-06,
+      "loss": 0.6245,
+      "step": 6234
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 5.763001277087024e-06,
+      "loss": 0.5103,
+      "step": 6235
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 5.756473193027612e-06,
+      "loss": 0.3879,
+      "step": 6236
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 5.749948327336948e-06,
+      "loss": 0.7455,
+      "step": 6237
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 5.7434266811062845e-06,
+      "loss": 0.5196,
+      "step": 6238
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 5.736908255426324e-06,
+      "loss": 0.611,
+      "step": 6239
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 5.7303930513872394e-06,
+      "loss": 0.677,
+      "step": 6240
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 5.723881070078657e-06,
+      "loss": 0.5453,
+      "step": 6241
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 5.717372312589675e-06,
+      "loss": 0.6188,
+      "step": 6242
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 5.710866780008839e-06,
+      "loss": 0.529,
+      "step": 6243
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 5.704364473424162e-06,
+      "loss": 0.3134,
+      "step": 6244
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 5.697865393923121e-06,
+      "loss": 0.8244,
+      "step": 6245
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 5.691369542592648e-06,
+      "loss": 0.7329,
+      "step": 6246
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 5.684876920519136e-06,
+      "loss": 0.6318,
+      "step": 6247
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 5.67838752878844e-06,
+      "loss": 0.8167,
+      "step": 6248
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 5.671901368485871e-06,
+      "loss": 0.7819,
+      "step": 6249
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 5.665418440696202e-06,
+      "loss": 0.5321,
+      "step": 6250
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 5.658938746503667e-06,
+      "loss": 0.6368,
+      "step": 6251
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 5.652462286991958e-06,
+      "loss": 0.6745,
+      "step": 6252
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 5.645989063244217e-06,
+      "loss": 0.6649,
+      "step": 6253
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 5.639519076343072e-06,
+      "loss": 0.8183,
+      "step": 6254
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 5.633052327370575e-06,
+      "loss": 0.5068,
+      "step": 6255
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 5.626588817408255e-06,
+      "loss": 0.572,
+      "step": 6256
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 5.620128547537101e-06,
+      "loss": 0.5694,
+      "step": 6257
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 5.613671518837543e-06,
+      "loss": 0.6294,
+      "step": 6258
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 5.607217732389503e-06,
+      "loss": 0.5883,
+      "step": 6259
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 5.600767189272332e-06,
+      "loss": 0.6936,
+      "step": 6260
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 5.59431989056484e-06,
+      "loss": 0.6085,
+      "step": 6261
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 5.5878758373452996e-06,
+      "loss": 0.6487,
+      "step": 6262
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 5.5814350306914455e-06,
+      "loss": 0.5289,
+      "step": 6263
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 5.5749974716804595e-06,
+      "loss": 0.5074,
+      "step": 6264
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 5.568563161388998e-06,
+      "loss": 0.3749,
+      "step": 6265
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 5.562132100893164e-06,
+      "loss": 0.6909,
+      "step": 6266
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 5.555704291268499e-06,
+      "loss": 0.6731,
+      "step": 6267
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 5.549279733590027e-06,
+      "loss": 0.7853,
+      "step": 6268
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 5.5428584289322075e-06,
+      "loss": 0.7984,
+      "step": 6269
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 5.536440378368982e-06,
+      "loss": 0.7617,
+      "step": 6270
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 5.530025582973727e-06,
+      "loss": 0.7525,
+      "step": 6271
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 5.523614043819289e-06,
+      "loss": 0.8414,
+      "step": 6272
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 5.51720576197794e-06,
+      "loss": 0.5465,
+      "step": 6273
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 5.510800738521441e-06,
+      "loss": 0.6732,
+      "step": 6274
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 5.5043989745209855e-06,
+      "loss": 0.346,
+      "step": 6275
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 5.498000471047246e-06,
+      "loss": 0.4741,
+      "step": 6276
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 5.491605229170329e-06,
+      "loss": 0.7187,
+      "step": 6277
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 5.485213249959808e-06,
+      "loss": 0.6587,
+      "step": 6278
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 5.478824534484692e-06,
+      "loss": 0.6335,
+      "step": 6279
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 5.472439083813455e-06,
+      "loss": 0.5624,
+      "step": 6280
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 5.466056899014041e-06,
+      "loss": 0.7793,
+      "step": 6281
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 5.459677981153827e-06,
+      "loss": 0.5761,
+      "step": 6282
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 5.453302331299651e-06,
+      "loss": 0.6054,
+      "step": 6283
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 5.44692995051781e-06,
+      "loss": 0.5512,
+      "step": 6284
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 5.440560839874026e-06,
+      "loss": 0.7817,
+      "step": 6285
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 5.4341950004335175e-06,
+      "loss": 0.6853,
+      "step": 6286
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 5.4278324332609295e-06,
+      "loss": 0.6464,
+      "step": 6287
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 5.42147313942036e-06,
+      "loss": 0.6013,
+      "step": 6288
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 5.415117119975369e-06,
+      "loss": 0.6271,
+      "step": 6289
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 5.4087643759889675e-06,
+      "loss": 0.753,
+      "step": 6290
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 5.402414908523598e-06,
+      "loss": 0.535,
+      "step": 6291
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 5.39606871864119e-06,
+      "loss": 0.659,
+      "step": 6292
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 5.389725807403101e-06,
+      "loss": 0.6585,
+      "step": 6293
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 5.383386175870145e-06,
+      "loss": 0.6034,
+      "step": 6294
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 5.377049825102593e-06,
+      "loss": 0.5651,
+      "step": 6295
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 5.370716756160157e-06,
+      "loss": 0.62,
+      "step": 6296
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 5.364386970102011e-06,
+      "loss": 0.6207,
+      "step": 6297
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 5.358060467986775e-06,
+      "loss": 0.6553,
+      "step": 6298
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 5.351737250872515e-06,
+      "loss": 0.9014,
+      "step": 6299
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 5.345417319816759e-06,
+      "loss": 0.5287,
+      "step": 6300
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 5.339100675876477e-06,
+      "loss": 0.5847,
+      "step": 6301
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 5.332787320108087e-06,
+      "loss": 0.6258,
+      "step": 6302
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 5.326477253567469e-06,
+      "loss": 0.6399,
+      "step": 6303
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 5.3201704773099385e-06,
+      "loss": 0.6972,
+      "step": 6304
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 5.313866992390268e-06,
+      "loss": 0.6851,
+      "step": 6305
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 5.307566799862685e-06,
+      "loss": 0.519,
+      "step": 6306
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 5.3012699007808535e-06,
+      "loss": 0.632,
+      "step": 6307
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 5.294976296197895e-06,
+      "loss": 0.5038,
+      "step": 6308
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 5.288685987166381e-06,
+      "loss": 0.6915,
+      "step": 6309
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 5.2823989747383285e-06,
+      "loss": 0.5501,
+      "step": 6310
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 5.276115259965203e-06,
+      "loss": 0.5418,
+      "step": 6311
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 5.2698348438979195e-06,
+      "loss": 0.5189,
+      "step": 6312
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 5.26355772758684e-06,
+      "loss": 0.5809,
+      "step": 6313
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 5.257283912081782e-06,
+      "loss": 0.4401,
+      "step": 6314
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 5.251013398431998e-06,
+      "loss": 0.5457,
+      "step": 6315
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 5.244746187686197e-06,
+      "loss": 0.693,
+      "step": 6316
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 5.238482280892531e-06,
+      "loss": 0.5229,
+      "step": 6317
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 5.2322216790986175e-06,
+      "loss": 0.649,
+      "step": 6318
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 5.2259643833514896e-06,
+      "loss": 0.6361,
+      "step": 6319
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 5.219710394697646e-06,
+      "loss": 0.5639,
+      "step": 6320
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 5.213459714183033e-06,
+      "loss": 0.5003,
+      "step": 6321
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 5.2072123428530435e-06,
+      "loss": 0.668,
+      "step": 6322
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 5.200968281752503e-06,
+      "loss": 0.5197,
+      "step": 6323
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 5.194727531925717e-06,
+      "loss": 0.7233,
+      "step": 6324
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 5.188490094416396e-06,
+      "loss": 0.4293,
+      "step": 6325
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 5.182255970267719e-06,
+      "loss": 0.7273,
+      "step": 6326
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 5.1760251605223075e-06,
+      "loss": 0.5107,
+      "step": 6327
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 5.169797666222226e-06,
+      "loss": 0.444,
+      "step": 6328
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 5.163573488408996e-06,
+      "loss": 0.5792,
+      "step": 6329
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 5.157352628123577e-06,
+      "loss": 0.7206,
+      "step": 6330
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 5.151135086406356e-06,
+      "loss": 0.6314,
+      "step": 6331
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 5.1449208642971925e-06,
+      "loss": 0.7665,
+      "step": 6332
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 5.138709962835375e-06,
+      "loss": 0.4977,
+      "step": 6333
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 5.132502383059637e-06,
+      "loss": 0.6623,
+      "step": 6334
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.126298126008175e-06,
+      "loss": 0.565,
+      "step": 6335
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.1200971927186095e-06,
+      "loss": 0.4686,
+      "step": 6336
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 5.113899584228002e-06,
+      "loss": 0.3614,
+      "step": 6337
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.107705301572871e-06,
+      "loss": 0.4838,
+      "step": 6338
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.101514345789171e-06,
+      "loss": 0.6971,
+      "step": 6339
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.095326717912313e-06,
+      "loss": 0.4958,
+      "step": 6340
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 5.0891424189771385e-06,
+      "loss": 0.5949,
+      "step": 6341
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 5.082961450017943e-06,
+      "loss": 0.7202,
+      "step": 6342
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 5.076783812068439e-06,
+      "loss": 0.6199,
+      "step": 6343
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 5.0706095061618085e-06,
+      "loss": 0.6296,
+      "step": 6344
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 5.064438533330673e-06,
+      "loss": 0.5918,
+      "step": 6345
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 5.058270894607095e-06,
+      "loss": 0.6863,
+      "step": 6346
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 5.052106591022571e-06,
+      "loss": 0.6568,
+      "step": 6347
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 5.0459456236080505e-06,
+      "loss": 0.562,
+      "step": 6348
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 5.0397879933939105e-06,
+      "loss": 0.5974,
+      "step": 6349
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 5.033633701409976e-06,
+      "loss": 0.5677,
+      "step": 6350
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 5.027482748685533e-06,
+      "loss": 0.5835,
+      "step": 6351
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 5.0213351362492836e-06,
+      "loss": 0.6043,
+      "step": 6352
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 5.015190865129382e-06,
+      "loss": 0.6464,
+      "step": 6353
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 5.009049936353427e-06,
+      "loss": 0.76,
+      "step": 6354
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 5.002912350948436e-06,
+      "loss": 0.6035,
+      "step": 6355
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.9967781099409e-06,
+      "loss": 0.5967,
+      "step": 6356
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.990647214356736e-06,
+      "loss": 0.6326,
+      "step": 6357
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.984519665221296e-06,
+      "loss": 0.592,
+      "step": 6358
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.978395463559377e-06,
+      "loss": 0.578,
+      "step": 6359
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.972274610395217e-06,
+      "loss": 0.5582,
+      "step": 6360
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.966157106752495e-06,
+      "loss": 0.5414,
+      "step": 6361
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.9600429536543255e-06,
+      "loss": 0.6428,
+      "step": 6362
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.953932152123269e-06,
+      "loss": 0.5058,
+      "step": 6363
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.94782470318132e-06,
+      "loss": 0.54,
+      "step": 6364
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.9417206078499115e-06,
+      "loss": 0.7846,
+      "step": 6365
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.935619867149921e-06,
+      "loss": 0.8408,
+      "step": 6366
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.929522482101662e-06,
+      "loss": 0.8202,
+      "step": 6367
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.923428453724887e-06,
+      "loss": 0.587,
+      "step": 6368
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.917337783038786e-06,
+      "loss": 0.6536,
+      "step": 6369
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.91125047106199e-06,
+      "loss": 0.569,
+      "step": 6370
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.9051665188125654e-06,
+      "loss": 0.8437,
+      "step": 6371
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.899085927308017e-06,
+      "loss": 0.595,
+      "step": 6372
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.89300869756529e-06,
+      "loss": 0.5607,
+      "step": 6373
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.8869348306007675e-06,
+      "loss": 0.6972,
+      "step": 6374
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.880864327430268e-06,
+      "loss": 0.5682,
+      "step": 6375
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.8747971890690384e-06,
+      "loss": 0.7222,
+      "step": 6376
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.868733416531796e-06,
+      "loss": 0.6898,
+      "step": 6377
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.862673010832652e-06,
+      "loss": 0.5792,
+      "step": 6378
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.856615972985179e-06,
+      "loss": 0.5021,
+      "step": 6379
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.850562304002382e-06,
+      "loss": 0.7035,
+      "step": 6380
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.844512004896701e-06,
+      "loss": 0.5802,
+      "step": 6381
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.83846507668001e-06,
+      "loss": 0.5278,
+      "step": 6382
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.8324215203636395e-06,
+      "loss": 0.6872,
+      "step": 6383
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.826381336958321e-06,
+      "loss": 0.4565,
+      "step": 6384
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.820344527474249e-06,
+      "loss": 0.5496,
+      "step": 6385
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.814311092921042e-06,
+      "loss": 0.4598,
+      "step": 6386
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.8082810343077514e-06,
+      "loss": 0.4499,
+      "step": 6387
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.802254352642882e-06,
+      "loss": 0.8521,
+      "step": 6388
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.796231048934366e-06,
+      "loss": 0.6252,
+      "step": 6389
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.790211124189544e-06,
+      "loss": 0.4932,
+      "step": 6390
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.78419457941523e-06,
+      "loss": 0.4258,
+      "step": 6391
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.7781814156176525e-06,
+      "loss": 0.5805,
+      "step": 6392
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.772171633802469e-06,
+      "loss": 0.7138,
+      "step": 6393
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.766165234974798e-06,
+      "loss": 0.7128,
+      "step": 6394
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.760162220139172e-06,
+      "loss": 0.5149,
+      "step": 6395
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.754162590299549e-06,
+      "loss": 0.6089,
+      "step": 6396
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.74816634645934e-06,
+      "loss": 0.665,
+      "step": 6397
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.742173489621374e-06,
+      "loss": 0.4819,
+      "step": 6398
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.736184020787937e-06,
+      "loss": 0.5067,
+      "step": 6399
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.730197940960723e-06,
+      "loss": 0.4682,
+      "step": 6400
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.7242152511408785e-06,
+      "loss": 0.8045,
+      "step": 6401
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.718235952328959e-06,
+      "loss": 0.6867,
+      "step": 6402
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.7122600455249686e-06,
+      "loss": 0.6248,
+      "step": 6403
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.706287531728357e-06,
+      "loss": 0.5821,
+      "step": 6404
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.7003184119379845e-06,
+      "loss": 0.8067,
+      "step": 6405
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.694352687152154e-06,
+      "loss": 0.6206,
+      "step": 6406
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.688390358368605e-06,
+      "loss": 0.6064,
+      "step": 6407
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.6824314265844855e-06,
+      "loss": 0.7489,
+      "step": 6408
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.6764758927963976e-06,
+      "loss": 0.906,
+      "step": 6409
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.670523758000381e-06,
+      "loss": 0.3437,
+      "step": 6410
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.664575023191886e-06,
+      "loss": 0.6596,
+      "step": 6411
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.658629689365809e-06,
+      "loss": 0.6303,
+      "step": 6412
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.652687757516477e-06,
+      "loss": 0.429,
+      "step": 6413
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.646749228637625e-06,
+      "loss": 0.4444,
+      "step": 6414
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.64081410372246e-06,
+      "loss": 0.7506,
+      "step": 6415
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.634882383763586e-06,
+      "loss": 0.6371,
+      "step": 6416
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.62895406975305e-06,
+      "loss": 0.6958,
+      "step": 6417
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.623029162682329e-06,
+      "loss": 0.705,
+      "step": 6418
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.617107663542328e-06,
+      "loss": 0.6288,
+      "step": 6419
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.611189573323388e-06,
+      "loss": 0.6748,
+      "step": 6420
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.605274893015271e-06,
+      "loss": 0.6621,
+      "step": 6421
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.5993636236071755e-06,
+      "loss": 0.6367,
+      "step": 6422
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.5934557660877255e-06,
+      "loss": 0.7563,
+      "step": 6423
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.587551321444978e-06,
+      "loss": 0.571,
+      "step": 6424
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.581650290666415e-06,
+      "loss": 0.4439,
+      "step": 6425
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.575752674738951e-06,
+      "loss": 0.7922,
+      "step": 6426
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.569858474648927e-06,
+      "loss": 0.8046,
+      "step": 6427
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.563967691382117e-06,
+      "loss": 0.4856,
+      "step": 6428
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.558080325923719e-06,
+      "loss": 0.5928,
+      "step": 6429
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.5521963792583585e-06,
+      "loss": 0.5988,
+      "step": 6430
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.546315852370095e-06,
+      "loss": 0.7,
+      "step": 6431
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.540438746242412e-06,
+      "loss": 0.528,
+      "step": 6432
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.5345650618582205e-06,
+      "loss": 0.577,
+      "step": 6433
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.528694800199859e-06,
+      "loss": 0.9133,
+      "step": 6434
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.5228279622490985e-06,
+      "loss": 0.6815,
+      "step": 6435
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.5169645489871295e-06,
+      "loss": 0.6688,
+      "step": 6436
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.511104561394577e-06,
+      "loss": 0.6163,
+      "step": 6437
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.505248000451492e-06,
+      "loss": 0.4003,
+      "step": 6438
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.499394867137346e-06,
+      "loss": 0.3983,
+      "step": 6439
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.493545162431043e-06,
+      "loss": 0.6828,
+      "step": 6440
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.487698887310915e-06,
+      "loss": 0.705,
+      "step": 6441
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.481856042754712e-06,
+      "loss": 0.611,
+      "step": 6442
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.476016629739624e-06,
+      "loss": 0.6831,
+      "step": 6443
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.4701806492422535e-06,
+      "loss": 0.4307,
+      "step": 6444
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.464348102238633e-06,
+      "loss": 0.7747,
+      "step": 6445
