diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9025 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 1.0164,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6417580093987654e-06,
+      "loss": 0.8405,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.772049881597968e-06,
+      "loss": 0.8834,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.283516018797531e-06,
+      "loss": 0.7104,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.455900236804067e-06,
+      "loss": 0.7056,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.413807890996732e-06,
+      "loss": 0.6901,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0223707272628328e-05,
+      "loss": 0.6461,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0925274028196296e-05,
+      "loss": 0.6526,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1544099763195936e-05,
+      "loss": 0.6329,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2097658246202834e-05,
+      "loss": 0.6054,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2598412805139713e-05,
+      "loss": 0.6179,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3055565900395498e-05,
+      "loss": 0.6471,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3476105981837633e-05,
+      "loss": 0.5956,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3865465282027091e-05,
+      "loss": 0.5981,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4227950118402035e-05,
+      "loss": 0.5962,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4567032037595061e-05,
+      "loss": 0.5875,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4885550540243259e-05,
+      "loss": 0.6061,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5185857772594698e-05,
+      "loss": 0.6096,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5469924045428557e-05,
+      "loss": 0.574,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5739416255601598e-05,
+      "loss": 0.552,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.599575715422629e-05,
+      "loss": 0.5557,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.624017081453848e-05,
+      "loss": 0.6003,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.647371798448217e-05,
+      "loss": 0.5835,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6697323909794263e-05,
+      "loss": 0.6021,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6911800473608135e-05,
+      "loss": 0.5963,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.71178639912364e-05,
+      "loss": 0.5938,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.73161496447939e-05,
+      "loss": 0.517,
+      "step": 27
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7507223291425857e-05,
+      "loss": 0.5493,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.769159119851282e-05,
+      "loss": 0.5902,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.78697081278008e-05,
+      "loss": 0.5252,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8041984093485214e-05,
+      "loss": 0.5722,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.820879004699383e-05,
+      "loss": 0.5961,
+      "step": 32
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.837046268673768e-05,
+      "loss": 0.5576,
+      "step": 33
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8527308549642024e-05,
+      "loss": 0.562,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8679607509432393e-05,
+      "loss": 0.5313,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8827615781993465e-05,
+      "loss": 0.5574,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.897156851886858e-05,
+      "loss": 0.5605,
+      "step": 37
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9111682054827324e-05,
+      "loss": 0.5571,
+      "step": 38
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9248155863435598e-05,
+      "loss": 0.5393,
+      "step": 39
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9381174265000365e-05,
+      "loss": 0.5389,
+      "step": 40
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.951090792358832e-05,
+      "loss": 0.5289,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9637515163625058e-05,
+      "loss": 0.5324,
+      "step": 42
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9761143131554592e-05,
+      "loss": 0.5228,
+      "step": 43
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9881928823937242e-05,
+      "loss": 0.5573,
+      "step": 44
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.5393,
+      "step": 45
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.556,
+      "step": 46
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986254295532648e-05,
+      "loss": 0.5285,
+      "step": 47
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9972508591065294e-05,
+      "loss": 0.5285,
+      "step": 48
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.995876288659794e-05,
+      "loss": 0.5307,
+      "step": 49
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9945017182130586e-05,
+      "loss": 0.5138,
+      "step": 50
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9931271477663232e-05,
+      "loss": 0.5678,
+      "step": 51
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9917525773195878e-05,
+      "loss": 0.5319,
+      "step": 52
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9903780068728524e-05,
+      "loss": 0.4967,
+      "step": 53
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.989003436426117e-05,
+      "loss": 0.528,
+      "step": 54
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9876288659793816e-05,
+      "loss": 0.5264,
+      "step": 55
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9862542955326462e-05,
+      "loss": 0.5381,
+      "step": 56
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9848797250859108e-05,
+      "loss": 0.5466,
+      "step": 57
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9835051546391754e-05,
+      "loss": 0.5334,
+      "step": 58
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.98213058419244e-05,
+      "loss": 0.5395,
+      "step": 59
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9807560137457046e-05,
+      "loss": 0.5405,
+      "step": 60
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9793814432989692e-05,
+      "loss": 0.5182,
+      "step": 61
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9780068728522338e-05,
+      "loss": 0.5166,
+      "step": 62
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9766323024054984e-05,
+      "loss": 0.503,
+      "step": 63
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.975257731958763e-05,