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.458518989704221e-06,
+      "loss": 0.4537,
+      "step": 6446
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.45269331261392e-06,
+      "loss": 0.4698,
+      "step": 6447
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.446871071942025e-06,
+      "loss": 0.5734,
+      "step": 6448
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.441052268662269e-06,
+      "loss": 0.5632,
+      "step": 6449
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.435236903747822e-06,
+      "loss": 0.6447,
+      "step": 6450
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.429424978171268e-06,
+      "loss": 0.6541,
+      "step": 6451
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.423616492904611e-06,
+      "loss": 0.552,
+      "step": 6452
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.417811448919306e-06,
+      "loss": 0.5214,
+      "step": 6453
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.412009847186194e-06,
+      "loss": 0.4713,
+      "step": 6454
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.406211688675566e-06,
+      "loss": 0.4242,
+      "step": 6455
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.40041697435713e-06,
+      "loss": 0.4701,
+      "step": 6456
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.394625705200011e-06,
+      "loss": 0.7006,
+      "step": 6457
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.388837882172783e-06,
+      "loss": 0.6701,
+      "step": 6458
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.3830535062434205e-06,
+      "loss": 0.4817,
+      "step": 6459
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.377272578379316e-06,
+      "loss": 0.7939,
+      "step": 6460
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.3714950995473015e-06,
+      "loss": 0.4635,
+      "step": 6461
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.365721070713622e-06,
+      "loss": 0.7861,
+      "step": 6462
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.359950492843964e-06,
+      "loss": 0.7788,
+      "step": 6463
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.354183366903416e-06,
+      "loss": 0.5987,
+      "step": 6464
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.3484196938565035e-06,
+      "loss": 0.8171,
+      "step": 6465
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.342659474667155e-06,
+      "loss": 0.4195,
+      "step": 6466
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.336902710298737e-06,
+      "loss": 0.7327,
+      "step": 6467
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.331149401714035e-06,
+      "loss": 0.5366,
+      "step": 6468
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.325399549875264e-06,
+      "loss": 0.5867,
+      "step": 6469
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.31965315574405e-06,
+      "loss": 0.7175,
+      "step": 6470
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.3139102202814476e-06,
+      "loss": 0.6061,
+      "step": 6471
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.308170744447917e-06,
+      "loss": 0.3446,
+      "step": 6472
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.302434729203359e-06,
+      "loss": 0.5968,
+      "step": 6473
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.296702175507092e-06,
+      "loss": 0.7992,
+      "step": 6474
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.290973084317854e-06,
+      "loss": 0.5625,
+      "step": 6475
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.285247456593799e-06,
+      "loss": 0.5622,
+      "step": 6476
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.2795252932925106e-06,
+      "loss": 0.6664,
+      "step": 6477
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.273806595370972e-06,
+      "loss": 0.6009,
+      "step": 6478
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.268091363785617e-06,
+      "loss": 0.7819,
+      "step": 6479
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.262379599492283e-06,
+      "loss": 0.5475,
+      "step": 6480
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.256671303446227e-06,
+      "loss": 0.6811,
+      "step": 6481
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.250966476602131e-06,
+      "loss": 0.5619,
+      "step": 6482
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.245265119914096e-06,
+      "loss": 0.4889,
+      "step": 6483
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.239567234335625e-06,
+      "loss": 0.7432,
+      "step": 6484
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.233872820819678e-06,
+      "loss": 0.5063,
+      "step": 6485
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.228181880318602e-06,
+      "loss": 0.6779,
+      "step": 6486
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.222494413784178e-06,
+      "loss": 0.5437,
+      "step": 6487
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.216810422167597e-06,
+      "loss": 0.8235,
+      "step": 6488
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.211129906419478e-06,
+      "loss": 0.4435,
+      "step": 6489
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.205452867489854e-06,
+      "loss": 0.6448,
+      "step": 6490
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.199779306328172e-06,
+      "loss": 0.7018,
+      "step": 6491
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.194109223883311e-06,
+      "loss": 0.6558,
+      "step": 6492
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.18844262110355e-06,
+      "loss": 0.6289,
+      "step": 6493
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.182779498936604e-06,
+      "loss": 0.5806,
+      "step": 6494
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.177119858329592e-06,
+      "loss": 0.7029,
+      "step": 6495
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.171463700229061e-06,
+      "loss": 0.6405,
+      "step": 6496
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 4.165811025580965e-06,
+      "loss": 0.6574,
+      "step": 6497
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 4.1601618353306835e-06,
+      "loss": 0.5512,
+      "step": 6498
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 4.154516130423014e-06,
+      "loss": 0.8079,
+      "step": 6499
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 4.148873911802165e-06,
+      "loss": 0.6864,
+      "step": 6500
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 4.143235180411767e-06,
+      "loss": 0.8736,
+      "step": 6501
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 4.137599937194861e-06,
+      "loss": 0.504,
+      "step": 6502
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 4.131968183093912e-06,
+      "loss": 0.4796,
+      "step": 6503
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 4.1263399190508e-06,
+      "loss": 0.6266,
+      "step": 6504
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.120715146006818e-06,
+      "loss": 0.3843,
+      "step": 6505
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.115093864902675e-06,
+      "loss": 0.5386,
+      "step": 6506
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.109476076678501e-06,
+      "loss": 0.5254,
+      "step": 6507
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.103861782273835e-06,
+      "loss": 0.5738,
+      "step": 6508
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 4.098250982627638e-06,
+      "loss": 0.506,
+      "step": 6509
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 4.092643678678284e-06,
+      "loss": 0.6612,
+      "step": 6510
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 4.087039871363563e-06,
+      "loss": 0.5001,
+      "step": 6511
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 4.0814395616206755e-06,
+      "loss": 0.6418,
+      "step": 6512
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.075842750386244e-06,
+      "loss": 0.5402,
+      "step": 6513
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.070249438596302e-06,
+      "loss": 0.7144,
+      "step": 6514
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.0646596271862985e-06,
+      "loss": 0.6889,
+      "step": 6515
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.059073317091092e-06,
+      "loss": 0.6092,
+      "step": 6516
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 4.0534905092449785e-06,
+      "loss": 0.553,
+      "step": 6517
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 4.047911204581631e-06,
+      "loss": 0.5538,
+      "step": 6518
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 4.042335404034161e-06,
+      "loss": 0.6814,
+      "step": 6519
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 4.036763108535097e-06,
+      "loss": 0.616,
+      "step": 6520
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 4.031194319016354e-06,
+      "loss": 0.4847,
+      "step": 6521
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 4.025629036409303e-06,
+      "loss": 0.6195,
+      "step": 6522
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 4.020067261644703e-06,
+      "loss": 0.3893,
+      "step": 6523
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 4.014508995652713e-06,
+      "loss": 0.9007,
+      "step": 6524
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 4.008954239362928e-06,
+      "loss": 0.6523,
+      "step": 6525
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 4.003402993704353e-06,
+      "loss": 0.77,
+      "step": 6526
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.9978552596053904e-06,
+      "loss": 0.7332,
+      "step": 6527
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.992311037993881e-06,
+      "loss": 0.6198,
+      "step": 6528
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.9867703297970656e-06,
+      "loss": 0.6461,
+      "step": 6529
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.9812331359415785e-06,
+      "loss": 0.5095,
+      "step": 6530
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.9756994573534964e-06,
+      "loss": 0.5978,
+      "step": 6531
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.970169294958284e-06,
+      "loss": 0.6132,
+      "step": 6532
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.964642649680844e-06,
+      "loss": 0.7221,
+      "step": 6533
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.9591195224454676e-06,
+      "loss": 0.651,
+      "step": 6534
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.95359991417587e-06,
+      "loss": 0.6787,
+      "step": 6535
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.948083825795168e-06,
+      "loss": 0.5324,
+      "step": 6536
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.942571258225891e-06,
+      "loss": 0.6156,
+      "step": 6537
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.93706221239e-06,
+      "loss": 0.7088,
+      "step": 6538
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.93155668920884e-06,
+      "loss": 0.6496,
+      "step": 6539
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.926054689603181e-06,
+      "loss": 0.6001,
+      "step": 6540
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.920556214493204e-06,
+      "loss": 0.5366,
+      "step": 6541
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.915061264798489e-06,
+      "loss": 0.5748,
+      "step": 6542
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.909569841438032e-06,
+      "loss": 0.5239,
+      "step": 6543
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.904081945330257e-06,
+      "loss": 0.6101,
+      "step": 6544
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.8985975773929736e-06,
+      "loss": 0.6467,
+      "step": 6545
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.893116738543409e-06,
+      "loss": 0.6992,
+      "step": 6546
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.887639429698214e-06,
+      "loss": 0.7286,
+      "step": 6547
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.882165651773415e-06,
+      "loss": 0.3978,
+      "step": 6548
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.8766954056844855e-06,
+      "loss": 0.4544,
+      "step": 6549
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.871228692346293e-06,
+      "loss": 0.561,
+      "step": 6550
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.865765512673109e-06,
+      "loss": 0.6774,
+      "step": 6551
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.860305867578618e-06,
+      "loss": 0.64,
+      "step": 6552
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.854849757975917e-06,
+      "loss": 0.6175,
+      "step": 6553
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.849397184777506e-06,
+      "loss": 0.667,
+      "step": 6554
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.8439481488953014e-06,
+      "loss": 0.6568,
+      "step": 6555
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.838502651240616e-06,
+      "loss": 0.451,
+      "step": 6556
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.833060692724186e-06,
+      "loss": 0.5752,
+      "step": 6557
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.82762227425614e-06,
+      "loss": 0.8309,
+      "step": 6558
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.822187396746027e-06,
+      "loss": 0.63,
+      "step": 6559
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.8167560611027944e-06,
+      "loss": 0.6719,
+      "step": 6560
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.8113282682348067e-06,
+      "loss": 0.628,
+      "step": 6561
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.8059040190498273e-06,
+      "loss": 0.6231,
+      "step": 6562
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.8004833144550334e-06,
+      "loss": 0.607,
+      "step": 6563
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.795066155357002e-06,
+      "loss": 0.6064,
+      "step": 6564
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.7896525426617287e-06,
+      "loss": 0.5751,
+      "step": 6565
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.784242477274602e-06,
+      "loss": 0.6387,
+      "step": 6566
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.778835960100427e-06,
+      "loss": 0.5488,
+      "step": 6567
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.773432992043413e-06,
+      "loss": 0.3949,
+      "step": 6568
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.7680335740071767e-06,
+      "loss": 0.6973,
+      "step": 6569
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.7626377068947365e-06,
+      "loss": 0.6001,
+      "step": 6570
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.7572453916085214e-06,
+      "loss": 0.6599,
+      "step": 6571
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.7518566290503626e-06,
+      "loss": 0.5138,
+      "step": 6572
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.7464714201215028e-06,
+      "loss": 0.6504,
+      "step": 6573
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.7410897657225887e-06,
+      "loss": 0.5139,
+      "step": 6574
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.735711666753658e-06,
+      "loss": 0.4636,
+      "step": 6575
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.7303371241141916e-06,
+      "loss": 0.607,
+      "step": 6576
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.724966138703029e-06,
+      "loss": 0.7327,
+      "step": 6577
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.7195987114184453e-06,
+      "loss": 0.4925,
+      "step": 6578
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.714234843158107e-06,
+      "loss": 0.658,
+      "step": 6579
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.70887453481909e-06,
+      "loss": 0.604,
+      "step": 6580
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.7035177872978833e-06,
+      "loss": 0.7922,
+      "step": 6581
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.6981646014903696e-06,
+      "loss": 0.777,
+      "step": 6582
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.6928149782918337e-06,
+      "loss": 0.5194,
+      "step": 6583
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.6874689185969676e-06,
+      "loss": 0.5888,
+      "step": 6584
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.6821264232998726e-06,
+      "loss": 0.7092,
+      "step": 6585
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.6767874932940453e-06,
+      "loss": 0.6851,
+      "step": 6586
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.6714521294724024e-06,
+      "loss": 0.6063,
+      "step": 6587
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.6661203327272494e-06,
+      "loss": 0.684,
+      "step": 6588
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.6607921039502892e-06,
+      "loss": 0.6391,
+      "step": 6589
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.6554674440326426e-06,
+      "loss": 0.5439,
+      "step": 6590
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.650146353864825e-06,
+      "loss": 0.4117,
+      "step": 6591
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.6448288343367663e-06,
+      "loss": 0.675,
+      "step": 6592
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.6395148863377858e-06,
+      "loss": 0.6028,
+      "step": 6593
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.6342045107566168e-06,
+      "loss": 0.5245,
+      "step": 6594
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.6288977084813767e-06,
+      "loss": 0.5713,
+      "step": 6595
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.6235944803995973e-06,
+      "loss": 0.5707,
+      "step": 6596
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.6182948273982276e-06,
+      "loss": 0.4404,
+      "step": 6597
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.612998750363597e-06,
+      "loss": 0.6073,
+      "step": 6598
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.607706250181439e-06,
+      "loss": 0.748,
+      "step": 6599
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.6024173277369067e-06,
+      "loss": 0.5855,
+      "step": 6600
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.5971319839145257e-06,
+      "loss": 0.6361,
+      "step": 6601
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.591850219598239e-06,
+      "loss": 0.755,
+      "step": 6602
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.5865720356714066e-06,
+      "loss": 0.5061,
+      "step": 6603
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.5812974330167643e-06,
+      "loss": 0.6891,
+      "step": 6604
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.5760264125164616e-06,
+      "loss": 0.4893,
+      "step": 6605
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.5707589750520495e-06,
+      "loss": 0.5734,
+      "step": 6606
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.5654951215044615e-06,
+      "loss": 0.654,
+      "step": 6607
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.560234852754063e-06,
+      "loss": 0.5242,
+      "step": 6608
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.5549781696806005e-06,
+      "loss": 0.5633,
+      "step": 6609
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.549725073163218e-06,
+      "loss": 0.4616,
+      "step": 6610
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.544475564080471e-06,
+      "loss": 0.5811,
+      "step": 6611
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.5392296433103126e-06,
+      "loss": 0.4937,
+      "step": 6612
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.5339873117300774e-06,
+      "loss": 0.5443,
+      "step": 6613
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.528748570216531e-06,
+      "loss": 0.6242,
+      "step": 6614
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.523513419645816e-06,
+      "loss": 0.736,
+      "step": 6615
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.5182818608934826e-06,
+      "loss": 0.5835,
+      "step": 6616
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.513053894834481e-06,
+      "loss": 0.5056,
+      "step": 6617
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.5078295223431536e-06,
+      "loss": 0.8062,
+      "step": 6618
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.502608744293248e-06,
+      "loss": 0.595,
+      "step": 6619
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.4973915615579113e-06,
+      "loss": 0.6449,
+      "step": 6620
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.4921779750096863e-06,
+      "loss": 0.6586,
+      "step": 6621
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.4869679855205115e-06,
+      "loss": 0.5587,
+      "step": 6622
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.481761593961733e-06,
+      "loss": 0.5835,
+      "step": 6623
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.476558801204088e-06,
+      "loss": 0.7125,
+      "step": 6624
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.4713596081177134e-06,
+      "loss": 0.6623,
+      "step": 6625
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.466164015572143e-06,
+      "loss": 0.5735,
+      "step": 6626
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.4609720244363098e-06,
+      "loss": 0.4951,
+      "step": 6627
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.455783635578547e-06,
+      "loss": 0.6914,
+      "step": 6628
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.450598849866579e-06,
+      "loss": 0.6039,
+      "step": 6629
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.445417668167536e-06,
+      "loss": 0.618,
+      "step": 6630
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.440240091347935e-06,
+      "loss": 0.5131,
+      "step": 6631
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.4350661202737e-06,
+      "loss": 0.3645,
+      "step": 6632
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.4298957558101446e-06,
+      "loss": 0.7518,
+      "step": 6633
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.4247289988219853e-06,
+      "loss": 0.5093,
+      "step": 6634
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.419565850173331e-06,
+      "loss": 0.5796,
+      "step": 6635
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.414406310727691e-06,
+      "loss": 0.6587,
+      "step": 6636
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.409250381347964e-06,
+      "loss": 0.5355,
+      "step": 6637
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.404098062896452e-06,
+      "loss": 0.6092,
+      "step": 6638
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.3989493562348455e-06,
+      "loss": 0.516,
+      "step": 6639
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.393804262224251e-06,
+      "loss": 0.3046,
+      "step": 6640
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.388662781725141e-06,
+      "loss": 0.8114,
+      "step": 6641
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.3835249155974023e-06,
+      "loss": 0.7202,
+      "step": 6642
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.378390664700312e-06,
+      "loss": 0.6291,
+      "step": 6643
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.373260029892547e-06,
+      "loss": 0.8085,
+      "step": 6644
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.3681330120321706e-06,
+      "loss": 0.7799,
+      "step": 6645
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.363009611976664e-06,
+      "loss": 0.5187,
+      "step": 6646
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3578898305828665e-06,
+      "loss": 0.6239,
+      "step": 6647
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3527736687070376e-06,
+      "loss": 0.657,
+      "step": 6648
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3476611272048294e-06,
+      "loss": 0.6461,
+      "step": 6649
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3425522069312763e-06,
+      "loss": 0.8117,
+      "step": 6650
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.3374469087408293e-06,
+      "loss": 0.4951,
+      "step": 6651
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.332345233487319e-06,
+      "loss": 0.5626,
+      "step": 6652
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.32724718202396e-06,
+      "loss": 0.5641,
+      "step": 6653
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.3221527552033755e-06,
+      "loss": 0.6313,
+      "step": 6654
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.317061953877576e-06,
+      "loss": 0.573,
+      "step": 6655
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.3119747788979777e-06,
+      "loss": 0.6711,
+      "step": 6656
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.3068912311153804e-06,
+      "loss": 0.5989,
+      "step": 6657
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.3018113113799764e-06,
+      "loss": 0.6399,
+      "step": 6658
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.29673502054135e-06,
+      "loss": 0.5086,
+      "step": 6659
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.2916623594484803e-06,
+      "loss": 0.5007,
+      "step": 6660
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.2865933289497356e-06,
+      "loss": 0.3663,
+      "step": 6661
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.281527929892897e-06,
+      "loss": 0.68,
+      "step": 6662
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2764661631251166e-06,
+      "loss": 0.6646,
+      "step": 6663
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2714080294929477e-06,
+      "loss": 0.7824,
+      "step": 6664
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.266353529842328e-06,
+      "loss": 0.7823,
+      "step": 6665
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2613026650185897e-06,
+      "loss": 0.7471,
+      "step": 6666
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.256255435866473e-06,
+      "loss": 0.7505,
+      "step": 6667
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.2512118432300924e-06,
+      "loss": 0.8413,
+      "step": 6668
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.246171887952959e-06,
+      "loss": 0.5354,
+      "step": 6669
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.2411355708779846e-06,
+      "loss": 0.6726,
+      "step": 6670
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.236102892847448e-06,
+      "loss": 0.3366,
+      "step": 6671
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.23107385470304e-06,
+      "loss": 0.4693,
+      "step": 6672
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.226048457285846e-06,
+      "loss": 0.7129,
+      "step": 6673
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.2210267014363303e-06,
+      "loss": 0.666,
+      "step": 6674
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.216008587994354e-06,
+      "loss": 0.6479,
+      "step": 6675
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.210994117799174e-06,
+      "loss": 0.5508,
+      "step": 6676
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.2059832916894102e-06,
+      "loss": 0.7721,
+      "step": 6677
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.200976110503115e-06,
+      "loss": 0.5705,
+      "step": 6678
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.195972575077705e-06,
+      "loss": 0.5953,
+      "step": 6679
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.1909726862499906e-06,
+      "loss": 0.5459,
+      "step": 6680
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.185976444856173e-06,
+      "loss": 0.7658,
+      "step": 6681
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.18098385173185e-06,
+      "loss": 0.6725,
+      "step": 6682
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.175994907711999e-06,
+      "loss": 0.6431,
+      "step": 6683
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.1710096136309958e-06,
+      "loss": 0.5952,
+      "step": 6684
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.1660279703225966e-06,
+      "loss": 0.6128,
+      "step": 6685
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.1610499786199592e-06,
+      "loss": 0.7592,
+      "step": 6686
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.1560756393556183e-06,
+      "loss": 0.5284,
+      "step": 6687
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.1511049533615084e-06,
+      "loss": 0.6449,
+      "step": 6688
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.1461379214689446e-06,
+      "loss": 0.6551,
+      "step": 6689
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.141174544508635e-06,
+      "loss": 0.5941,
+      "step": 6690
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.1362148233106762e-06,
+      "loss": 0.555,
+      "step": 6691
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.1312587587045528e-06,
+      "loss": 0.6185,
+      "step": 6692
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.126306351519137e-06,
+      "loss": 0.6143,
+      "step": 6693
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.1213576025826947e-06,
+      "loss": 0.6577,
+      "step": 6694
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.116412512722869e-06,
+      "loss": 0.8798,
+      "step": 6695
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.111471082766701e-06,
+      "loss": 0.5248,
+      "step": 6696
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.106533313540616e-06,
+      "loss": 0.5817,
+      "step": 6697
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.1015992058704285e-06,
+      "loss": 0.6069,
+      "step": 6698
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.0966687605813366e-06,
+      "loss": 0.6274,
+      "step": 6699
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.0917419784979334e-06,
+      "loss": 0.693,
+      "step": 6700
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.0868188604441905e-06,
+      "loss": 0.699,
+      "step": 6701
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.0818994072434743e-06,
+      "loss": 0.5185,
+      "step": 6702
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.076983619718529e-06,
+      "loss": 0.6199,
+      "step": 6703
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.0720714986915007e-06,
+      "loss": 0.4961,
+      "step": 6704
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.067163044983906e-06,
+      "loss": 0.6807,
+      "step": 6705
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.0622582594166587e-06,
+      "loss": 0.5438,
+      "step": 6706
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.057357142810055e-06,
+      "loss": 0.5431,
+      "step": 6707
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.0524596959837754e-06,
+      "loss": 0.515,
+      "step": 6708
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.047565919756892e-06,
+      "loss": 0.5754,
+      "step": 6709
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.042675814947868e-06,
+      "loss": 0.4263,
+      "step": 6710
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.037789382374537e-06,
+      "loss": 0.5379,
+      "step": 6711
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.032906622854126e-06,
+      "loss": 0.6795,
+      "step": 6712
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.0280275372032484e-06,
+      "loss": 0.5162,
+      "step": 6713
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.023152126237902e-06,
+      "loss": 0.6436,
+      "step": 6714
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.0182803907734785e-06,
+      "loss": 0.6385,
+      "step": 6715
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.0134123316247514e-06,
+      "loss": 0.5518,
+      "step": 6716
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.0085479496058584e-06,
+      "loss": 0.4844,
+      "step": 6717
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.003687245530351e-06,
+      "loss": 0.6673,
+      "step": 6718
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 2.998830220211149e-06,
+      "loss": 0.5156,
+      "step": 6719
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 2.993976874460561e-06,
+      "loss": 0.7174,
+      "step": 6720
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 2.9891272090902867e-06,
+      "loss": 0.4164,
+      "step": 6721
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 2.9842812249114088e-06,
+      "loss": 0.7191,
+      "step": 6722
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 2.979438922734379e-06,
+      "loss": 0.5136,
+      "step": 6723
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 2.974600303369046e-06,
+      "loss": 0.4405,
+      "step": 6724
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 2.9697653676246412e-06,
+      "loss": 0.5675,
+      "step": 6725
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 2.9649341163097865e-06,
+      "loss": 0.7126,
+      "step": 6726
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 2.960106550232475e-06,
+      "loss": 0.6254,
+      "step": 6727
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 2.955282670200099e-06,
+      "loss": 0.7492,
+      "step": 6728
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 2.950462477019408e-06,
+      "loss": 0.4779,
+      "step": 6729
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 2.9456459714965606e-06,
+      "loss": 0.6401,
+      "step": 6730
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 2.940833154437084e-06,
+      "loss": 0.5522,
+      "step": 6731
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 2.936024026645906e-06,
+      "loss": 0.4611,
+      "step": 6732
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 2.931218588927315e-06,
+      "loss": 0.3641,
+      "step": 6733
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 2.926416842085003e-06,
+      "loss": 0.4766,
+      "step": 6734
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 2.9216187869220202e-06,
+      "loss": 0.6833,
+      "step": 6735
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 2.916824424240819e-06,
+      "loss": 0.4986,
+      "step": 6736
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 2.9120337548432324e-06,
+      "loss": 0.5953,
+      "step": 6737
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 2.9072467795304733e-06,
+      "loss": 0.7143,
+      "step": 6738
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 2.902463499103131e-06,
+      "loss": 0.6257,
+      "step": 6739
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 2.8976839143611877e-06,
+      "loss": 0.6223,
+      "step": 6740
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 2.892908026103991e-06,
+      "loss": 0.5824,
+      "step": 6741
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 2.8881358351302872e-06,
+      "loss": 0.6682,
+      "step": 6742
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 2.8833673422381998e-06,
+      "loss": 0.6529,
+      "step": 6743
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 2.878602548225229e-06,
+      "loss": 0.5498,
+      "step": 6744
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 2.873841453888257e-06,
+      "loss": 0.5742,
+      "step": 6745
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 2.869084060023558e-06,
+      "loss": 0.5538,
+      "step": 6746
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 2.8643303674267623e-06,
+      "loss": 0.5807,
+      "step": 6747
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 2.8595803768929117e-06,
+      "loss": 0.6028,
+      "step": 6748
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 2.8548340892164127e-06,
+      "loss": 0.6364,
+      "step": 6749
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 2.850091505191049e-06,
+      "loss": 0.7549,
+      "step": 6750
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 2.8453526256099953e-06,
+      "loss": 0.5861,
+      "step": 6751
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 2.840617451265801e-06,
+      "loss": 0.5689,
+      "step": 6752
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 2.835885982950395e-06,
+      "loss": 0.6183,
+      "step": 6753
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 2.831158221455088e-06,
+      "loss": 0.5818,
+      "step": 6754
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 2.8264341675705713e-06,
+      "loss": 0.5739,
+      "step": 6755
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 2.8217138220869187e-06,
+      "loss": 0.5642,
+      "step": 6756
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 2.816997185793577e-06,
+      "loss": 0.5255,
+      "step": 6757
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 2.8122842594793753e-06,
+      "loss": 0.6294,
+      "step": 6758
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 2.8075750439325284e-06,
+      "loss": 0.4992,
+      "step": 6759
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 2.8028695399406195e-06,
+      "loss": 0.533,
+      "step": 6760
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 2.7981677482906226e-06,
+      "loss": 0.7685,
+      "step": 6761
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 2.7934696697688806e-06,
+      "loss": 0.8266,
+      "step": 6762
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 2.788775305161123e-06,
+      "loss": 0.8091,
+      "step": 6763
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 2.7840846552524557e-06,
+      "loss": 0.5744,
+      "step": 6764
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 2.779397720827359e-06,
+      "loss": 0.6434,
+      "step": 6765
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 2.774714502669698e-06,
+      "loss": 0.5473,
+      "step": 6766
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 2.7700350015627153e-06,
+      "loss": 0.8252,
+      "step": 6767
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 2.7653592182890266e-06,
+      "loss": 0.5811,
+      "step": 6768
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 2.760687153630634e-06,
+      "loss": 0.5499,
+      "step": 6769
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 2.75601880836891e-06,
+      "loss": 0.6864,
+      "step": 6770
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 2.7513541832846102e-06,
+      "loss": 0.5706,
+      "step": 6771
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 2.7466932791578664e-06,
+      "loss": 0.7013,
+      "step": 6772
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 2.742036096768183e-06,
+      "loss": 0.675,
+      "step": 6773
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 2.7373826368944587e-06,
+      "loss": 0.5735,
+      "step": 6774
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 2.7327329003149464e-06,
+      "loss": 0.4917,
+      "step": 6775
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 2.7280868878072907e-06,
+      "loss": 0.6945,
+      "step": 6776
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 2.7234446001485116e-06,
+      "loss": 0.5713,
+      "step": 6777
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 2.7188060381150044e-06,
+      "loss": 0.5142,
+      "step": 6778
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 2.714171202482538e-06,
+      "loss": 0.6762,
+      "step": 6779
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 2.709540094026275e-06,
+      "loss": 0.441,
+      "step": 6780
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 2.704912713520727e-06,
+      "loss": 0.5451,
+      "step": 6781
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 2.7002890617398026e-06,
+      "loss": 0.4458,
+      "step": 6782
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 2.69566913945678e-06,
+      "loss": 0.4417,
+      "step": 6783
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 2.6910529474443086e-06,
+      "loss": 0.848,
+      "step": 6784
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 2.6910529474443086e-06,
+      "loss": 0.6186,
+      "step": 6785
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 2.6864404864744315e-06,
+      "loss": 0.493,
+      "step": 6786
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 2.6818317573185566e-06,
+      "loss": 0.413,
+      "step": 6787
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 2.6772267607474562e-06,
+      "loss": 0.5658,
+      "step": 6788
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 2.6726254975312947e-06,
+      "loss": 0.6959,
+      "step": 6789
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 2.668027968439607e-06,
+      "loss": 0.7059,
+      "step": 6790
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 2.663434174241297e-06,
+      "loss": 0.5055,
+      "step": 6791
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 2.658844115704662e-06,
+      "loss": 0.6164,
+      "step": 6792
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 2.654257793597359e-06,
+      "loss": 0.6594,
+      "step": 6793
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 2.6496752086864195e-06,
+      "loss": 0.4804,
+      "step": 6794
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 2.6450963617382584e-06,
+      "loss": 0.5029,
+      "step": 6795
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 2.6405212535186517e-06,
+      "loss": 0.4605,
+      "step": 6796
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 2.635949884792774e-06,
+      "loss": 0.7909,
+      "step": 6797
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 2.6313822563251543e-06,
+      "loss": 0.6773,
+      "step": 6798
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 2.626818368879708e-06,
+      "loss": 0.6073,
+      "step": 6799
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 2.6222582232197064e-06,
+      "loss": 0.5703,
+      "step": 6800
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 2.6177018201078097e-06,
+      "loss": 0.7988,
+      "step": 6801
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 2.613149160306058e-06,
+      "loss": 0.6092,
+      "step": 6802
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 2.6086002445758566e-06,
+      "loss": 0.6086,
+      "step": 6803
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 2.60405507367798e-06,
+      "loss": 0.7469,
+      "step": 6804
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 2.5995136483725922e-06,
+      "loss": 0.8921,
+      "step": 6805
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 2.5949759694192056e-06,
+      "loss": 0.3338,
+      "step": 6806
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 2.5904420375767207e-06,
+      "loss": 0.6629,
+      "step": 6807
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 2.585911853603426e-06,
+      "loss": 0.6292,
+      "step": 6808
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 2.5813854182569625e-06,
+      "loss": 0.431,
+      "step": 6809
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 2.5768627322943467e-06,
+      "loss": 0.4375,
+      "step": 6810
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 2.5723437964719794e-06,
+      "loss": 0.7435,
+      "step": 6811
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 2.5678286115456086e-06,
+      "loss": 0.6351,
+      "step": 6812
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 2.563317178270394e-06,
+      "loss": 0.7017,
+      "step": 6813
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 2.5588094974008355e-06,
+      "loss": 0.6992,
+      "step": 6814
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 2.5543055696908206e-06,
+      "loss": 0.6287,
+      "step": 6815
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 2.5498053958936036e-06,
+      "loss": 0.663,
+      "step": 6816
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 2.545308976761815e-06,
+      "loss": 0.6612,
+      "step": 6817
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 2.5408163130474504e-06,
+      "loss": 0.6353,
+      "step": 6818
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 2.5363274055018843e-06,
+      "loss": 0.7472,
+      "step": 6819
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 2.531842254875863e-06,
+      "loss": 0.5604,
+      "step": 6820
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 2.527360861919498e-06,
+      "loss": 0.4247,
+      "step": 6821
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 2.5228832273822777e-06,
+      "loss": 0.7904,
+      "step": 6822
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 2.51840935201306e-06,
+      "loss": 0.7949,
+      "step": 6823
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 2.5139392365600783e-06,
+      "loss": 0.4735,
+      "step": 6824
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 2.5094728817709275e-06,
+      "loss": 0.5918,
+      "step": 6825
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 2.505010288392587e-06,
+      "loss": 0.5807,
+      "step": 6826
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 2.5005514571713935e-06,
+      "loss": 0.7,
+      "step": 6827
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 2.496096388853064e-06,
+      "loss": 0.5291,
+      "step": 6828
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 2.491645084182681e-06,
+      "loss": 0.5734,
+      "step": 6829
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 2.4871975439047012e-06,
+      "loss": 0.9125,
+      "step": 6830
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 2.482753768762952e-06,
+      "loss": 0.6723,
+      "step": 6831
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 2.478313759500625e-06,
+      "loss": 0.6678,
+      "step": 6832
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 2.4738775168602817e-06,
+      "loss": 0.6027,
+      "step": 6833
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 2.469445041583873e-06,
+      "loss": 0.3866,
+      "step": 6834
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 2.4650163344126925e-06,
+      "loss": 0.4062,
+      "step": 6835
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 2.4605913960874192e-06,
+      "loss": 0.6773,
+      "step": 6836
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 2.456170227348101e-06,
+      "loss": 0.6973,
+      "step": 6837
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 2.451752828934148e-06,
+      "loss": 0.6183,
+      "step": 6838
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 2.4473392015843423e-06,
+      "loss": 0.6609,
+      "step": 6839
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 2.4429293460368513e-06,
+      "loss": 0.4376,
+      "step": 6840
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 2.4385232630291856e-06,
+      "loss": 0.7653,
+      "step": 6841
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 2.4341209532982384e-06,
+      "loss": 0.4505,
+      "step": 6842
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 2.429722417580274e-06,
+      "loss": 0.464,
+      "step": 6843
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 2.425327656610915e-06,
+      "loss": 0.5767,
+      "step": 6844
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 2.4209366711251734e-06,
+      "loss": 0.5558,
+      "step": 6845
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 2.4165494618574093e-06,
+      "loss": 0.6346,
+      "step": 6846
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 2.4121660295413547e-06,
+      "loss": 0.6595,
+      "step": 6847
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 2.407786374910115e-06,
+      "loss": 0.5399,
+      "step": 6848
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 2.4034104986961627e-06,
+      "loss": 0.5153,
+      "step": 6849
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 2.3990384016313314e-06,
+      "loss": 0.4787,
+      "step": 6850
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 2.394670084446843e-06,
+      "loss": 0.4323,
+      "step": 6851
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 2.39030554787327e-06,
+      "loss": 0.4784,
+      "step": 6852
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 2.3859447926405485e-06,
+      "loss": 0.701,
+      "step": 6853
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 2.3815878194779927e-06,
+      "loss": 0.6704,
+      "step": 6854
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 2.377234629114275e-06,
+      "loss": 0.4789,
+      "step": 6855
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 2.3728852222774543e-06,
+      "loss": 0.79,
+      "step": 6856
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.3685395996949385e-06,
+      "loss": 0.4589,
+      "step": 6857
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.364197762093509e-06,
+      "loss": 0.7966,
+      "step": 6858
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.359859710199308e-06,
+      "loss": 0.7655,
+      "step": 6859
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 2.3555254447378467e-06,
+      "loss": 0.5931,
+      "step": 6860
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 2.3511949664340173e-06,
+      "loss": 0.8128,
+      "step": 6861
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 2.3468682760120626e-06,
+      "loss": 0.4182,
+      "step": 6862
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 2.3425453741955927e-06,
+      "loss": 0.7192,
+      "step": 6863
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 2.338226261707599e-06,
+      "loss": 0.5352,
+      "step": 6864
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 2.3339109392704146e-06,
+      "loss": 0.5911,
+      "step": 6865
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 2.3295994076057537e-06,
+      "loss": 0.7066,
+      "step": 6866
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 2.3252916674347036e-06,
+      "loss": 0.6007,
+      "step": 6867
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 2.320987719477702e-06,
+      "loss": 0.3456,