+      "loss": 0.5571,
+      "step": 64
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9738831615120276e-05,
+      "loss": 0.5386,
+      "step": 65
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9725085910652922e-05,
+      "loss": 0.5634,
+      "step": 66
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9711340206185568e-05,
+      "loss": 0.5213,
+      "step": 67
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9697594501718217e-05,
+      "loss": 0.4895,
+      "step": 68
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.968384879725086e-05,
+      "loss": 0.5555,
+      "step": 69
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9670103092783506e-05,
+      "loss": 0.5213,
+      "step": 70
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9656357388316152e-05,
+      "loss": 0.555,
+      "step": 71
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9642611683848798e-05,
+      "loss": 0.5115,
+      "step": 72
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9628865979381444e-05,
+      "loss": 0.512,
+      "step": 73
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.961512027491409e-05,
+      "loss": 0.5093,
+      "step": 74
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9601374570446736e-05,
+      "loss": 0.5029,
+      "step": 75
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9587628865979382e-05,
+      "loss": 0.4886,
+      "step": 76
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9573883161512028e-05,
+      "loss": 0.5042,
+      "step": 77
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9560137457044674e-05,
+      "loss": 0.4847,
+      "step": 78
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.954639175257732e-05,
+      "loss": 0.5512,
+      "step": 79
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9532646048109966e-05,
+      "loss": 0.5082,
+      "step": 80
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9518900343642612e-05,
+      "loss": 0.5274,
+      "step": 81
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9505154639175258e-05,
+      "loss": 0.5256,
+      "step": 82
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9491408934707904e-05,
+      "loss": 0.4952,
+      "step": 83
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.947766323024055e-05,
+      "loss": 0.5067,
+      "step": 84
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9463917525773196e-05,
+      "loss": 0.515,
+      "step": 85
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9450171821305842e-05,
+      "loss": 0.489,
+      "step": 86
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.943642611683849e-05,
+      "loss": 0.5025,
+      "step": 87
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9422680412371134e-05,
+      "loss": 0.4947,
+      "step": 88
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.940893470790378e-05,
+      "loss": 0.5165,
+      "step": 89
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9395189003436426e-05,
+      "loss": 0.473,
+      "step": 90
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9381443298969072e-05,
+      "loss": 0.4868,
+      "step": 91
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.936769759450172e-05,
+      "loss": 0.5153,
+      "step": 92
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9353951890034364e-05,
+      "loss": 0.4833,
+      "step": 93
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.934020618556701e-05,
+      "loss": 0.5242,
+      "step": 94
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9326460481099656e-05,
+      "loss": 0.5138,
+      "step": 95
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9312714776632306e-05,
+      "loss": 0.4941,
+      "step": 96
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9298969072164952e-05,
+      "loss": 0.5316,
+      "step": 97
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9285223367697598e-05,
+      "loss": 0.4951,
+      "step": 98
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9271477663230244e-05,
+      "loss": 0.5495,
+      "step": 99
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.925773195876289e-05,
+      "loss": 0.4812,
+      "step": 100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9243986254295536e-05,
+      "loss": 0.4714,
+      "step": 101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9230240549828182e-05,
+      "loss": 0.5283,
+      "step": 102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9216494845360828e-05,
+      "loss": 0.4746,
+      "step": 103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9202749140893474e-05,
+      "loss": 0.4857,
+      "step": 104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.918900343642612e-05,
+      "loss": 0.5127,
+      "step": 105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9175257731958766e-05,
+      "loss": 0.506,
+      "step": 106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9161512027491412e-05,
+      "loss": 0.4739,
+      "step": 107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9147766323024058e-05,
+      "loss": 0.4897,
+      "step": 108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9134020618556704e-05,
+      "loss": 0.4829,
+      "step": 109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.912027491408935e-05,
+      "loss": 0.5053,
+      "step": 110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9106529209621996e-05,
+      "loss": 0.4803,
+      "step": 111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9092783505154642e-05,
+      "loss": 0.4718,
+      "step": 112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9079037800687288e-05,
+      "loss": 0.495,
+      "step": 113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9065292096219934e-05,
+      "loss": 0.4947,
+      "step": 114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9051546391752577e-05,
+      "loss": 0.4839,
+      "step": 115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9037800687285223e-05,
+      "loss": 0.4998,
+      "step": 116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.902405498281787e-05,
+      "loss": 0.507,
+      "step": 117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9010309278350515e-05,
+      "loss": 0.4538,
+      "step": 118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.899656357388316e-05,
+      "loss": 0.4727,
+      "step": 119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8982817869415807e-05,
+      "loss": 0.4843,
+      "step": 120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8969072164948453e-05,
+      "loss": 0.5025,