+      "step": 6868
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 2.3166875644545644e-06,
+      "loss": 0.6033,
+      "step": 6869
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 2.3123912030844713e-06,
+      "loss": 0.7968,
+      "step": 6870
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 2.308098636085945e-06,
+      "loss": 0.5706,
+      "step": 6871
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 2.303809864176909e-06,
+      "loss": 0.5512,
+      "step": 6872
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 2.2995248880746336e-06,
+      "loss": 0.657,
+      "step": 6873
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 2.2952437084957563e-06,
+      "loss": 0.6007,
+      "step": 6874
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 2.290966326156274e-06,
+      "loss": 0.7832,
+      "step": 6875
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 2.2866927417715597e-06,
+      "loss": 0.5385,
+      "step": 6876
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.282422956056343e-06,
+      "loss": 0.6686,
+      "step": 6877
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.2781569697247236e-06,
+      "loss": 0.5714,
+      "step": 6878
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.2738947834901604e-06,
+      "loss": 0.4759,
+      "step": 6879
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 2.2696363980654807e-06,
+      "loss": 0.739,
+      "step": 6880
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 2.265381814162873e-06,
+      "loss": 0.4927,
+      "step": 6881
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 2.261131032493896e-06,
+      "loss": 0.6657,
+      "step": 6882
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 2.2568840537694646e-06,
+      "loss": 0.5392,
+      "step": 6883
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 2.252640878699863e-06,
+      "loss": 0.8203,
+      "step": 6884
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.248401507994738e-06,
+      "loss": 0.4354,
+      "step": 6885
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.2441659423630967e-06,
+      "loss": 0.6391,
+      "step": 6886
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.2399341825133202e-06,
+      "loss": 0.7072,
+      "step": 6887
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.235706229153137e-06,
+      "loss": 0.6461,
+      "step": 6888
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 2.2314820829896552e-06,
+      "loss": 0.6271,
+      "step": 6889
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 2.227261744729334e-06,
+      "loss": 0.5718,
+      "step": 6890
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 2.223045215078004e-06,
+      "loss": 0.6848,
+      "step": 6891
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 2.2188324947408525e-06,
+      "loss": 0.6327,
+      "step": 6892
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 2.2146235844224343e-06,
+      "loss": 0.6511,
+      "step": 6893
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 2.210418484826665e-06,
+      "loss": 0.5552,
+      "step": 6894
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 2.206217196656826e-06,
+      "loss": 0.8094,
+      "step": 6895
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 2.202019720615553e-06,
+      "loss": 0.6817,
+      "step": 6896
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 2.1978260574048533e-06,
+      "loss": 0.8613,
+      "step": 6897
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 2.193636207726091e-06,
+      "loss": 0.4935,
+      "step": 6898
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 2.189450172279997e-06,
+      "loss": 0.4785,
+      "step": 6899
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 2.1852679517666592e-06,
+      "loss": 0.6244,
+      "step": 6900
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 2.181089546885531e-06,
+      "loss": 0.3833,
+      "step": 6901
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 2.1769149583354243e-06,
+      "loss": 0.5336,
+      "step": 6902
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 2.1727441868145153e-06,
+      "loss": 0.5149,
+      "step": 6903
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 2.1685772330203497e-06,
+      "loss": 0.5581,
+      "step": 6904
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.164414097649814e-06,
+      "loss": 0.5044,
+      "step": 6905
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.160254781399176e-06,
+      "loss": 0.6551,
+      "step": 6906
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.1560992849640556e-06,
+      "loss": 0.4855,
+      "step": 6907
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 2.1519476090394344e-06,
+      "loss": 0.6332,
+      "step": 6908
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 2.1477997543196513e-06,
+      "loss": 0.5283,
+      "step": 6909
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 2.1436557214984278e-06,
+      "loss": 0.7089,
+      "step": 6910
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 2.1395155112688136e-06,
+      "loss": 0.6832,
+      "step": 6911
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 2.1353791243232417e-06,
+      "loss": 0.5935,
+      "step": 6912
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 2.131246561353495e-06,
+      "loss": 0.5341,
+      "step": 6913
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 2.12711782305072e-06,
+      "loss": 0.5529,
+      "step": 6914
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 2.122992910105431e-06,
+      "loss": 0.6901,
+      "step": 6915
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 2.1188718232074995e-06,
+      "loss": 0.6003,
+      "step": 6916
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 2.1147545630461367e-06,
+      "loss": 0.4693,
+      "step": 6917
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 2.1106411303099455e-06,
+      "loss": 0.6178,
+      "step": 6918
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 2.106531525686861e-06,
+      "loss": 0.3917,
+      "step": 6919
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 2.1024257498642037e-06,
+      "loss": 0.8955,
+      "step": 6920
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 2.0983238035286345e-06,
+      "loss": 0.6369,
+      "step": 6921
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 2.094225687366186e-06,
+      "loss": 0.763,
+      "step": 6922
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 2.0901314020622344e-06,
+      "loss": 0.72,
+      "step": 6923
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 2.0860409483015318e-06,
+      "loss": 0.6071,
+      "step": 6924
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 2.081954326768176e-06,
+      "loss": 0.6387,
+      "step": 6925
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 2.0778715381456426e-06,
+      "loss": 0.5051,
+      "step": 6926
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 2.073792583116749e-06,
+      "loss": 0.5849,
+      "step": 6927
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 2.0697174623636794e-06,
+      "loss": 0.6176,
+      "step": 6928
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 2.0656461765679648e-06,
+      "loss": 0.7107,
+      "step": 6929
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 2.0615787264105094e-06,
+      "loss": 0.6465,
+      "step": 6930
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 2.0575151125715757e-06,
+      "loss": 0.6731,
+      "step": 6931
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 2.053455335730775e-06,
+      "loss": 0.5352,
+      "step": 6932
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 2.049399396567081e-06,
+      "loss": 0.6054,
+      "step": 6933
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 2.0453472957588342e-06,
+      "loss": 0.6849,
+      "step": 6934
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 2.04129903398371e-06,
+      "loss": 0.6365,
+      "step": 6935
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 2.037254611918768e-06,
+      "loss": 0.6015,
+      "step": 6936
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 2.03321403024041e-06,
+      "loss": 0.5219,
+      "step": 6937
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 2.029177289624404e-06,
+      "loss": 0.5658,
+      "step": 6938
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 2.02514439074587e-06,
+      "loss": 0.5137,
+      "step": 6939
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 2.0211153342792887e-06,
+      "loss": 0.6025,
+      "step": 6940
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 2.017090120898485e-06,
+      "loss": 0.6409,
+      "step": 6941
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 2.0130687512766673e-06,
+      "loss": 0.6975,
+      "step": 6942
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 2.0090512260863807e-06,
+      "loss": 0.7203,
+      "step": 6943
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 2.005037545999533e-06,
+      "loss": 0.394,
+      "step": 6944
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 2.0010277116873895e-06,
+      "loss": 0.4397,
+      "step": 6945
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.997021723820572e-06,
+      "loss": 0.5563,
+      "step": 6946
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.993019583069056e-06,
+      "loss": 0.6743,
+      "step": 6947
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.9890212901021804e-06,
+      "loss": 0.6321,
+      "step": 6948
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.985026845588633e-06,
+      "loss": 0.612,
+      "step": 6949
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.9810362501964665e-06,
+      "loss": 0.6665,
+      "step": 6950
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.9770495045930813e-06,
+      "loss": 0.6558,
+      "step": 6951
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.973066609445237e-06,
+      "loss": 0.4531,
+      "step": 6952
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.969087565419053e-06,
+      "loss": 0.5651,
+      "step": 6953
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.9651123731799974e-06,
+      "loss": 0.8268,
+      "step": 6954
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.961141033392902e-06,
+      "loss": 0.6118,
+      "step": 6955
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.957173546721949e-06,
+      "loss": 0.663,
+      "step": 6956
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.953209913830675e-06,
+      "loss": 0.6381,
+      "step": 6957
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.9492501353819803e-06,
+      "loss": 0.6212,
+      "step": 6958
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.9452942120381097e-06,
+      "loss": 0.5991,
+      "step": 6959
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.94134214446067e-06,
+      "loss": 0.6119,
+      "step": 6960
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.9373939333106245e-06,
+      "loss": 0.5745,
+      "step": 6961
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.933449579248284e-06,
+      "loss": 0.623,
+      "step": 6962
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.9295090829333228e-06,
+      "loss": 0.5513,
+      "step": 6963
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.9255724450247674e-06,
+      "loss": 0.3741,
+      "step": 6964
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.921639666180991e-06,
+      "loss": 0.6975,
+      "step": 6965
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.917710747059734e-06,
+      "loss": 0.5891,
+      "step": 6966
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.9137856883180846e-06,
+      "loss": 0.6559,
+      "step": 6967
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.909864490612484e-06,
+      "loss": 0.5121,
+      "step": 6968
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.9059471545987335e-06,
+      "loss": 0.6387,
+      "step": 6969
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.9020336809319806e-06,
+      "loss": 0.5215,
+      "step": 6970
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 1.8981240702667353e-06,
+      "loss": 0.4672,
+      "step": 6971
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.8942183232568522e-06,
+      "loss": 0.5982,
+      "step": 6972
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.890316440555548e-06,
+      "loss": 0.732,
+      "step": 6973
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.8864184228153948e-06,
+      "loss": 0.4796,
+      "step": 6974
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.8825242706883074e-06,
+      "loss": 0.6469,
+      "step": 6975
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.8786339848255591e-06,
+      "loss": 0.5878,
+      "step": 6976
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.8747475658777825e-06,
+      "loss": 0.7773,
+      "step": 6977
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.870865014494952e-06,
+      "loss": 0.761,
+      "step": 6978
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.866986331326412e-06,
+      "loss": 0.5021,
+      "step": 6979
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.8631115170208463e-06,
+      "loss": 0.5758,
+      "step": 6980
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.8592405722262924e-06,
+      "loss": 0.7029,
+      "step": 6981
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.8553734975901405e-06,
+      "loss": 0.6801,
+      "step": 6982
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.8515102937591428e-06,
+      "loss": 0.589,
+      "step": 6983
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.8476509613793907e-06,
+      "loss": 0.687,
+      "step": 6984
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.8437955010963459e-06,
+      "loss": 0.6256,
+      "step": 6985
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.839943913554809e-06,
+      "loss": 0.5331,
+      "step": 6986
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.836096199398929e-06,
+      "loss": 0.4203,
+      "step": 6987
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.8322523592722157e-06,
+      "loss": 0.6775,
+      "step": 6988
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.8284123938175273e-06,
+      "loss": 0.5989,
+      "step": 6989
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.8245763036770864e-06,
+      "loss": 0.5185,
+      "step": 6990
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 1.8207440894924488e-06,
+      "loss": 0.562,
+      "step": 6991
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.8169157519045355e-06,
+      "loss": 0.5688,
+      "step": 6992
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.8130912915536065e-06,
+      "loss": 0.4279,
+      "step": 6993
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.8092707090792776e-06,
+      "loss": 0.6094,
+      "step": 6994
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.8054540051205326e-06,
+      "loss": 0.7344,
+      "step": 6995
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.8016411803156857e-06,
+      "loss": 0.5862,
+      "step": 6996
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.7978322353024101e-06,
+      "loss": 0.6306,
+      "step": 6997
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.7940271707177353e-06,
+      "loss": 0.756,
+      "step": 6998
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.790225987198027e-06,
+      "loss": 0.4987,
+      "step": 6999
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.7864286853790101e-06,
+      "loss": 0.6801,
+      "step": 7000
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.7826352658957712e-06,
+      "loss": 0.4845,
+      "step": 7001
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.778845729382736e-06,
+      "loss": 0.5659,
+      "step": 7002
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.775060076473678e-06,
+      "loss": 0.6484,
+      "step": 7003
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.7712783078017326e-06,
+      "loss": 0.5248,
+      "step": 7004
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.767500423999363e-06,
+      "loss": 0.5514,
+      "step": 7005
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.763726425698417e-06,
+      "loss": 0.4626,
+      "step": 7006
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.7599563135300644e-06,
+      "loss": 0.5719,
+      "step": 7007
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.7561900881248373e-06,
+      "loss": 0.4903,
+      "step": 7008
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.7524277501126175e-06,
+      "loss": 0.535,
+      "step": 7009
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.7486693001226268e-06,
+      "loss": 0.6207,
+      "step": 7010
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.7449147387834514e-06,
+      "loss": 0.7279,
+      "step": 7011
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.7411640667230168e-06,
+      "loss": 0.5814,
+      "step": 7012
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.7374172845686015e-06,
+      "loss": 0.5037,
+      "step": 7013
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.7336743929468351e-06,
+      "loss": 0.7982,
+      "step": 7014
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 1.7336743929468351e-06,
+      "loss": 0.5966,
+      "step": 7015
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.729935392483692e-06,
+      "loss": 0.6332,
+      "step": 7016
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.726200283804502e-06,
+      "loss": 0.6488,
+      "step": 7017
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.7224690675339356e-06,
+      "loss": 0.5524,
+      "step": 7018
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.7187417442960208e-06,
+      "loss": 0.5777,
+      "step": 7019
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.7150183147141286e-06,
+      "loss": 0.7069,
+      "step": 7020
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.711298779410983e-06,
+      "loss": 0.6541,
+      "step": 7021
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.7075831390086534e-06,
+      "loss": 0.5749,
+      "step": 7022
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.703871394128559e-06,
+      "loss": 0.4949,
+      "step": 7023
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.7001635453914699e-06,
+      "loss": 0.6971,
+      "step": 7024
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.6964595934175015e-06,
+      "loss": 0.6017,
+      "step": 7025
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.6927595388261164e-06,
+      "loss": 0.6221,
+      "step": 7026
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 1.6890633822361308e-06,
+      "loss": 0.5086,
+      "step": 7027
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.6853711242657e-06,
+      "loss": 0.3621,
+      "step": 7028
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.6816827655323413e-06,
+      "loss": 0.7459,
+      "step": 7029
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.6779983066529032e-06,
+      "loss": 0.5259,
+      "step": 7030
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.674317748243595e-06,
+      "loss": 0.5796,
+      "step": 7031
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.6706410909199666e-06,
+      "loss": 0.6671,
+      "step": 7032
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.6669683352969146e-06,
+      "loss": 0.535,
+      "step": 7033
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 1.6632994819886977e-06,
+      "loss": 0.6067,
+      "step": 7034
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.6596345316089002e-06,
+      "loss": 0.513,
+      "step": 7035
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.655973484770465e-06,
+      "loss": 0.3007,
+      "step": 7036
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.652316342085683e-06,
+      "loss": 0.8053,
+      "step": 7037
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.6486631041661842e-06,
+      "loss": 0.7122,
+      "step": 7038
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.6450137716229635e-06,
+      "loss": 0.6147,
+      "step": 7039
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.6413683450663463e-06,
+      "loss": 0.8086,
+      "step": 7040
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.6377268251060062e-06,
+      "loss": 0.7802,
+      "step": 7041
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.6340892123509643e-06,
+      "loss": 0.5111,
+      "step": 7042
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.630455507409595e-06,
+      "loss": 0.6317,
+      "step": 7043
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.6268257108896069e-06,
+      "loss": 0.6562,
+      "step": 7044
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.623199823398075e-06,
+      "loss": 0.6414,
+      "step": 7045
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.6195778455414035e-06,
+      "loss": 0.8124,
+      "step": 7046
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.615959777925341e-06,
+      "loss": 0.5029,
+      "step": 7047
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.6123456211549926e-06,
+      "loss": 0.5613,
+      "step": 7048
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.6087353758348023e-06,
+      "loss": 0.5613,
+      "step": 7049
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.6051290425685706e-06,
+      "loss": 0.6189,
+      "step": 7050
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.6015266219594289e-06,
+      "loss": 0.5735,
+      "step": 7051
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.5979281146098674e-06,
+      "loss": 0.672,
+      "step": 7052
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.5943335211217102e-06,
+      "loss": 0.5892,
+      "step": 7053
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.5907428420961269e-06,
+      "loss": 0.6306,
+      "step": 7054
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.5871560781336503e-06,
+      "loss": 0.5058,
+      "step": 7055
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.58357322983414e-06,
+      "loss": 0.4913,
+      "step": 7056
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.579994297796808e-06,
+      "loss": 0.3619,
+      "step": 7057
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 1.5764192826202118e-06,
+      "loss": 0.6751,
+      "step": 7058
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.5728481849022452e-06,
+      "loss": 0.6683,
+      "step": 7059
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.569281005240153e-06,
+      "loss": 0.7822,
+      "step": 7060
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.5657177442305354e-06,
+      "loss": 0.7779,
+      "step": 7061
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.5621584024693214e-06,
+      "loss": 0.7563,
+      "step": 7062
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.5586029805517933e-06,
+      "loss": 0.7414,
+      "step": 7063
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.5550514790725723e-06,
+      "loss": 0.8394,
+      "step": 7064
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.5515038986256226e-06,
+      "loss": 0.539,
+      "step": 7065
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.5479602398042636e-06,
+      "loss": 0.6663,
+      "step": 7066
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.5444205032011493e-06,
+      "loss": 0.3407,
+      "step": 7067
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.5408846894082785e-06,
+      "loss": 0.4646,
+      "step": 7068
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.5373527990169973e-06,
+      "loss": 0.7097,
+      "step": 7069
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 1.5338248326179943e-06,
+      "loss": 0.6558,
+      "step": 7070
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.5303007908013007e-06,
+      "loss": 0.6334,
+      "step": 7071
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.5267806741562917e-06,
+      "loss": 0.5463,
+      "step": 7072
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.5232644832716857e-06,
+      "loss": 0.7751,
+      "step": 7073
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.5197522187355483e-06,
+      "loss": 0.5773,
+      "step": 7074
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.5162438811352847e-06,
+      "loss": 0.5965,
+      "step": 7075
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.512739471057642e-06,
+      "loss": 0.531,
+      "step": 7076
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.5092389890887154e-06,
+      "loss": 0.7602,
+      "step": 7077
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 1.5057424358139365e-06,
+      "loss": 0.669,
+      "step": 7078
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.5022498118180879e-06,
+      "loss": 0.6418,
+      "step": 7079
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.4987611176852878e-06,
+      "loss": 0.5902,
+      "step": 7080
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.4952763539990006e-06,
+      "loss": 0.6025,
+      "step": 7081
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.4917955213420349e-06,
+      "loss": 0.745,
+      "step": 7082
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.488318620296536e-06,
+      "loss": 0.5249,
+      "step": 7083
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.4848456514440001e-06,
+      "loss": 0.6447,
+      "step": 7084
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.481376615365257e-06,
+      "loss": 0.6498,
+      "step": 7085
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.4779115126404868e-06,
+      "loss": 0.5846,
+      "step": 7086
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.4744503438492069e-06,
+      "loss": 0.5421,
+      "step": 7087
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.4709931095702733e-06,
+      "loss": 0.6126,
+      "step": 7088
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.467539810381896e-06,
+      "loss": 0.6168,
+      "step": 7089
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.4640904468616129e-06,
+      "loss": 0.645,
+      "step": 7090
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.4606450195863153e-06,
+      "loss": 0.8671,
+      "step": 7091
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.4572035291322256e-06,
+      "loss": 0.5245,
+      "step": 7092
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.4537659760749167e-06,
+      "loss": 0.5731,
+      "step": 7093
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.450332360989301e-06,
+      "loss": 0.603,
+      "step": 7094
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.4469026844496276e-06,
+      "loss": 0.6164,
+      "step": 7095
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.44347694702949e-06,
+      "loss": 0.6931,
+      "step": 7096
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.4400551493018195e-06,
+      "loss": 0.6856,
+      "step": 7097
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.4366372918389076e-06,
+      "loss": 0.5197,
+      "step": 7098
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.4332233752123532e-06,
+      "loss": 0.6134,
+      "step": 7099
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.429813399993124e-06,
+      "loss": 0.496,
+      "step": 7100
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.4264073667515144e-06,
+      "loss": 0.6674,
+      "step": 7101
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.4230052760571655e-06,
+      "loss": 0.5428,
+      "step": 7102
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.419607128479053e-06,
+      "loss": 0.529,
+      "step": 7103
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.4162129245855083e-06,
+      "loss": 0.4986,
+      "step": 7104
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.4128226649441829e-06,
+      "loss": 0.5613,
+      "step": 7105
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.4094363501220813e-06,
+      "loss": 0.4296,
+      "step": 7106
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.4060539806855456e-06,
+      "loss": 0.5391,
+      "step": 7107
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.402675557200253e-06,
+      "loss": 0.6842,
+      "step": 7108
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.39930108023123e-06,
+      "loss": 0.508,
+      "step": 7109
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.3959305503428443e-06,
+      "loss": 0.6326,
+      "step": 7110
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.392563968098784e-06,
+      "loss": 0.6356,
+      "step": 7111
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.3892013340621012e-06,
+      "loss": 0.5462,
+      "step": 7112
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.385842648795166e-06,
+      "loss": 0.494,
+      "step": 7113
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 1.3824879128597118e-06,
+      "loss": 0.6679,
+      "step": 7114
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.3791371268167901e-06,
+      "loss": 0.513,
+      "step": 7115
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.3757902912268111e-06,
+      "loss": 0.7183,
+      "step": 7116
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.3724474066494991e-06,
+      "loss": 0.4089,
+      "step": 7117
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.3691084736439431e-06,
+      "loss": 0.7072,
+      "step": 7118
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.3657734927685489e-06,
+      "loss": 0.5142,
+      "step": 7119
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.3624424645810874e-06,
+      "loss": 0.4395,
+      "step": 7120
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.359115389638646e-06,
+      "loss": 0.5616,
+      "step": 7121
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.355792268497663e-06,
+      "loss": 0.699,
+      "step": 7122
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.3524731017139053e-06,
+      "loss": 0.6123,
+      "step": 7123
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.349157889842484e-06,
+      "loss": 0.7464,
+      "step": 7124
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.3458466334378527e-06,
+      "loss": 0.4802,
+      "step": 7125
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.3425393330538022e-06,
+      "loss": 0.6386,
+      "step": 7126
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.3392359892434564e-06,
+      "loss": 0.5448,
+      "step": 7127
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.3359366025592819e-06,
+      "loss": 0.457,
+      "step": 7128
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.3326411735530785e-06,
+      "loss": 0.3517,
+      "step": 7129
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.3293497027759893e-06,
+      "loss": 0.469,
+      "step": 7130
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.3260621907784931e-06,
+      "loss": 0.6719,
+      "step": 7131
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.322778638110414e-06,
+      "loss": 0.4709,
+      "step": 7132
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 1.319499045320899e-06,
+      "loss": 0.5889,
+      "step": 7133
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.3162234129584479e-06,
+      "loss": 0.7063,
+      "step": 7134
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.3129517415708781e-06,
+      "loss": 0.6064,
+      "step": 7135
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.3096840317053744e-06,
+      "loss": 0.623,
+      "step": 7136
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.3064202839084328e-06,
+      "loss": 0.574,
+      "step": 7137
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.3031604987259e-06,
+      "loss": 0.6673,
+      "step": 7138
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.2999046767029538e-06,
+      "loss": 0.6362,
+      "step": 7139
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.2966528183841114e-06,
+      "loss": 0.5449,
+      "step": 7140
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 1.2934049243132297e-06,
+      "loss": 0.5636,
+      "step": 7141
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.2901609950335014e-06,
+      "loss": 0.5531,
+      "step": 7142
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.2869210310874486e-06,
+      "loss": 0.5745,
+      "step": 7143
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.283685033016943e-06,
+      "loss": 0.5871,
+      "step": 7144
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.2804530013631854e-06,
+      "loss": 0.6174,
+      "step": 7145
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.2772249366667099e-06,
+      "loss": 0.7541,
+      "step": 7146
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.2740008394673957e-06,
+      "loss": 0.58,
+      "step": 7147
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.2707807103044561e-06,
+      "loss": 0.5666,
+      "step": 7148
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 1.267564549716435e-06,
+      "loss": 0.6116,
+      "step": 7149
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.2643523582412192e-06,
+      "loss": 0.5802,
+      "step": 7150
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.2611441364160286e-06,
+      "loss": 0.5544,
+      "step": 7151
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.2579398847774177e-06,
+      "loss": 0.5534,
+      "step": 7152
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.2547396038612824e-06,
+      "loss": 0.5133,
+      "step": 7153
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.2515432942028503e-06,
+      "loss": 0.6194,
+      "step": 7154
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.248350956336683e-06,
+      "loss": 0.4907,
+      "step": 7155
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.2451625907966864e-06,
+      "loss": 0.5276,
+      "step": 7156
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 1.2419781981160895e-06,
+      "loss": 0.7559,
+      "step": 7157
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.2387977788274664e-06,
+      "loss": 0.8257,
+      "step": 7158
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.2356213334627276e-06,
+      "loss": 0.8099,
+      "step": 7159
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.2324488625531094e-06,
+      "loss": 0.5704,
+      "step": 7160
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.229280366629193e-06,
+      "loss": 0.6366,
+      "step": 7161
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.2261158462208877e-06,
+      "loss": 0.539,
+      "step": 7162
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.2229553018574452e-06,
+      "loss": 0.8092,
+      "step": 7163
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.2197987340674454e-06,
+      "loss": 0.585,
+      "step": 7164
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 1.2166461433788051e-06,
+      "loss": 0.5449,
+      "step": 7165
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.2134975303187828e-06,
+      "loss": 0.6833,
+      "step": 7166
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.2103528954139547e-06,
+      "loss": 0.5519,
+      "step": 7167
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.2072122391902585e-06,
+      "loss": 0.696,
+      "step": 7168
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 1.2040755621729405e-06,
+      "loss": 0.6678,
+      "step": 7169
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.2009428648865922e-06,
+      "loss": 0.5597,
+      "step": 7170
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.1978141478551392e-06,
+      "loss": 0.4801,
+      "step": 7171
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.1946894116018404e-06,
+      "loss": 0.6897,
+      "step": 7172
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.1915686566492973e-06,
+      "loss": 0.5549,
+      "step": 7173
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.1884518835194343e-06,
+      "loss": 0.4992,
+      "step": 7174
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.1853390927335117e-06,
+      "loss": 0.6662,
+      "step": 7175
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.1822302848121246e-06,
+      "loss": 0.4374,
+      "step": 7176
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 1.1791254602752045e-06,
+      "loss": 0.5355,
+      "step": 7177
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 1.1760246196420132e-06,
+      "loss": 0.4419,
+      "step": 7178
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 1.172927763431156e-06,
+      "loss": 0.4373,
+      "step": 7179
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 1.1698348921605595e-06,
+      "loss": 0.8313,
+      "step": 7180
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 1.1667460063474883e-06,
+      "loss": 0.6207,
+      "step": 7181
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 1.16366110650854e-06,
+      "loss": 0.4828,
+      "step": 7182
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 1.1605801931596439e-06,
+      "loss": 0.3957,
+      "step": 7183
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 1.1575032668160706e-06,
+      "loss": 0.5518,
+      "step": 7184
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 1.1544303279924174e-06,
+      "loss": 0.6975,
+      "step": 7185
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 1.1513613772026171e-06,
+      "loss": 0.7003,
+      "step": 7186
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 1.1482964149599263e-06,
+      "loss": 0.5024,
+      "step": 7187
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 1.1452354417769484e-06,
+      "loss": 0.6102,
+      "step": 7188
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 1.1421784581656075e-06,
+      "loss": 0.6519,
+      "step": 7189
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 1.139125464637175e-06,
+      "loss": 0.4676,
+      "step": 7190
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 1.136076461702243e-06,
+      "loss": 0.5001,
+      "step": 7191
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 1.1330314498707444e-06,
+      "loss": 0.4565,
+      "step": 7192
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 1.1299904296519281e-06,
+      "loss": 0.7886,
+      "step": 7193
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 1.1269534015543899e-06,
+      "loss": 0.6764,
+      "step": 7194
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 1.1239203660860648e-06,
+      "loss": 0.6039,
+      "step": 7195
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 1.120891323754203e-06,
+      "loss": 0.5633,
+      "step": 7196
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 1.1178662750653961e-06,
+      "loss": 0.8081,
+      "step": 7197
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 1.1148452205255677e-06,
+      "loss": 0.5982,
+      "step": 7198
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 1.1118281606399662e-06,
+      "loss": 0.5925,
+      "step": 7199
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 1.108815095913182e-06,
+      "loss": 0.7379,
+      "step": 7200
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 1.105806026849135e-06,
+      "loss": 0.8891,
+      "step": 7201
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 1.1028009539510692e-06,
+      "loss": 0.3293,
+      "step": 7202
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 1.0997998777215663e-06,
+      "loss": 0.6458,
+      "step": 7203
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 1.0968027986625469e-06,
+      "loss": 0.6252,
+      "step": 7204
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 1.0938097172752404e-06,
+      "loss": 0.4215,
+      "step": 7205
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 1.0908206340602356e-06,
+      "loss": 0.4357,
+      "step": 7206
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 1.0878355495174354e-06,
+      "loss": 0.7365,
+      "step": 7207
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 1.084854464146076e-06,
+      "loss": 0.6242,
+      "step": 7208
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 1.081877378444729e-06,
+      "loss": 0.7002,
+      "step": 7209
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 1.0789042929112925e-06,
+      "loss": 0.6938,
+      "step": 7210
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 1.0759352080429996e-06,
+      "loss": 0.6122,
+      "step": 7211
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 1.0729701243364144e-06,
+      "loss": 0.6567,
+      "step": 7212
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 1.070009042287426e-06,
+      "loss": 0.6458,
+      "step": 7213
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 1.0670519623912605e-06,
+      "loss": 0.6192,
+      "step": 7214
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 1.0640988851424699e-06,
+      "loss": 0.7391,
+      "step": 7215
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 1.061149811034945e-06,
+      "loss": 0.552,
+      "step": 7216
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 1.0582047405618966e-06,
+      "loss": 0.4261,
+      "step": 7217
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 1.05526367421587e-06,
+      "loss": 0.7838,
+      "step": 7218
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 1.0523266124887465e-06,
+      "loss": 0.7898,
+      "step": 7219
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 1.0493935558717276e-06,
+      "loss": 0.4699,
+      "step": 7220
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 1.0464645048553545e-06,
+      "loss": 0.5854,
+      "step": 7221
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 1.0435394599294907e-06,
+      "loss": 0.5799,
+      "step": 7222
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 1.0406184215833336e-06,
+      "loss": 0.6876,
+      "step": 7223
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 1.0377013903054094e-06,
+      "loss": 0.5205,
+      "step": 7224
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 1.0347883665835777e-06,
+      "loss": 0.5665,
+      "step": 7225
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 1.0318793509050239e-06,
+      "loss": 0.9115,
+      "step": 7226
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 1.0289743437562643e-06,
+      "loss": 0.6641,
+      "step": 7227
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 1.0260733456231407e-06,
+      "loss": 0.6607,
+      "step": 7228
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.023176356990832e-06,
+      "loss": 0.6111,
+      "step": 7229
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.0202833783438448e-06,
+      "loss": 0.386,
+      "step": 7230
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.0173944101660033e-06,
+      "loss": 0.3881,
+      "step": 7231
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.0145094529404875e-06,
+      "loss": 0.6862,
+      "step": 7232
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 1.0116285071497755e-06,
+      "loss": 0.6954,
+      "step": 7233
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 1.008751573275693e-06,
+      "loss": 0.6045,
+      "step": 7234
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 1.0058786517993913e-06,
+      "loss": 0.6738,
+      "step": 7235
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 1.0030097432013497e-06,
+      "loss": 0.4288,
+      "step": 7236
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 1.0001448479613734e-06,
+      "loss": 0.7593,
+      "step": 7237
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.97283966558607e-07,
+      "loss": 0.4552,
+      "step": 7238
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.944270994715122e-07,
+      "loss": 0.4588,
+      "step": 7239
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 9.915742471778816e-07,
+      "loss": 0.571,
+      "step": 7240
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.88725410154842e-07,
+      "loss": 0.5485,
+      "step": 7241
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.858805888788374e-07,
+      "loss": 0.6443,
+      "step": 7242
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.830397838256567e-07,
+      "loss": 0.6574,
+      "step": 7243
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 9.80202995470411e-07,
+      "loss": 0.5341,
+      "step": 7244
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.773702242875244e-07,
+      "loss": 0.5135,
+      "step": 7245
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.7454147075077e-07,
+      "loss": 0.4869,
+      "step": 7246
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.717167353332363e-07,
+      "loss": 0.4337,
+      "step": 7247
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 9.688960185073453e-07,
+      "loss": 0.4774,
+      "step": 7248
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 9.660793207448476e-07,
+      "loss": 0.6939,
+      "step": 7249
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 9.632666425168219e-07,
+      "loss": 0.6595,
+      "step": 7250
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 9.60457984293664e-07,
+      "loss": 0.4877,
+      "step": 7251
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 9.57653346545112e-07,
+      "loss": 0.7968,
+      "step": 7252
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 9.548527297402216e-07,
+      "loss": 0.4746,
+      "step": 7253
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 9.520561343473844e-07,
+      "loss": 0.8048,
+      "step": 7254
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 9.492635608343125e-07,
+      "loss": 0.7564,
+      "step": 7255
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 9.46475009668052e-07,
+      "loss": 0.6121,
+      "step": 7256
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 9.436904813149627e-07,
+      "loss": 0.815,
+      "step": 7257
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 9.409099762407447e-07,
+      "loss": 0.411,
+      "step": 7258
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 9.381334949104231e-07,
+      "loss": 0.7192,
+      "step": 7259
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 9.353610377883514e-07,
+      "loss": 0.5293,
+      "step": 7260
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 9.325926053382006e-07,
+      "loss": 0.5802,
+      "step": 7261
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 9.298281980229834e-07,
+      "loss": 0.7043,
+      "step": 7262
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 9.270678163050217e-07,
+      "loss": 0.6025,
+      "step": 7263
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 9.243114606459741e-07,
+      "loss": 0.3521,
+      "step": 7264
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 9.215591315068301e-07,
+      "loss": 0.6065,
+      "step": 7265
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 9.188108293478992e-07,
+      "loss": 0.78,
+      "step": 7266
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 9.160665546288194e-07,
+      "loss": 0.5707,
+      "step": 7267
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 9.133263078085597e-07,
+      "loss": 0.5457,
+      "step": 7268
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 9.105900893453978e-07,
+      "loss": 0.6529,
+      "step": 7269
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 9.078578996969594e-07,
+      "loss": 0.6005,
+      "step": 7270
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 9.051297393201874e-07,
+      "loss": 0.7783,
+      "step": 7271
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 9.024056086713472e-07,
+      "loss": 0.5248,
+      "step": 7272
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.996855082060384e-07,
+      "loss": 0.6723,
+      "step": 7273
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.969694383791805e-07,
+      "loss": 0.572,
+      "step": 7274
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.942573996450182e-07,
+      "loss": 0.4789,
+      "step": 7275
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 8.91549392457125e-07,
+      "loss": 0.7312,
+      "step": 7276
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.888454172684025e-07,
+      "loss": 0.498,
+      "step": 7277
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.861454745310722e-07,
+      "loss": 0.6667,
+      "step": 7278
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.83449564696684e-07,
+      "loss": 0.5306,
+      "step": 7279
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.807576882161134e-07,
+      "loss": 0.8163,
+      "step": 7280
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.780698455395642e-07,
+      "loss": 0.4415,
+      "step": 7281
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.753860371165573e-07,
+      "loss": 0.6433,
+      "step": 7282
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.727062633959476e-07,
+      "loss": 0.7032,
+      "step": 7283
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.7003052482591e-07,
+      "loss": 0.6535,
+      "step": 7284
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.673588218539502e-07,
+      "loss": 0.6263,
+      "step": 7285
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.646911549268888e-07,
+      "loss": 0.5733,
+      "step": 7286
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.620275244908827e-07,
+      "loss": 0.6893,
+      "step": 7287
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 8.593679309914088e-07,
+      "loss": 0.6237,
+      "step": 7288
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.567123748732642e-07,
+      "loss": 0.6514,
+      "step": 7289
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.540608565805769e-07,
+      "loss": 0.5487,
+      "step": 7290