+      "step": 121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.89553264604811e-05,
+      "loss": 0.4918,
+      "step": 122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8941580756013745e-05,
+      "loss": 0.4756,
+      "step": 123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8927835051546394e-05,
+      "loss": 0.4967,
+      "step": 124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.891408934707904e-05,
+      "loss": 0.4992,
+      "step": 125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8900343642611686e-05,
+      "loss": 0.4812,
+      "step": 126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8886597938144332e-05,
+      "loss": 0.4898,
+      "step": 127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.887285223367698e-05,
+      "loss": 0.4952,
+      "step": 128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8859106529209624e-05,
+      "loss": 0.5155,
+      "step": 129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.884536082474227e-05,
+      "loss": 0.4854,
+      "step": 130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8831615120274916e-05,
+      "loss": 0.4719,
+      "step": 131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8817869415807562e-05,
+      "loss": 0.4739,
+      "step": 132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.880412371134021e-05,
+      "loss": 0.5014,
+      "step": 133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8790378006872854e-05,
+      "loss": 0.4532,
+      "step": 134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.87766323024055e-05,
+      "loss": 0.5012,
+      "step": 135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8762886597938147e-05,
+      "loss": 0.5185,
+      "step": 136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8749140893470793e-05,
+      "loss": 0.4896,
+      "step": 137
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.873539518900344e-05,
+      "loss": 0.4623,
+      "step": 138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8721649484536085e-05,
+      "loss": 0.485,
+      "step": 139
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.870790378006873e-05,
+      "loss": 0.4778,
+      "step": 140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8694158075601377e-05,
+      "loss": 0.484,
+      "step": 141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8680412371134023e-05,
+      "loss": 0.4778,
+      "step": 142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.4735,
+      "step": 143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8652920962199315e-05,
+      "loss": 0.4585,
+      "step": 144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863917525773196e-05,
+      "loss": 0.4654,
+      "step": 145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8625429553264607e-05,
+      "loss": 0.4889,
+      "step": 146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8611683848797253e-05,
+      "loss": 0.4872,
+      "step": 147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.85979381443299e-05,
+      "loss": 0.5021,
+      "step": 148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8584192439862545e-05,
+      "loss": 0.5002,
+      "step": 149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.857044673539519e-05,
+      "loss": 0.4914,
+      "step": 150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8556701030927837e-05,
+      "loss": 0.4969,
+      "step": 151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8542955326460483e-05,
+      "loss": 0.4913,
+      "step": 152
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.852920962199313e-05,
+      "loss": 0.4969,
+      "step": 153
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8515463917525775e-05,
+      "loss": 0.42,
+      "step": 154
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.850171821305842e-05,
+      "loss": 0.4916,
+      "step": 155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8487972508591067e-05,
+      "loss": 0.4561,
+      "step": 156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8474226804123713e-05,
+      "loss": 0.4757,
+      "step": 157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.846048109965636e-05,
+      "loss": 0.4832,
+      "step": 158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8446735395189005e-05,
+      "loss": 0.4674,
+      "step": 159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.843298969072165e-05,
+      "loss": 0.5089,
+      "step": 160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8419243986254297e-05,
+      "loss": 0.5119,
+      "step": 161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8405498281786943e-05,
+      "loss": 0.4584,
+      "step": 162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.839175257731959e-05,
+      "loss": 0.5134,
+      "step": 163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8378006872852235e-05,
+      "loss": 0.467,
+      "step": 164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.836426116838488e-05,
+      "loss": 0.4638,
+      "step": 165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8350515463917527e-05,
+      "loss": 0.4812,
+      "step": 166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8336769759450173e-05,
+      "loss": 0.4681,
+      "step": 167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.832302405498282e-05,
+      "loss": 0.4621,
+      "step": 168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8309278350515465e-05,
+      "loss": 0.4476,
+      "step": 169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.829553264604811e-05,
+      "loss": 0.4661,
+      "step": 170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8281786941580757e-05,
+      "loss": 0.4921,
+      "step": 171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8268041237113403e-05,
+      "loss": 0.4609,
+      "step": 172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.825429553264605e-05,
+      "loss": 0.4791,
+      "step": 173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8240549828178695e-05,
+      "loss": 0.4732,
+      "step": 174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.822680412371134e-05,
+      "loss": 0.4692,
+      "step": 175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8213058419243987e-05,
+      "loss": 0.5002,
+      "step": 176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8199312714776633e-05,
+      "loss": 0.4618,
+      "step": 177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.818556701030928e-05,
+      "loss": 0.4686,
+      "step": 178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8171821305841925e-05,
+      "loss": 0.4904,