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.51413376556806e-07,
+      "loss": 0.8135,
+      "step": 7291
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.487699352447143e-07,
+      "loss": 0.6888,
+      "step": 7292
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.461305330864117e-07,
+      "loss": 0.8509,
+      "step": 7293
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.434951705233201e-07,
+      "loss": 0.4791,
+      "step": 7294
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.408638479961844e-07,
+      "loss": 0.4651,
+      "step": 7295
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.38236565945083e-07,
+      "loss": 0.6158,
+      "step": 7296
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.356133248094172e-07,
+      "loss": 0.3772,
+      "step": 7297
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.329941250279e-07,
+      "loss": 0.542,
+      "step": 7298
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.303789670385837e-07,
+      "loss": 0.5065,
+      "step": 7299
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 8.277678512788351e-07,
+      "loss": 0.5626,
+      "step": 7300
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.251607781853466e-07,
+      "loss": 0.4996,
+      "step": 7301
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.225577481941443e-07,
+      "loss": 0.6495,
+      "step": 7302
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.199587617405663e-07,
+      "loss": 0.4826,
+      "step": 7303
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 8.173638192592731e-07,
+      "loss": 0.6277,
+      "step": 7304
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.147729211842564e-07,
+      "loss": 0.5358,
+      "step": 7305
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.121860679488307e-07,
+      "loss": 0.703,
+      "step": 7306
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.096032599856307e-07,
+      "loss": 0.6806,
+      "step": 7307
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 8.070244977266162e-07,
+      "loss": 0.5955,
+      "step": 7308
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.044497816030755e-07,
+      "loss": 0.534,
+      "step": 7309
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 8.018791120456087e-07,
+      "loss": 0.5458,
+      "step": 7310
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 7.993124894841441e-07,
+      "loss": 0.6849,
+      "step": 7311
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 7.967499143479384e-07,
+      "loss": 0.5992,
+      "step": 7312
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 7.941913870655654e-07,
+      "loss": 0.4733,
+      "step": 7313
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 7.916369080649272e-07,
+      "loss": 0.6086,
+      "step": 7314
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 7.890864777732487e-07,
+      "loss": 0.3976,
+      "step": 7315
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 7.865400966170638e-07,
+      "loss": 0.884,
+      "step": 7316
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 7.83997765022243e-07,
+      "loss": 0.6403,
+      "step": 7317
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 7.814594834139849e-07,
+      "loss": 0.7537,
+      "step": 7318
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 7.789252522167945e-07,
+      "loss": 0.7183,
+      "step": 7319
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 7.763950718545132e-07,
+      "loss": 0.6106,
+      "step": 7320
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 7.738689427502943e-07,
+      "loss": 0.6336,
+      "step": 7321
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 7.713468653266193e-07,
+      "loss": 0.4928,
+      "step": 7322
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 7.688288400052896e-07,
+      "loss": 0.5736,
+      "step": 7323
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 7.663148672074321e-07,
+      "loss": 0.6147,
+      "step": 7324
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.63804947353497e-07,
+      "loss": 0.7053,
+      "step": 7325
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.612990808632508e-07,
+      "loss": 0.6346,
+      "step": 7326
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.587972681557892e-07,
+      "loss": 0.6572,
+      "step": 7327
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.562995096495163e-07,
+      "loss": 0.5278,
+      "step": 7328
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.538058057621756e-07,
+      "loss": 0.5987,
+      "step": 7329
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.513161569108252e-07,
+      "loss": 0.6847,
+      "step": 7330
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 7.488305635118404e-07,
+      "loss": 0.6374,
+      "step": 7331
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.463490259809247e-07,
+      "loss": 0.5897,
+      "step": 7332
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.438715447331018e-07,
+      "loss": 0.5226,
+      "step": 7333
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.413981201827125e-07,
+      "loss": 0.565,
+      "step": 7334
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.389287527434286e-07,
+      "loss": 0.5092,
+      "step": 7335
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.364634428282307e-07,
+      "loss": 0.5914,
+      "step": 7336
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.340021908494337e-07,
+      "loss": 0.6355,
+      "step": 7337
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.315449972186639e-07,
+      "loss": 0.7005,
+      "step": 7338
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.290918623468729e-07,
+      "loss": 0.7215,
+      "step": 7339
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.266427866443354e-07,
+      "loss": 0.3883,
+      "step": 7340
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.24197770520646e-07,
+      "loss": 0.4369,
+      "step": 7341
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.217568143847136e-07,
+      "loss": 0.552,
+      "step": 7342
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 7.193199186447814e-07,
+      "loss": 0.6686,
+      "step": 7343
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.168870837084035e-07,
+      "loss": 0.6213,
+      "step": 7344
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.144583099824547e-07,
+      "loss": 0.609,
+      "step": 7345
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.120335978731351e-07,
+      "loss": 0.6692,
+      "step": 7346
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.096129477859647e-07,
+      "loss": 0.654,
+      "step": 7347
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.071963601257836e-07,
+      "loss": 0.4499,
+      "step": 7348
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.047838352967518e-07,
+      "loss": 0.5566,
+      "step": 7349
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.023753737023492e-07,
+      "loss": 0.8312,
+      "step": 7350
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 6.999709757453788e-07,
+      "loss": 0.614,
+      "step": 7351
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 6.975706418279576e-07,
+      "loss": 0.6671,
+      "step": 7352
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 6.951743723515342e-07,
+      "loss": 0.6372,
+      "step": 7353
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 6.927821677168655e-07,
+      "loss": 0.629,
+      "step": 7354
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 6.903940283240374e-07,
+      "loss": 0.5953,
+      "step": 7355
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 6.880099545724522e-07,
+      "loss": 0.604,
+      "step": 7356
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 6.856299468608301e-07,
+      "loss": 0.5559,
+      "step": 7357
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 6.832540055872139e-07,
+      "loss": 0.6153,
+      "step": 7358
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 6.808821311489688e-07,
+      "loss": 0.5417,
+      "step": 7359
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 6.785143239427721e-07,
+      "loss": 0.3759,
+      "step": 7360
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 6.761505843646348e-07,
+      "loss": 0.6886,
+      "step": 7361
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 6.737909128098685e-07,
+      "loss": 0.5851,
+      "step": 7362
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 6.714353096731213e-07,
+      "loss": 0.6522,
+      "step": 7363
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 6.690837753483531e-07,
+      "loss": 0.501,
+      "step": 7364
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 6.66736310228841e-07,
+      "loss": 0.6405,
+      "step": 7365
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 6.643929147071876e-07,
+      "loss": 0.5128,
+      "step": 7366
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 6.620535891753155e-07,
+      "loss": 0.4469,
+      "step": 7367
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 6.597183340244562e-07,
+      "loss": 0.5938,
+      "step": 7368
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 6.57387149645175e-07,
+      "loss": 0.7194,
+      "step": 7369
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 6.550600364273407e-07,
+      "loss": 0.4755,
+      "step": 7370
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 6.527369947601558e-07,
+      "loss": 0.6425,
+      "step": 7371
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 6.504180250321345e-07,
+      "loss": 0.5953,
+      "step": 7372
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 6.481031276311167e-07,
+      "loss": 0.7742,
+      "step": 7373
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 6.457923029442425e-07,
+      "loss": 0.7559,
+      "step": 7374
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 6.434855513579946e-07,
+      "loss": 0.4928,
+      "step": 7375
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 6.41182873258156e-07,
+      "loss": 0.5701,
+      "step": 7376
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 6.388842690298436e-07,
+      "loss": 0.6929,
+      "step": 7377
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 6.365897390574832e-07,
+      "loss": 0.6692,
+      "step": 7378
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 6.342992837248235e-07,
+      "loss": 0.5787,
+      "step": 7379
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 6.320129034149275e-07,
+      "loss": 0.6705,
+      "step": 7380
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 6.297305985101781e-07,
+      "loss": 0.6112,
+      "step": 7381
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 6.274523693922757e-07,
+      "loss": 0.5368,
+      "step": 7382
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 6.25178216442246e-07,
+      "loss": 0.4148,
+      "step": 7383
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 6.229081400404263e-07,
+      "loss": 0.6624,
+      "step": 7384
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 6.20642140566477e-07,
+      "loss": 0.5881,
+      "step": 7385
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 6.18380218399367e-07,
+      "loss": 0.5114,
+      "step": 7386
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 6.161223739173911e-07,
+      "loss": 0.5565,
+      "step": 7387
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 6.138686074981637e-07,
+      "loss": 0.5609,
+      "step": 7388
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 6.116189195186139e-07,
+      "loss": 0.4219,
+      "step": 7389
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 6.093733103549881e-07,
+      "loss": 0.5943,
+      "step": 7390
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 6.071317803828525e-07,
+      "loss": 0.7298,
+      "step": 7391
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 6.048943299770848e-07,
+      "loss": 0.5667,
+      "step": 7392
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 6.026609595118943e-07,
+      "loss": 0.6137,
+      "step": 7393
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 6.004316693607903e-07,
+      "loss": 0.7462,
+      "step": 7394
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 5.982064598966164e-07,
+      "loss": 0.4901,
+      "step": 7395
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 5.95985331491522e-07,
+      "loss": 0.6688,
+      "step": 7396
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 5.937682845169823e-07,
+      "loss": 0.4591,
+      "step": 7397
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 5.915553193437784e-07,
+      "loss": 0.5601,
+      "step": 7398
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 5.893464363420198e-07,
+      "loss": 0.6391,
+      "step": 7399
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 5.871416358811305e-07,
+      "loss": 0.5135,
+      "step": 7400
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 5.849409183298488e-07,
+      "loss": 0.5493,
+      "step": 7401
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 5.82744284056233e-07,
+      "loss": 0.4564,
+      "step": 7402
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 5.805517334276562e-07,
+      "loss": 0.5703,
+      "step": 7403
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 5.783632668108107e-07,
+      "loss": 0.4925,
+      "step": 7404
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 5.761788845717069e-07,
+      "loss": 0.5261,
+      "step": 7405
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 5.739985870756665e-07,
+      "loss": 0.6133,
+      "step": 7406
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 5.718223746873363e-07,
+      "loss": 0.7299,
+      "step": 7407
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 5.6965024777067e-07,
+      "loss": 0.58,
+      "step": 7408
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 5.674822066889462e-07,
+      "loss": 0.5038,
+      "step": 7409
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 5.653182518047556e-07,
+      "loss": 0.7857,
+      "step": 7410
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 5.631583834800113e-07,
+      "loss": 0.5909,
+      "step": 7411
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 5.610026020759329e-07,
+      "loss": 0.6256,
+      "step": 7412
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 5.588509079530651e-07,
+      "loss": 0.6422,
+      "step": 7413
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 5.567033014712674e-07,
+      "loss": 0.5452,
+      "step": 7414
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 5.545597829897165e-07,
+      "loss": 0.5724,
+      "step": 7415
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 5.524203528668975e-07,
+      "loss": 0.6967,
+      "step": 7416
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 5.502850114606217e-07,
+      "loss": 0.6595,
+      "step": 7417
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 5.481537591280117e-07,
+      "loss": 0.5736,
+      "step": 7418
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 5.460265962255073e-07,
+      "loss": 0.4917,
+      "step": 7419
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 5.43903523108863e-07,
+      "loss": 0.6957,
+      "step": 7420
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 5.417845401331528e-07,
+      "loss": 0.5931,
+      "step": 7421
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 5.39669647652763e-07,
+      "loss": 0.6105,
+      "step": 7422
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 5.375588460213993e-07,
+      "loss": 0.5098,
+      "step": 7423
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 5.354521355920767e-07,
+      "loss": 0.3663,
+      "step": 7424
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 5.333495167171353e-07,
+      "loss": 0.7439,
+      "step": 7425
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 5.312509897482221e-07,
+      "loss": 0.5154,
+      "step": 7426
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 5.291565550363031e-07,
+      "loss": 0.5853,
+      "step": 7427
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 5.27066212931665e-07,
+      "loss": 0.6584,
+      "step": 7428
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 5.249799637839032e-07,
+      "loss": 0.5318,
+      "step": 7429
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 5.228978079419272e-07,
+      "loss": 0.5972,
+      "step": 7430
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 5.208197457539726e-07,
+      "loss": 0.5176,
+      "step": 7431
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 5.187457775675781e-07,
+      "loss": 0.2913,
+      "step": 7432
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 5.166759037296049e-07,
+      "loss": 0.7987,
+      "step": 7433
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 5.146101245862261e-07,
+      "loss": 0.7064,
+      "step": 7434
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 5.125484404829295e-07,
+      "loss": 0.6125,
+      "step": 7435
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 5.104908517645251e-07,
+      "loss": 0.7943,
+      "step": 7436
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 5.084373587751323e-07,
+      "loss": 0.7825,
+      "step": 7437
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 5.063879618581818e-07,
+      "loss": 0.5075,
+      "step": 7438
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 5.043426613564245e-07,
+      "loss": 0.6239,
+      "step": 7439
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 5.023014576119284e-07,
+      "loss": 0.6511,
+      "step": 7440
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 5.002643509660649e-07,
+      "loss": 0.6354,
+      "step": 7441
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 4.982313417595391e-07,
+      "loss": 0.8064,
+      "step": 7442
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 4.962024303323543e-07,
+      "loss": 0.4824,
+      "step": 7443
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 4.94177617023836e-07,
+      "loss": 0.5573,
+      "step": 7444
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 4.921569021726191e-07,
+      "loss": 0.5486,
+      "step": 7445
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 4.901402861166581e-07,
+      "loss": 0.6056,
+      "step": 7446
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 4.881277691932224e-07,
+      "loss": 0.5699,
+      "step": 7447
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 4.861193517388923e-07,
+      "loss": 0.6705,
+      "step": 7448
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 4.841150340895662e-07,
+      "loss": 0.5903,
+      "step": 7449
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 4.821148165804507e-07,
+      "loss": 0.6285,
+      "step": 7450
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 4.801186995460699e-07,
+      "loss": 0.5004,
+      "step": 7451
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 4.78126683320268e-07,
+      "loss": 0.4793,
+      "step": 7452
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 4.7613876823619784e-07,
+      "loss": 0.3562,
+      "step": 7453
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 4.741549546263241e-07,
+      "loss": 0.6656,
+      "step": 7454
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 4.7217524282242853e-07,
+      "loss": 0.662,
+      "step": 7455
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 4.701996331556074e-07,
+      "loss": 0.7742,
+      "step": 7456
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 4.68228125956266e-07,
+      "loss": 0.7715,
+      "step": 7457
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 4.6626072155413204e-07,
+      "loss": 0.7545,
+      "step": 7458
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 4.642974202782424e-07,
+      "loss": 0.742,
+      "step": 7459
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 4.623382224569456e-07,
+      "loss": 0.8229,
+      "step": 7460
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 4.6038312841791e-07,
+      "loss": 0.535,
+      "step": 7461
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 4.584321384881074e-07,
+      "loss": 0.6622,
+      "step": 7462
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 4.564852529938379e-07,
+      "loss": 0.3278,
+      "step": 7463
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 4.5454247226069944e-07,
+      "loss": 0.4664,
+      "step": 7464
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 4.526037966136154e-07,
+      "loss": 0.717,
+      "step": 7465
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 4.506692263768153e-07,
+      "loss": 0.6423,
+      "step": 7466
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 4.4873876187384587e-07,
+      "loss": 0.6388,
+      "step": 7467
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 4.468124034275656e-07,
+      "loss": 0.5445,
+      "step": 7468
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 4.4489015136015e-07,
+      "loss": 0.7596,
+      "step": 7469
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 4.429720059930809e-07,
+      "loss": 0.5609,
+      "step": 7470
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 4.4105796764715714e-07,
+      "loss": 0.5926,
+      "step": 7471
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 4.391480366424949e-07,
+      "loss": 0.5319,
+      "step": 7472
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 4.372422132985138e-07,
+      "loss": 0.75,
+      "step": 7473
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 4.353404979339532e-07,
+      "loss": 0.6544,
+      "step": 7474
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 4.3344289086686696e-07,
+      "loss": 0.6327,
+      "step": 7475
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 4.315493924146152e-07,
+      "loss": 0.5865,
+      "step": 7476
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 4.296600028938752e-07,
+      "loss": 0.5995,
+      "step": 7477
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 4.2777472262063856e-07,
+      "loss": 0.7391,
+      "step": 7478
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 4.2589355191020586e-07,
+      "loss": 0.5202,
+      "step": 7479
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 4.2401649107719486e-07,
+      "loss": 0.6362,
+      "step": 7480
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 4.2214354043552674e-07,
+      "loss": 0.6458,
+      "step": 7481
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 4.202747002984481e-07,
+      "loss": 0.5832,
+      "step": 7482
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 4.1840997097850887e-07,
+      "loss": 0.5488,
+      "step": 7483
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 4.1654935278757636e-07,
+      "loss": 0.604,
+      "step": 7484
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 4.1469284603682377e-07,
+      "loss": 0.6122,
+      "step": 7485
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 4.128404510367473e-07,
+      "loss": 0.644,
+      "step": 7486
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 4.109921680971435e-07,
+      "loss": 0.8778,
+      "step": 7487
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 4.0914799752712895e-07,
+      "loss": 0.509,
+      "step": 7488
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 4.0730793963512923e-07,
+      "loss": 0.5594,
+      "step": 7489
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 4.054719947288899e-07,
+      "loss": 0.591,
+      "step": 7490
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 4.0364016311545683e-07,
+      "loss": 0.619,
+      "step": 7491
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 4.0181244510119063e-07,
+      "loss": 0.6984,
+      "step": 7492
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 3.999888409917718e-07,
+      "loss": 0.6763,
+      "step": 7493
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 3.9816935109218413e-07,
+      "loss": 0.5164,
+      "step": 7494
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 3.963539757067314e-07,
+      "loss": 0.6097,
+      "step": 7495
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 3.9454271513902075e-07,
+      "loss": 0.4905,
+      "step": 7496
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 3.927355696919738e-07,
+      "loss": 0.6653,
+      "step": 7497
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 3.909325396678293e-07,
+      "loss": 0.5478,
+      "step": 7498
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 3.891336253681321e-07,
+      "loss": 0.5343,
+      "step": 7499
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 3.8733882709373617e-07,
+      "loss": 0.5005,
+      "step": 7500
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 3.8554814514481505e-07,
+      "loss": 0.5466,
+      "step": 7501
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 3.837615798208544e-07,
+      "loss": 0.428,
+      "step": 7502
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 3.8197913142064e-07,
+      "loss": 0.5349,
+      "step": 7503
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 3.80200800242278e-07,
+      "loss": 0.6826,
+      "step": 7504
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 3.784265865831804e-07,
+      "loss": 0.5094,
+      "step": 7505
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 3.7665649074008213e-07,
+      "loss": 0.6345,
+      "step": 7506
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 3.7489051300901576e-07,
+      "loss": 0.6333,
+      "step": 7507
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 3.731286536853368e-07,
+      "loss": 0.5319,
+      "step": 7508
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 3.713709130636983e-07,
+      "loss": 0.488,
+      "step": 7509
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 3.6961729143807355e-07,
+      "loss": 0.6631,
+      "step": 7510
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 3.6786778910175015e-07,
+      "loss": 0.4934,
+      "step": 7511
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 3.6612240634731897e-07,
+      "loss": 0.7061,
+      "step": 7512
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 3.6438114346668826e-07,
+      "loss": 0.395,
+      "step": 7513
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 3.626440007510723e-07,
+      "loss": 0.7032,
+      "step": 7514
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 3.6091097849099696e-07,
+      "loss": 0.4931,
+      "step": 7515
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 3.5918207697629714e-07,
+      "loss": 0.4397,
+      "step": 7516
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 3.5745729649613034e-07,
+      "loss": 0.5591,
+      "step": 7517
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 3.5573663733894924e-07,
+      "loss": 0.6895,
+      "step": 7518
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 3.5402009979252916e-07,
+      "loss": 0.6129,
+      "step": 7519
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 3.5230768414394597e-07,
+      "loss": 0.7356,
+      "step": 7520
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 3.505993906795929e-07,
+      "loss": 0.4718,
+      "step": 7521
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 3.4889521968517467e-07,
+      "loss": 0.6293,
+      "step": 7522
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 3.471951714457022e-07,
+      "loss": 0.5409,
+      "step": 7523
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 3.4549924624549813e-07,
+      "loss": 0.4538,
+      "step": 7524
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 3.4380744436819654e-07,
+      "loss": 0.3452,
+      "step": 7525
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 3.4211976609674344e-07,
+      "loss": 0.4761,
+      "step": 7526
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 3.404362117133908e-07,
+      "loss": 0.669,
+      "step": 7527
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 3.3875678149970504e-07,
+      "loss": 0.4697,
+      "step": 7528
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 3.370814757365587e-07,
+      "loss": 0.5783,
+      "step": 7529
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 3.3541029470413877e-07,
+      "loss": 0.7093,
+      "step": 7530
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 3.337432386819411e-07,
+      "loss": 0.6121,
+      "step": 7531
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 3.3208030794877055e-07,
+      "loss": 0.6204,
+      "step": 7532
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 3.304215027827434e-07,
+      "loss": 0.5696,
+      "step": 7533
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 3.2876682346128505e-07,
+      "loss": 0.663,
+      "step": 7534
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 3.2711627026112967e-07,
+      "loss": 0.6404,
+      "step": 7535
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 3.254698434583231e-07,
+      "loss": 0.539,
+      "step": 7536
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 3.238275433282229e-07,
+      "loss": 0.5507,
+      "step": 7537
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 3.2218937014549257e-07,
+      "loss": 0.5454,
+      "step": 7538
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 3.205553241841047e-07,
+      "loss": 0.562,
+      "step": 7539
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 3.189254057173491e-07,
+      "loss": 0.5898,
+      "step": 7540
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 3.172996150178159e-07,
+      "loss": 0.627,
+      "step": 7541
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 3.156779523574127e-07,
+      "loss": 0.7563,
+      "step": 7542
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 3.140604180073503e-07,
+      "loss": 0.5929,
+      "step": 7543
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 3.12447012238154e-07,
+      "loss": 0.5656,
+      "step": 7544
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 3.108377353196551e-07,
+      "loss": 0.6167,
+      "step": 7545
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 3.092325875209967e-07,
+      "loss": 0.5775,
+      "step": 7546
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 3.0763156911063074e-07,
+      "loss": 0.5528,
+      "step": 7547
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 3.06034680356318e-07,
+      "loss": 0.55,
+      "step": 7548
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 3.0444192152512816e-07,
+      "loss": 0.5079,
+      "step": 7549
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 3.028532928834427e-07,
+      "loss": 0.6115,
+      "step": 7550
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 3.012687946969489e-07,
+      "loss": 0.4966,
+      "step": 7551
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.9968842723064603e-07,
+      "loss": 0.5147,
+      "step": 7552
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.9811219074883654e-07,
+      "loss": 0.7602,
+      "step": 7553
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.9654008551514847e-07,
+      "loss": 0.8217,
+      "step": 7554
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.949721117924964e-07,
+      "loss": 0.8007,
+      "step": 7555
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.9340826984311785e-07,
+      "loss": 0.5667,
+      "step": 7556
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.918485599285592e-07,
+      "loss": 0.6385,
+      "step": 7557
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.9029298230967006e-07,
+      "loss": 0.5382,
+      "step": 7558
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.887415372466118e-07,
+      "loss": 0.813,
+      "step": 7559
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.871942249988574e-07,
+      "loss": 0.5684,
+      "step": 7560
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.856510458251832e-07,
+      "loss": 0.5414,
+      "step": 7561
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.841119999836772e-07,
+      "loss": 0.6878,
+      "step": 7562
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.825770877317363e-07,
+      "loss": 0.5483,
+      "step": 7563
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.810463093260635e-07,
+      "loss": 0.7072,
+      "step": 7564
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.795196650226789e-07,
+      "loss": 0.6728,
+      "step": 7565
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.779971550769006e-07,
+      "loss": 0.5588,
+      "step": 7566
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.76478779743361e-07,
+      "loss": 0.4685,
+      "step": 7567
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7496453927599586e-07,
+      "loss": 0.6982,
+      "step": 7568
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7345443392805534e-07,
+      "loss": 0.5602,
+      "step": 7569
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7194846395209863e-07,
+      "loss": 0.5044,
+      "step": 7570
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.704466295999908e-07,
+      "loss": 0.6641,
+      "step": 7571
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.6894893112290043e-07,
+      "loss": 0.434,
+      "step": 7572
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.674553687713105e-07,
+      "loss": 0.5385,
+      "step": 7573
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.6596594279501283e-07,
+      "loss": 0.4406,
+      "step": 7574
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.644806534430999e-07,
+      "loss": 0.4312,
+      "step": 7575
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.6299950096398406e-07,
+      "loss": 0.8384,
+      "step": 7576
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.615224856053783e-07,
+      "loss": 0.6178,
+      "step": 7577
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.6004960761430165e-07,
+      "loss": 0.4808,
+      "step": 7578
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.5858086723708764e-07,
+      "loss": 0.3962,
+      "step": 7579
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.5711626471937313e-07,
+      "loss": 0.5569,
+      "step": 7580
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.556558003061038e-07,
+      "loss": 0.691,
+      "step": 7581
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.541994742415371e-07,
+      "loss": 0.6964,
+      "step": 7582
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.52747286769231e-07,
+      "loss": 0.4963,
+      "step": 7583
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.5129923813206056e-07,
+      "loss": 0.6007,
+      "step": 7584
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.4985532857219604e-07,
+      "loss": 0.6548,
+      "step": 7585
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.484155583311276e-07,
+      "loss": 0.4838,
+      "step": 7586
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.469799276496515e-07,
+      "loss": 0.4912,
+      "step": 7587
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.4554843676786455e-07,
+      "loss": 0.445,
+      "step": 7588
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.4412108592517524e-07,
+      "loss": 0.7721,
+      "step": 7589
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.4269787536030097e-07,
+      "loss": 0.6761,
+      "step": 7590
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.4127880531126524e-07,
+      "loss": 0.602,
+      "step": 7591
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.3986387601539764e-07,
+      "loss": 0.5577,
+      "step": 7592
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.384530877093422e-07,
+      "loss": 0.8001,
+      "step": 7593
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.3704644062903793e-07,
+      "loss": 0.6102,
+      "step": 7594
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.3564393500974391e-07,
+      "loss": 0.5894,
+      "step": 7595
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.3424557108601974e-07,
+      "loss": 0.7376,
+      "step": 7596
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.3285134909173112e-07,
+      "loss": 0.8892,
+      "step": 7597
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.3146126926005818e-07,
+      "loss": 0.3265,
+      "step": 7598
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.3007533182347885e-07,
+      "loss": 0.6433,
+      "step": 7599
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.2869353701378825e-07,
+      "loss": 0.6179,
+      "step": 7600
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.2731588506207934e-07,
+      "loss": 0.4225,
+      "step": 7601
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.259423761987567e-07,
+      "loss": 0.4291,
+      "step": 7602
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.2457301065353377e-07,
+      "loss": 0.7349,
+      "step": 7603
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.2320778865543012e-07,
+      "loss": 0.629,
+      "step": 7604
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.2184671043276872e-07,
+      "loss": 0.6883,
+      "step": 7605
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.2048977621318135e-07,
+      "loss": 0.6924,
+      "step": 7606
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.1913698622361144e-07,
+      "loss": 0.6161,
+      "step": 7607
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.177883406903003e-07,
+      "loss": 0.6363,
+      "step": 7608
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.1644383983880357e-07,
+      "loss": 0.6435,
+      "step": 7609
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.1510348389398305e-07,
+      "loss": 0.6222,
+      "step": 7610
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.1376727308000387e-07,
+      "loss": 0.7309,
+      "step": 7611
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.124352076203373e-07,
+      "loss": 0.5398,
+      "step": 7612
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.1110728773776622e-07,
+      "loss": 0.4274,
+      "step": 7613
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.0978351365437687e-07,
+      "loss": 0.7743,
+      "step": 7614
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.0846388559156438e-07,
+      "loss": 0.7877,
+      "step": 7615
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.071484037700272e-07,
+      "loss": 0.4547,
+      "step": 7616
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.0583706840977268e-07,
+      "loss": 0.5865,
+      "step": 7617
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.0452987973011151e-07,
+      "loss": 0.5705,
+      "step": 7618
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.032268379496688e-07,
+      "loss": 0.6748,
+      "step": 7619
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.0192794328636743e-07,
+      "loss": 0.5156,
+      "step": 7620
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.0063319595743923e-07,
+      "loss": 0.5543,
+      "step": 7621
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 1.9934259617942485e-07,
+      "loss": 0.8959,
+      "step": 7622
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 1.9805614416816553e-07,
+      "loss": 0.654,
+      "step": 7623
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 1.967738401388225e-07,
+      "loss": 0.6578,
+      "step": 7624
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 1.9549568430584365e-07,
+      "loss": 0.5972,
+      "step": 7625
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 1.9422167688299963e-07,
+      "loss": 0.3802,
+      "step": 7626
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 1.9295181808335616e-07,
+      "loss": 0.3747,
+      "step": 7627
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 1.9168610811929332e-07,
+      "loss": 0.6804,
+      "step": 7628
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 1.904245472024918e-07,
+      "loss": 0.6913,
+      "step": 7629
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 1.8916713554394394e-07,
+      "loss": 0.6005,
+      "step": 7630
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 1.8791387335393985e-07,
+      "loss": 0.659,
+      "step": 7631
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 1.8666476084208129e-07,
+      "loss": 0.424,
+      "step": 7632
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 1.8541979821727616e-07,
+      "loss": 0.7465,
+      "step": 7633
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 1.8417898568773572e-07,
+      "loss": 0.4512,
+      "step": 7634
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 1.8294232346098284e-07,
+      "loss": 0.4544,
+      "step": 7635
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 1.8170981174384094e-07,
+      "loss": 0.5622,
+      "step": 7636
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 1.8048145074243682e-07,
+      "loss": 0.5485,
+      "step": 7637
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 1.7925724066221173e-07,
+      "loss": 0.6351,
+      "step": 7638
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 1.7803718170790185e-07,
+      "loss": 0.6419,
+      "step": 7639
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 1.7682127408355787e-07,
+      "loss": 0.5268,
+      "step": 7640
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 1.7560951799253655e-07,
+      "loss": 0.5096,
+      "step": 7641
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 1.7440191363749249e-07,
+      "loss": 0.4689,
+      "step": 7642
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 1.7319846122039185e-07,
+      "loss": 0.4199,
+      "step": 7643
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 1.7199916094250423e-07,
+      "loss": 0.4698,
+      "step": 7644
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 1.7080401300441084e-07,
+      "loss": 0.6908,
+      "step": 7645
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 1.6961301760598514e-07,
+      "loss": 0.6589,
+      "step": 7646
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 1.6842617494642054e-07,
+      "loss": 0.4696,
+      "step": 7647
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 1.6724348522421107e-07,
+      "loss": 0.7706,
+      "step": 7648
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 1.6606494863714572e-07,
+      "loss": 0.4508,
+      "step": 7649
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 1.6489056538233627e-07,
+      "loss": 0.7859,
+      "step": 7650
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 1.6372033565618672e-07,
+      "loss": 0.7567,
+      "step": 7651
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 1.625542596544155e-07,
+      "loss": 0.589,
+      "step": 7652
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 1.613923375720361e-07,
+      "loss": 0.7968,
+      "step": 7653
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 1.602345696033819e-07,
+      "loss": 0.3964,
+      "step": 7654
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 1.5908095594207583e-07,
+      "loss": 0.7036,
+      "step": 7655
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 1.5793149678105523e-07,
+      "loss": 0.5242,
+      "step": 7656
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 1.5678619231255798e-07,
+      "loss": 0.5785,
+      "step": 7657
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 1.556450427281364e-07,
+      "loss": 0.704,
+      "step": 7658
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 1.5450804821863507e-07,
+      "loss": 0.5895,
+      "step": 7659
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 1.5337520897421298e-07,
+      "loss": 0.348,
+      "step": 7660
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.5224652518432691e-07,
+      "loss": 0.5941,
+      "step": 7661
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.5112199703774809e-07,
+      "loss": 0.7808,
+      "step": 7662
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.5000162472254275e-07,
+      "loss": 0.5632,
+      "step": 7663
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.4888540842608878e-07,
+      "loss": 0.5401,
+      "step": 7664
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 1.477733483350674e-07,
+      "loss": 0.6556,
+      "step": 7665
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 1.4666544463546317e-07,
+      "loss": 0.5956,
+      "step": 7666
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 1.4556169751256955e-07,
+      "loss": 0.7728,
+      "step": 7667
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 1.4446210715097775e-07,
+      "loss": 0.525,
+      "step": 7668
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 1.433666737345879e-07,
+      "loss": 0.6635,
+      "step": 7669
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 1.42275397446609e-07,
+      "loss": 0.5711,
+      "step": 7670
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 1.4118827846954785e-07,
+      "loss": 0.4758,
+      "step": 7671
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 1.4010531698521735e-07,
+      "loss": 0.7318,
+      "step": 7672
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 1.3902651317473924e-07,
+      "loss": 0.4811,
+      "step": 7673
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 1.3795186721853591e-07,
+      "loss": 0.6696,
+      "step": 7674
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 1.368813792963386e-07,
+      "loss": 0.5267,
+      "step": 7675
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 1.3581504958717628e-07,
+      "loss": 0.8135,
+      "step": 7676
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 1.3475287826938686e-07,
+      "loss": 0.4333,
+      "step": 7677
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 1.336948655206144e-07,
+      "loss": 0.6456,
+      "step": 7678
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 1.3264101151780616e-07,
+      "loss": 0.6918,
+      "step": 7679
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 1.3159131643721012e-07,
+      "loss": 0.6387,
+      "step": 7680
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 1.3054578045438305e-07,
+      "loss": 0.624,
+      "step": 7681
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 1.2950440374418226e-07,
+      "loss": 0.5675,
+      "step": 7682
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 1.284671864807796e-07,
+      "loss": 0.6929,
+      "step": 7683
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 1.2743412883763627e-07,
+      "loss": 0.6202,
+      "step": 7684
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 1.2640523098752799e-07,
+      "loss": 0.6419,
+      "step": 7685
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 1.2538049310253097e-07,
+      "loss": 0.5402,
+      "step": 7686
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 1.243599153540248e-07,
+      "loss": 0.8055,
+      "step": 7687
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 1.2334349791270074e-07,
+      "loss": 0.6851,
+      "step": 7688
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 1.2233124094854498e-07,
+      "loss": 0.8509,
+      "step": 7689
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 1.213231446308527e-07,
+      "loss": 0.4906,
+      "step": 7690
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 1.2031920912822236e-07,
+      "loss": 0.4653,
+      "step": 7691
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 1.1931943460855577e-07,
+      "loss": 0.6121,
+      "step": 7692
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 1.1832382123905805e-07,
+      "loss": 0.3844,
+      "step": 7693
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 1.1733236918624325e-07,
+      "loss": 0.534,
+      "step": 7694
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 1.1634507861592314e-07,
+      "loss": 0.5046,
+      "step": 7695
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 1.1536194969321845e-07,
+      "loss": 0.5547,
+      "step": 7696
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.1438298258255043e-07,
+      "loss": 0.4997,
+      "step": 7697
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.1340817744764365e-07,
+      "loss": 0.6472,
+      "step": 7698
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.124375344515316e-07,
+      "loss": 0.4826,
+      "step": 7699
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.1147105375655109e-07,
+      "loss": 0.6312,
+      "step": 7700
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 1.1050873552433394e-07,
+      "loss": 0.5246,
+      "step": 7701
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 1.095505799158264e-07,
+      "loss": 0.7017,
+      "step": 7702
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 1.0859658709127529e-07,
+      "loss": 0.6722,
+      "step": 7703
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 1.0764675721022522e-07,
+      "loss": 0.5926,
+      "step": 7704
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.0670109043153797e-07,
+      "loss": 0.5363,
+      "step": 7705
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.0575958691336485e-07,
+      "loss": 0.5382,
+      "step": 7706
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.0482224681316599e-07,
+      "loss": 0.6826,
+      "step": 7707
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.0388907028771044e-07,
+      "loss": 0.5955,
+      "step": 7708
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.0296005749306504e-07,
+      "loss": 0.4534,
+      "step": 7709
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.0203520858459992e-07,
+      "loss": 0.6024,
+      "step": 7710
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.0111452371699415e-07,
+      "loss": 0.3934,
+      "step": 7711