+      "step": 179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.815807560137457e-05,
+      "loss": 0.4488,
+      "step": 180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8144329896907217e-05,
+      "loss": 0.4625,
+      "step": 181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8130584192439863e-05,
+      "loss": 0.4768,
+      "step": 182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.811683848797251e-05,
+      "loss": 0.4481,
+      "step": 183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8103092783505155e-05,
+      "loss": 0.4643,
+      "step": 184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.80893470790378e-05,
+      "loss": 0.4603,
+      "step": 185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8075601374570447e-05,
+      "loss": 0.4588,
+      "step": 186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8061855670103093e-05,
+      "loss": 0.4395,
+      "step": 187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.804810996563574e-05,
+      "loss": 0.4536,
+      "step": 188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8034364261168385e-05,
+      "loss": 0.4588,
+      "step": 189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.802061855670103e-05,
+      "loss": 0.4679,
+      "step": 190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8006872852233677e-05,
+      "loss": 0.4513,
+      "step": 191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7993127147766323e-05,
+      "loss": 0.4633,
+      "step": 192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.797938144329897e-05,
+      "loss": 0.4559,
+      "step": 193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7965635738831615e-05,
+      "loss": 0.4693,
+      "step": 194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7951890034364265e-05,
+      "loss": 0.4412,
+      "step": 195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.793814432989691e-05,
+      "loss": 0.4537,
+      "step": 196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7924398625429557e-05,
+      "loss": 0.4708,
+      "step": 197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7910652920962203e-05,
+      "loss": 0.4865,
+      "step": 198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.789690721649485e-05,
+      "loss": 0.428,
+      "step": 199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7883161512027495e-05,
+      "loss": 0.4864,
+      "step": 200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.786941580756014e-05,
+      "loss": 0.4868,
+      "step": 201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7855670103092787e-05,
+      "loss": 0.4503,
+      "step": 202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7841924398625433e-05,
+      "loss": 0.4416,
+      "step": 203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.782817869415808e-05,
+      "loss": 0.4555,
+      "step": 204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7814432989690725e-05,
+      "loss": 0.4586,
+      "step": 205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7800687285223368e-05,
+      "loss": 0.4505,
+      "step": 206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7786941580756014e-05,
+      "loss": 0.4787,
+      "step": 207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.777319587628866e-05,
+      "loss": 0.4513,
+      "step": 208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7759450171821306e-05,
+      "loss": 0.4638,
+      "step": 209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7745704467353952e-05,
+      "loss": 0.4793,
+      "step": 210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7731958762886598e-05,
+      "loss": 0.4608,
+      "step": 211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7718213058419244e-05,
+      "loss": 0.4517,
+      "step": 212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.770446735395189e-05,
+      "loss": 0.4559,
+      "step": 213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7690721649484536e-05,
+      "loss": 0.4564,
+      "step": 214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7676975945017182e-05,
+      "loss": 0.4632,
+      "step": 215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7663230240549828e-05,
+      "loss": 0.4607,
+      "step": 216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7649484536082474e-05,
+      "loss": 0.4615,
+      "step": 217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.763573883161512e-05,
+      "loss": 0.4623,
+      "step": 218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7621993127147766e-05,
+      "loss": 0.4979,
+      "step": 219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7608247422680412e-05,
+      "loss": 0.4605,
+      "step": 220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7594501718213058e-05,
+      "loss": 0.4901,
+      "step": 221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7580756013745704e-05,
+      "loss": 0.4555,
+      "step": 222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7567010309278353e-05,
+      "loss": 0.4546,
+      "step": 223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7553264604811e-05,
+      "loss": 0.4609,
+      "step": 224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7539518900343645e-05,
+      "loss": 0.4732,
+      "step": 225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.752577319587629e-05,
+      "loss": 0.4547,
+      "step": 226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7512027491408937e-05,
+      "loss": 0.4557,
+      "step": 227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7498281786941583e-05,
+      "loss": 0.4703,
+      "step": 228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.748453608247423e-05,
+      "loss": 0.4427,
+      "step": 229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7470790378006875e-05,
+      "loss": 0.4714,
+      "step": 230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.745704467353952e-05,
+      "loss": 0.4677,
+      "step": 231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7443298969072168e-05,
+      "loss": 0.433,
+      "step": 232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7429553264604814e-05,
+      "loss": 0.4391,
+      "step": 233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.741580756013746e-05,
+      "loss": 0.4438,
+      "step": 234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7402061855670106e-05,
+      "loss": 0.4475,
+      "step": 235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.738831615120275e-05,
+      "loss": 0.4369,