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.0019800304422455e-07,
+      "loss": 0.8787,
+      "step": 7712
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 9.928564671957408e-08,
+      "loss": 0.6327,
+      "step": 7713
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 9.8377454895629e-08,
+      "loss": 0.7566,
+      "step": 7714
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 9.747342772427892e-08,
+      "loss": 0.7042,
+      "step": 7715
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 9.657356535671957e-08,
+      "loss": 0.6186,
+      "step": 7716
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 9.567786794344446e-08,
+      "loss": 0.6323,
+      "step": 7717
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 9.478633563425599e-08,
+      "loss": 0.4971,
+      "step": 7718
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 9.389896857825709e-08,
+      "loss": 0.5801,
+      "step": 7719
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 9.301576692385405e-08,
+      "loss": 0.6106,
+      "step": 7720
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 9.213673081875651e-08,
+      "loss": 0.7002,
+      "step": 7721
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 9.126186040998297e-08,
+      "loss": 0.63,
+      "step": 7722
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 9.039115584384417e-08,
+      "loss": 0.6668,
+      "step": 7723
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 8.952461726596528e-08,
+      "loss": 0.527,
+      "step": 7724
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 8.866224482126651e-08,
+      "loss": 0.5964,
+      "step": 7725
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 8.780403865397691e-08,
+      "loss": 0.6865,
+      "step": 7726
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 8.694999890762611e-08,
+      "loss": 0.6348,
+      "step": 7727
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 8.610012572504711e-08,
+      "loss": 0.5833,
+      "step": 7728
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 8.525441924837619e-08,
+      "loss": 0.5182,
+      "step": 7729
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 8.44128796190502e-08,
+      "loss": 0.5501,
+      "step": 7730
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 8.357550697781768e-08,
+      "loss": 0.4993,
+      "step": 7731
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 8.274230146471939e-08,
+      "loss": 0.5895,
+      "step": 7732
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 8.191326321910774e-08,
+      "loss": 0.6356,
+      "step": 7733
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 8.108839237963017e-08,
+      "loss": 0.6926,
+      "step": 7734
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 8.026768908424576e-08,
+      "loss": 0.7193,
+      "step": 7735
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 7.94511534702086e-08,
+      "loss": 0.3855,
+      "step": 7736
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 7.863878567408444e-08,
+      "loss": 0.4413,
+      "step": 7737
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 7.78305858317313e-08,
+      "loss": 0.5426,
+      "step": 7738
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 7.70265540783216e-08,
+      "loss": 0.664,
+      "step": 7739
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 7.622669054831999e-08,
+      "loss": 0.6143,
+      "step": 7740
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 7.54309953755028e-08,
+      "loss": 0.6115,
+      "step": 7741
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 7.463946869294414e-08,
+      "loss": 0.669,
+      "step": 7742
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 7.385211063302422e-08,
+      "loss": 0.6547,
+      "step": 7743
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 7.306892132741827e-08,
+      "loss": 0.4327,
+      "step": 7744
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 7.228990090711596e-08,
+      "loss": 0.552,
+      "step": 7745
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 7.151504950240196e-08,
+      "loss": 0.8173,
+      "step": 7746
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 7.074436724286704e-08,
+      "loss": 0.6069,
+      "step": 7747
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 6.997785425739978e-08,
+      "loss": 0.6635,
+      "step": 7748
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 6.921551067419762e-08,
+      "loss": 0.6396,
+      "step": 7749
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 6.84573366207586e-08,
+      "loss": 0.6223,
+      "step": 7750
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 6.770333222388126e-08,
+      "loss": 0.5967,
+      "step": 7751
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 6.695349760967029e-08,
+      "loss": 0.6104,
+      "step": 7752
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 6.620783290353094e-08,
+      "loss": 0.5658,
+      "step": 7753
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 6.546633823016901e-08,
+      "loss": 0.6073,
+      "step": 7754
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 6.472901371359919e-08,
+      "loss": 0.5459,
+      "step": 7755
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 6.399585947713116e-08,
+      "loss": 0.3642,
+      "step": 7756
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 6.326687564338074e-08,
+      "loss": 0.6875,
+      "step": 7757
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 6.254206233426984e-08,
+      "loss": 0.5833,
+      "step": 7758
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 6.18214196710154e-08,
+      "loss": 0.6424,
+      "step": 7759
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 6.110494777414322e-08,
+      "loss": 0.4937,
+      "step": 7760
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 6.039264676347966e-08,
+      "loss": 0.6312,
+      "step": 7761
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 5.96845167581489e-08,
+      "loss": 0.514,
+      "step": 7762
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 5.8980557876583966e-08,
+      "loss": 0.4477,
+      "step": 7763
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 5.8280770236518456e-08,
+      "loss": 0.5919,
+      "step": 7764
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 5.758515395498931e-08,
+      "loss": 0.7166,
+      "step": 7765
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 5.689370914833403e-08,
+      "loss": 0.4776,
+      "step": 7766
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 5.6206435932187886e-08,
+      "loss": 0.6452,
+      "step": 7767
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 5.5523334421497816e-08,
+      "loss": 0.5785,
+      "step": 7768
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 5.484440473050856e-08,
+      "loss": 0.7768,
+      "step": 7769
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 5.416964697276261e-08,
+      "loss": 0.7484,
+      "step": 7770
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 5.349906126111692e-08,
+      "loss": 0.4936,
+      "step": 7771
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 5.2832647707717895e-08,
+      "loss": 0.5669,
+      "step": 7772
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 5.217040642402082e-08,
+      "loss": 0.6859,
+      "step": 7773
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 5.1512337520778775e-08,
+      "loss": 0.6562,
+      "step": 7774
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 5.0858441108056486e-08,
+      "loss": 0.5912,
+      "step": 7775
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 5.020871729520815e-08,
+      "loss": 0.6734,
+      "step": 7776
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 4.9563166190899625e-08,
+      "loss": 0.6146,
+      "step": 7777
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 4.892178790309454e-08,
+      "loss": 0.5228,
+      "step": 7778
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 4.82845825390571e-08,
+      "loss": 0.4019,
+      "step": 7779
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 4.765155020536316e-08,
+      "loss": 0.6588,
+      "step": 7780
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 4.7022691007878036e-08,
+      "loss": 0.5829,
+      "step": 7781
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 4.639800505177594e-08,
+      "loss": 0.5106,
+      "step": 7782
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 4.5777492441531625e-08,
+      "loss": 0.5511,
+      "step": 7783
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 4.5161153280923184e-08,
+      "loss": 0.5582,
+      "step": 7784
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 4.454898767302929e-08,
+      "loss": 0.421,
+      "step": 7785
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 4.3940995720231914e-08,
+      "loss": 0.5867,
+      "step": 7786
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 4.333717752421362e-08,
+      "loss": 0.7233,
+      "step": 7787
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 4.27375331859603e-08,
+      "loss": 0.5639,
+      "step": 7788
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 4.214206280575561e-08,
+      "loss": 0.6098,
+      "step": 7789
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 4.155076648319212e-08,
+      "loss": 0.7426,
+      "step": 7790
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 4.096364431716293e-08,
+      "loss": 0.4965,
+      "step": 7791
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 4.03806964058534e-08,
+      "loss": 0.6665,
+      "step": 7792
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 3.9801922846766095e-08,
+      "loss": 0.4548,
+      "step": 7793
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 3.922732373669025e-08,
+      "loss": 0.5596,
+      "step": 7794
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 3.865689917172954e-08,
+      "loss": 0.6377,
+      "step": 7795
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 3.809064924728267e-08,
+      "loss": 0.5019,
+      "step": 7796
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 3.7528574058051656e-08,
+      "loss": 0.5321,
+      "step": 7797
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 3.6970673698041856e-08,
+      "loss": 0.4466,
+      "step": 7798
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 3.641694826055364e-08,
+      "loss": 0.5677,
+      "step": 7799
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 3.586739783820181e-08,
+      "loss": 0.4928,
+      "step": 7800
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 3.5322022522887855e-08,
+      "loss": 0.5286,
+      "step": 7801
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 3.4780822405830474e-08,
+      "loss": 0.6029,
+      "step": 7802
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 3.424379757753504e-08,
+      "loss": 0.723,
+      "step": 7803
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 3.371094812782139e-08,
+      "loss": 0.5748,
+      "step": 7804
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 3.318227414580155e-08,
+      "loss": 0.4958,
+      "step": 7805
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 3.2657775719896476e-08,
+      "loss": 0.7901,
+      "step": 7806
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 3.213745293782211e-08,
+      "loss": 0.5995,
+      "step": 7807
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 3.162130588660051e-08,
+      "loss": 0.6165,
+      "step": 7808
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 3.110933465255705e-08,
+      "loss": 0.6381,
+      "step": 7809
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 3.0601539321312155e-08,
+      "loss": 0.5389,
+      "step": 7810
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 3.0097919977795094e-08,
+      "loss": 0.5678,
+      "step": 7811
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.9598476706232948e-08,
+      "loss": 0.6954,
+      "step": 7812
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.910320959015056e-08,
+      "loss": 0.6567,
+      "step": 7813
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.8612118712381674e-08,
+      "loss": 0.5574,
+      "step": 7814
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.8125204155057816e-08,
+      "loss": 0.4842,
+      "step": 7815
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.7642465999613842e-08,
+      "loss": 0.6884,
+      "step": 7816
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.7163904326785172e-08,
+      "loss": 0.5883,
+      "step": 7817
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.6689519216605007e-08,
+      "loss": 0.5948,
+      "step": 7818
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.6219310748415437e-08,
+      "loss": 0.5069,
+      "step": 7819
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.575327900085356e-08,
+      "loss": 0.358,
+      "step": 7820
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.52914240518598e-08,
+      "loss": 0.7343,
+      "step": 7821
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.483374597868071e-08,
+      "loss": 0.5033,
+      "step": 7822
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.4380244857857835e-08,
+      "loss": 0.5795,
+      "step": 7823
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.393092076523329e-08,
+      "loss": 0.6584,
+      "step": 7824
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.348577377596084e-08,
+      "loss": 0.5318,
+      "step": 7825
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.304480396448372e-08,
+      "loss": 0.6112,
+      "step": 7826
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.2608011404554043e-08,
+      "loss": 0.5078,
+      "step": 7827
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.2175396169221707e-08,
+      "loss": 0.2934,
+      "step": 7828
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.174695833083995e-08,
+      "loss": 0.7963,
+      "step": 7829
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.132269796105979e-08,
+      "loss": 0.7109,
+      "step": 7830
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.090261513084113e-08,
+      "loss": 0.6063,
+      "step": 7831
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.048670991043611e-08,
+      "loss": 0.7944,
+      "step": 7832
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.0074982369402973e-08,
+      "loss": 0.7682,
+      "step": 7833
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.9667432576603308e-08,
+      "loss": 0.5081,
+      "step": 7834
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.9264060600196478e-08,
+      "loss": 0.6073,
+      "step": 7835
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.886486650764241e-08,
+      "loss": 0.646,
+      "step": 7836
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.846985036570714e-08,
+      "loss": 0.6311,
+      "step": 7837
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 1.807901224045172e-08,
+      "loss": 0.796,
+      "step": 7838
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.7692352197240526e-08,
+      "loss": 0.4804,
+      "step": 7839
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.7309870300746823e-08,
+      "loss": 0.5492,
+      "step": 7840
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.693156661493056e-08,
+      "loss": 0.5507,
+      "step": 7841
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.6557441203066126e-08,
+      "loss": 0.6059,
+      "step": 7842
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.6187494127722914e-08,
+      "loss": 0.5591,
+      "step": 7843
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.5821725450770873e-08,
+      "loss": 0.6597,
+      "step": 7844
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.54601352333833e-08,
+      "loss": 0.595,
+      "step": 7845
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.5102723536034035e-08,
+      "loss": 0.6249,
+      "step": 7846
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.4749490418497492e-08,
+      "loss": 0.4928,
+      "step": 7847
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.440043593985141e-08,
+      "loss": 0.4783,
+      "step": 7848
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.405556015847409e-08,
+      "loss": 0.3457,
+      "step": 7849
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.3714863132041622e-08,
+      "loss": 0.662,
+      "step": 7850
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 1.3378344917533425e-08,
+      "loss": 0.6465,
+      "step": 7851
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 1.3046005571232256e-08,
+      "loss": 0.7662,
+      "step": 7852
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 1.2717845148718654e-08,
+      "loss": 0.7738,
+      "step": 7853
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 1.2393863704876496e-08,
+      "loss": 0.7423,
+      "step": 7854
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.207406129388744e-08,
+      "loss": 0.7323,
+      "step": 7855
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.1758437969239255e-08,
+      "loss": 0.8181,
+      "step": 7856
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.1446993783717497e-08,
+      "loss": 0.5229,
+      "step": 7857
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.1139728789408277e-08,
+      "loss": 0.6617,
+      "step": 7858
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 1.0836643037701044e-08,
+      "loss": 0.3213,
+      "step": 7859
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 1.053773657928303e-08,
+      "loss": 0.4675,
+      "step": 7860
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 1.024300946414758e-08,
+      "loss": 0.7083,
+      "step": 7861
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 9.952461741585817e-09,
+      "loss": 0.6388,
+      "step": 7862
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 9.666093460189429e-09,
+      "loss": 0.6292,
+      "step": 7863
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 9.383904667850662e-09,
+      "loss": 0.538,
+      "step": 7864
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 9.105895411762322e-09,
+      "loss": 0.7529,
+      "step": 7865
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 8.832065738426098e-09,
+      "loss": 0.5661,
+      "step": 7866
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 8.56241569363314e-09,
+      "loss": 0.593,
+      "step": 7867
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 8.296945322483484e-09,
+      "loss": 0.5277,
+      "step": 7868
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 8.035654669374947e-09,
+      "loss": 0.7574,
+      "step": 7869
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 7.77854377800591e-09,
+      "loss": 0.656,
+      "step": 7870
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 7.525612691375305e-09,
+      "loss": 0.6191,
+      "step": 7871
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 7.276861451788186e-09,
+      "loss": 0.5903,
+      "step": 7872
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 7.032290100844607e-09,
+      "loss": 0.6105,
+      "step": 7873
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 6.791898679450736e-09,
+      "loss": 0.7366,
+      "step": 7874
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.5556872278049745e-09,
+      "loss": 0.5123,
+      "step": 7875
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.323655785414606e-09,
+      "loss": 0.645,
+      "step": 7876
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.095804391087478e-09,
+      "loss": 0.6437,
+      "step": 7877
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 5.872133082929221e-09,
+      "loss": 0.5858,
+      "step": 7878
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.652641898346023e-09,
+      "loss": 0.5322,
+      "step": 7879
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.437330874050184e-09,
+      "loss": 0.5975,
+      "step": 7880
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.226200046049012e-09,
+      "loss": 0.6127,
+      "step": 7881
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 5.019249449653152e-09,
+      "loss": 0.6372,
+      "step": 7882
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 4.816479119471029e-09,
+      "loss": 0.8623,
+      "step": 7883
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 4.617889089419958e-09,
+      "loss": 0.5091,
+      "step": 7884
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 4.423479392709484e-09,
+      "loss": 0.5559,
+      "step": 7885
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 4.233250061855265e-09,
+      "loss": 0.5883,
+      "step": 7886
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.047201128670741e-09,
+      "loss": 0.6127,
+      "step": 7887
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 3.865332624272688e-09,
+      "loss": 0.6824,
+      "step": 7888
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 3.6876445790756662e-09,
+      "loss": 0.6771,
+      "step": 7889
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 3.5141370228003455e-09,
+      "loss": 0.5075,
+      "step": 7890
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.3448099844596294e-09,
+      "loss": 0.6102,
+      "step": 7891
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.179663492378082e-09,
+      "loss": 0.4838,
+      "step": 7892
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.018697574172502e-09,
+      "loss": 0.6633,
+      "step": 7893
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.8619122567657972e-09,
+      "loss": 0.5345,
+      "step": 7894
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.709307566375885e-09,
+      "loss": 0.5355,
+      "step": 7895
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.560883528526792e-09,
+      "loss": 0.4993,
+      "step": 7896
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.4166401680403295e-09,
+      "loss": 0.5517,
+      "step": 7897
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.2765775090444197e-09,
+      "loss": 0.4239,
+      "step": 7898
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.140695574959217e-09,
+      "loss": 0.5355,
+      "step": 7899
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.008994388513763e-09,
+      "loss": 0.6714,
+      "step": 7900
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.8814739717293305e-09,
+      "loss": 0.5035,
+      "step": 7901
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.7581343459388556e-09,
+      "loss": 0.6315,
+      "step": 7902
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 1.6389755317675059e-09,
+      "loss": 0.6218,
+      "step": 7903
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 1.5239975491437853e-09,
+      "loss": 0.5378,
+      "step": 7904
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 1.4132004172939806e-09,
+      "loss": 0.4692,
+      "step": 7905
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 1.306584154756041e-09,
+      "loss": 0.66,
+      "step": 7906
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.2041487793545968e-09,
+      "loss": 0.5146,
+      "step": 7907
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.10589430822039e-09,
+      "loss": 0.7021,
+      "step": 7908
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 1.0118207577930472e-09,
+      "loss": 0.4051,
+      "step": 7909
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 9.219281437988781e-10,
+      "loss": 0.7005,
+      "step": 7910
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 8.362164812758533e-10,
+      "loss": 0.4933,
+      "step": 7911
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 7.546857845569522e-10,
+      "loss": 0.4278,
+      "step": 7912
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 6.773360672757135e-10,
+      "loss": 0.5571,
+      "step": 7913
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 6.04167342374562e-10,
+      "loss": 0.6876,
+      "step": 7914
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 5.3517962208538e-10,
+      "loss": 0.6141,
+      "step": 7915
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 4.703729179461603e-10,
+      "loss": 0.735,
+      "step": 7916
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 4.0974724079823055e-10,
+      "loss": 0.4627,
+      "step": 7917
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 3.533026007779272e-10,
+      "loss": 0.6149,
+      "step": 7918
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 3.01039007327697e-10,
+      "loss": 0.5411,
+      "step": 7919
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.529564691877706e-10,
+      "loss": 0.4426,
+      "step": 7920
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7920,
+  "num_train_epochs": 20,
+  "save_steps": 1000000000,
+  "total_flos": 1.0707352932057088e+19,
+  "trial_name": null,
+  "trial_params": null
+}