+      "step": 236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7374570446735398e-05,
+      "loss": 0.4324,
+      "step": 237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7360824742268044e-05,
+      "loss": 0.4324,
+      "step": 238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.734707903780069e-05,
+      "loss": 0.4351,
+      "step": 239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.4425,
+      "step": 240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.731958762886598e-05,
+      "loss": 0.4744,
+      "step": 241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7305841924398628e-05,
+      "loss": 0.4511,
+      "step": 242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7292096219931274e-05,
+      "loss": 0.4399,
+      "step": 243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.727835051546392e-05,
+      "loss": 0.4388,
+      "step": 244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7264604810996566e-05,
+      "loss": 0.4465,
+      "step": 245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7250859106529212e-05,
+      "loss": 0.4392,
+      "step": 246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7237113402061858e-05,
+      "loss": 0.4651,
+      "step": 247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7223367697594504e-05,
+      "loss": 0.4643,
+      "step": 248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.720962199312715e-05,
+      "loss": 0.4554,
+      "step": 249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7195876288659796e-05,
+      "loss": 0.4812,
+      "step": 250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7182130584192442e-05,
+      "loss": 0.4211,
+      "step": 251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7168384879725088e-05,
+      "loss": 0.4429,
+      "step": 252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7154639175257734e-05,
+      "loss": 0.4539,
+      "step": 253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.714089347079038e-05,
+      "loss": 0.4614,
+      "step": 254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7127147766323026e-05,
+      "loss": 0.4625,
+      "step": 255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7113402061855672e-05,
+      "loss": 0.4323,
+      "step": 256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7099656357388318e-05,
+      "loss": 0.4402,
+      "step": 257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7085910652920964e-05,
+      "loss": 0.4408,
+      "step": 258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.707216494845361e-05,
+      "loss": 0.4573,
+      "step": 259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7058419243986256e-05,
+      "loss": 0.4715,
+      "step": 260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7044673539518902e-05,
+      "loss": 0.4279,
+      "step": 261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7030927835051548e-05,
+      "loss": 0.4642,
+      "step": 262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7017182130584194e-05,
+      "loss": 0.4183,
+      "step": 263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.700343642611684e-05,
+      "loss": 0.451,
+      "step": 264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6989690721649486e-05,
+      "loss": 0.4478,
+      "step": 265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6975945017182132e-05,
+      "loss": 0.4575,
+      "step": 266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6962199312714778e-05,
+      "loss": 0.4303,
+      "step": 267
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6948453608247424e-05,
+      "loss": 0.4277,
+      "step": 268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.693470790378007e-05,
+      "loss": 0.4515,
+      "step": 269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6920962199312716e-05,
+      "loss": 0.4285,
+      "step": 270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6907216494845362e-05,
+      "loss": 0.4387,
+      "step": 271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6893470790378008e-05,
+      "loss": 0.4379,
+      "step": 272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6879725085910654e-05,
+      "loss": 0.4444,
+      "step": 273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.68659793814433e-05,
+      "loss": 0.4371,
+      "step": 274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6852233676975946e-05,
+      "loss": 0.4538,
+      "step": 275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6838487972508592e-05,
+      "loss": 0.4312,
+      "step": 276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6824742268041238e-05,
+      "loss": 0.429,
+      "step": 277
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6810996563573884e-05,
+      "loss": 0.4668,
+      "step": 278
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.679725085910653e-05,
+      "loss": 0.4505,
+      "step": 279
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6783505154639176e-05,
+      "loss": 0.4603,
+      "step": 280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6769759450171822e-05,
+      "loss": 0.4323,
+      "step": 281
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.675601374570447e-05,
+      "loss": 0.4835,
+      "step": 282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6742268041237114e-05,
+      "loss": 0.4414,
+      "step": 283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.672852233676976e-05,
+      "loss": 0.4329,
+      "step": 284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6714776632302406e-05,
+      "loss": 0.4379,
+      "step": 285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6701030927835052e-05,
+      "loss": 0.4325,
+      "step": 286
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.66872852233677e-05,
+      "loss": 0.4483,
+      "step": 287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6673539518900344e-05,
+      "loss": 0.444,
+      "step": 288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665979381443299e-05,
+      "loss": 0.4583,
+      "step": 289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6646048109965636e-05,
+      "loss": 0.4349,
+      "step": 290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6632302405498283e-05,
+      "loss": 0.4367,
+      "step": 291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661855670103093e-05,
+      "loss": 0.4395,
+      "step": 292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6604810996563575e-05,
+      "loss": 0.4591,
+      "step": 293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6591065292096224e-05,
+      "loss": 0.4474,
+      "step": 294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.657731958762887e-05,
+      "loss": 0.4349,
+      "step": 295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6563573883161516e-05,
+      "loss": 0.4516,
+      "step": 296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654982817869416e-05,
+      "loss": 0.4432,
+      "step": 297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6536082474226805e-05,
+      "loss": 0.4513,
+      "step": 298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.652233676975945e-05,
+      "loss": 0.4543,
+      "step": 299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6508591065292097e-05,
+      "loss": 0.4129,
+      "step": 300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6494845360824743e-05,
+      "loss": 0.4442,
+      "step": 301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.648109965635739e-05,
+      "loss": 0.4205,
+      "step": 302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6467353951890035e-05,
+      "loss": 0.4483,
+      "step": 303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.645360824742268e-05,
+      "loss": 0.4214,
+      "step": 304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6439862542955327e-05,
+      "loss": 0.4546,
+      "step": 305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6426116838487973e-05,
+      "loss": 0.452,
+      "step": 306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.641237113402062e-05,
+      "loss": 0.4179,
+      "step": 307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6398625429553265e-05,
+      "loss": 0.4287,
+      "step": 308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.638487972508591e-05,
+      "loss": 0.4476,
+      "step": 309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6371134020618557e-05,
+      "loss": 0.4113,
+      "step": 310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6357388316151203e-05,
+      "loss": 0.4395,
+      "step": 311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.634364261168385e-05,
+      "loss": 0.4253,
+      "step": 312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6329896907216495e-05,
+      "loss": 0.467,
+      "step": 313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.631615120274914e-05,
+      "loss": 0.4587,
+      "step": 314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6302405498281787e-05,
+      "loss": 0.4317,
+      "step": 315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6288659793814433e-05,
+      "loss": 0.4656,
+      "step": 316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.627491408934708e-05,
+      "loss": 0.408,
+      "step": 317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6261168384879725e-05,
+      "loss": 0.4052,
+      "step": 318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.624742268041237e-05,
+      "loss": 0.4356,
+      "step": 319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6233676975945017e-05,
+      "loss": 0.4305,
+      "step": 320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6219931271477663e-05,
+      "loss": 0.4283,
+      "step": 321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6206185567010312e-05,
+      "loss": 0.4488,
+      "step": 322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.619243986254296e-05,
+      "loss": 0.4394,
+      "step": 323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6178694158075604e-05,
+      "loss": 0.4193,
+      "step": 324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.616494845360825e-05,
+      "loss": 0.3997,
+      "step": 325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6151202749140896e-05,
+      "loss": 0.4195,
+      "step": 326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6137457044673542e-05,
+      "loss": 0.4351,
+      "step": 327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.612371134020619e-05,
+      "loss": 0.4355,
+      "step": 328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6109965635738835e-05,
+      "loss": 0.426,
+      "step": 329
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.609621993127148e-05,
+      "loss": 0.4293,
+      "step": 330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6082474226804127e-05,
+      "loss": 0.4364,
+      "step": 331
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6068728522336773e-05,
+      "loss": 0.4377,
+      "step": 332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.605498281786942e-05,
+      "loss": 0.4316,
+      "step": 333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6041237113402065e-05,
+      "loss": 0.4261,
+      "step": 334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.602749140893471e-05,
+      "loss": 0.4457,
+      "step": 335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6013745704467357e-05,
+      "loss": 0.4529,
+      "step": 336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.4148,
+      "step": 337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.598625429553265e-05,
+      "loss": 0.4663,
+      "step": 338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5972508591065295e-05,
+      "loss": 0.4286,
+      "step": 339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.595876288659794e-05,
+      "loss": 0.4512,
+      "step": 340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5945017182130587e-05,
+      "loss": 0.4448,
+      "step": 341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.593127147766323e-05,
+      "loss": 0.412,
+      "step": 342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5917525773195875e-05,
+      "loss": 0.4242,
+      "step": 343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.590378006872852e-05,
+      "loss": 0.4237,
+      "step": 344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5890034364261167e-05,
+      "loss": 0.4355,
+      "step": 345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5876288659793813e-05,
+      "loss": 0.4479,
+      "step": 346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.586254295532646e-05,
+      "loss": 0.4343,
+      "step": 347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5848797250859105e-05,
+      "loss": 0.4123,
+      "step": 348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.583505154639175e-05,
+      "loss": 0.4041,
+      "step": 349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.58213058419244e-05,
+      "loss": 0.4318,
+      "step": 350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5807560137457047e-05,
+      "loss": 0.4375,
+      "step": 351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5793814432989693e-05,
+      "loss": 0.4164,
+      "step": 352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.578006872852234e-05,
+      "loss": 0.4449,
+      "step": 353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5766323024054985e-05,
+      "loss": 0.4349,
+      "step": 354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.575257731958763e-05,
+      "loss": 0.446,
+      "step": 355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5738831615120277e-05,
+      "loss": 0.4079,
+      "step": 356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5725085910652923e-05,
+      "loss": 0.4367,
+      "step": 357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.571134020618557e-05,
+      "loss": 0.436,
+      "step": 358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5697594501718215e-05,
+      "loss": 0.4204,
+      "step": 359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.568384879725086e-05,
+      "loss": 0.4185,
+      "step": 360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5670103092783507e-05,
+      "loss": 0.45,
+      "step": 361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5656357388316153e-05,
+      "loss": 0.4294,
+      "step": 362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.56426116838488e-05,
+      "loss": 0.4257,
+      "step": 363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5628865979381445e-05,
+      "loss": 0.4016,
+      "step": 364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.561512027491409e-05,
+      "loss": 0.4348,
+      "step": 365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5601374570446737e-05,
+      "loss": 0.4355,
+      "step": 366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5587628865979383e-05,
+      "loss": 0.4319,
+      "step": 367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.557388316151203e-05,
+      "loss": 0.3872,
+      "step": 368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5560137457044675e-05,
+      "loss": 0.4202,
+      "step": 369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.554639175257732e-05,
+      "loss": 0.4082,
+      "step": 370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5532646048109967e-05,
+      "loss": 0.4215,
+      "step": 371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5518900343642613e-05,
+      "loss": 0.4247,
+      "step": 372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.550515463917526e-05,
+      "loss": 0.3985,
+      "step": 373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5491408934707905e-05,
+      "loss": 0.4237,
+      "step": 374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.547766323024055e-05,
+      "loss": 0.4339,
+      "step": 375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5463917525773197e-05,
+      "loss": 0.4221,
+      "step": 376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5450171821305843e-05,
+      "loss": 0.4254,
+      "step": 377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.543642611683849e-05,
+      "loss": 0.4437,
+      "step": 378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5422680412371135e-05,
+      "loss": 0.4277,
+      "step": 379
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.540893470790378e-05,
+      "loss": 0.4381,
+      "step": 380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5395189003436427e-05,
+      "loss": 0.4127,
+      "step": 381
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5381443298969073e-05,
+      "loss": 0.4255,
+      "step": 382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.536769759450172e-05,
+      "loss": 0.4294,
+      "step": 383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5353951890034365e-05,
+      "loss": 0.4057,
+      "step": 384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.534020618556701e-05,
+      "loss": 0.4257,
+      "step": 385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5326460481099657e-05,
+      "loss": 0.4492,
+      "step": 386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5312714776632303e-05,
+      "loss": 0.4229,
+      "step": 387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.529896907216495e-05,
+      "loss": 0.4403,
+      "step": 388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5285223367697596e-05,
+      "loss": 0.4001,
+      "step": 389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.527147766323024e-05,
+      "loss": 0.4233,
+      "step": 390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5257731958762888e-05,
+      "loss": 0.4212,
+      "step": 391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5243986254295534e-05,
+      "loss": 0.4104,
+      "step": 392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.523024054982818e-05,
+      "loss": 0.4278,
+      "step": 393
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5216494845360826e-05,
+      "loss": 0.4113,
+      "step": 394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5202749140893472e-05,
+      "loss": 0.4282,
+      "step": 395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5189003436426118e-05,
+      "loss": 0.4074,
+      "step": 396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5175257731958764e-05,
+      "loss": 0.3818,
+      "step": 397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.516151202749141e-05,
+      "loss": 0.4134,
+      "step": 398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5147766323024056e-05,
+      "loss": 0.4233,
+      "step": 399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5134020618556702e-05,
+      "loss": 0.4201,
+      "step": 400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5120274914089348e-05,
+      "loss": 0.3935,
+      "step": 401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5106529209621994e-05,
+      "loss": 0.4354,
+      "step": 402
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.509278350515464e-05,
+      "loss": 0.4107,
+      "step": 403
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5079037800687286e-05,
+      "loss": 0.4216,
+      "step": 404
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5065292096219932e-05,
+      "loss": 0.3953,
+      "step": 405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5051546391752578e-05,
+      "loss": 0.4109,
+      "step": 406
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5037800687285224e-05,
+      "loss": 0.4363,
+      "step": 407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.502405498281787e-05,
+      "loss": 0.4187,
+      "step": 408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5010309278350516e-05,
+      "loss": 0.4279,
+      "step": 409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4996563573883164e-05,
+      "loss": 0.4219,
+      "step": 410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.498281786941581e-05,
+      "loss": 0.4172,
+      "step": 411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4969072164948456e-05,
+      "loss": 0.4436,
+      "step": 412
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4955326460481102e-05,
+      "loss": 0.4289,
+      "step": 413
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4941580756013748e-05,
+      "loss": 0.4059,
+      "step": 414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4927835051546394e-05,
+      "loss": 0.4253,
+      "step": 415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491408934707904e-05,
+      "loss": 0.4235,
+      "step": 416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4900343642611686e-05,
+      "loss": 0.4124,
+      "step": 417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4886597938144332e-05,
+      "loss": 0.399,
+      "step": 418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4872852233676978e-05,
+      "loss": 0.4077,
+      "step": 419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4859106529209624e-05,
+      "loss": 0.4088,
+      "step": 420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.484536082474227e-05,
+      "loss": 0.3903,
+      "step": 421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4831615120274916e-05,
+      "loss": 0.4117,
+      "step": 422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4817869415807562e-05,
+      "loss": 0.4214,
+      "step": 423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4804123711340208e-05,
+      "loss": 0.4085,
+      "step": 424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4790378006872854e-05,
+      "loss": 0.4259,
+      "step": 425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.47766323024055e-05,
+      "loss": 0.3896,
+      "step": 426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4762886597938146e-05,
+      "loss": 0.4327,
+      "step": 427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4749140893470792e-05,
+      "loss": 0.3763,
+      "step": 428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4735395189003438e-05,
+      "loss": 0.4159,
+      "step": 429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4721649484536084e-05,
+      "loss": 0.4273,
+      "step": 430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.470790378006873e-05,
+      "loss": 0.4116,
+      "step": 431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4694158075601378e-05,
+      "loss": 0.4141,
+      "step": 432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.468041237113402e-05,
+      "loss": 0.4203,
+      "step": 433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.3811,
+      "step": 434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4652920962199312e-05,
+      "loss": 0.4193,
+      "step": 435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4639175257731958e-05,
+      "loss": 0.4133,
+      "step": 436
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4625429553264604e-05,
+      "loss": 0.4103,
+      "step": 437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4611683848797252e-05,
+      "loss": 0.3851,
+      "step": 438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4597938144329898e-05,
+      "loss": 0.4085,
+      "step": 439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4584192439862544e-05,
+      "loss": 0.4275,
+      "step": 440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.457044673539519e-05,
+      "loss": 0.4405,
+      "step": 441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4556701030927836e-05,
+      "loss": 0.4213,
+      "step": 442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4542955326460482e-05,
+      "loss": 0.4701,
+      "step": 443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4529209621993128e-05,
+      "loss": 0.3969,
+      "step": 444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4515463917525774e-05,
+      "loss": 0.3998,
+      "step": 445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.450171821305842e-05,
+      "loss": 0.3883,
+      "step": 446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4487972508591066e-05,
+      "loss": 0.3722,
+      "step": 447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4474226804123712e-05,
+      "loss": 0.3989,
+      "step": 448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4460481099656358e-05,
+      "loss": 0.4196,
+      "step": 449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4446735395189004e-05,
+      "loss": 0.4139,
+      "step": 450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.443298969072165e-05,
+      "loss": 0.3856,
+      "step": 451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4419243986254296e-05,
+      "loss": 0.3771,
+      "step": 452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4405498281786942e-05,
+      "loss": 0.3936,
+      "step": 453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4391752577319588e-05,
+      "loss": 0.438,
+      "step": 454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4378006872852234e-05,
+      "loss": 0.3955,
+      "step": 455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.436426116838488e-05,
+      "loss": 0.4139,
+      "step": 456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4350515463917526e-05,
+      "loss": 0.3936,
+      "step": 457
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4336769759450172e-05,
+      "loss": 0.4046,
+      "step": 458
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4323024054982818e-05,
+      "loss": 0.3948,
+      "step": 459
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4309278350515466e-05,
+      "loss": 0.4078,
+      "step": 460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4295532646048112e-05,
+      "loss": 0.3929,
+      "step": 461
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4281786941580758e-05,
+      "loss": 0.4239,
+      "step": 462
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4268041237113404e-05,
+      "loss": 0.4052,
+      "step": 463
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.425429553264605e-05,
+      "loss": 0.4138,
+      "step": 464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4240549828178696e-05,
+      "loss": 0.3748,
+      "step": 465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4226804123711342e-05,
+      "loss": 0.4377,
+      "step": 466
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4213058419243988e-05,
+      "loss": 0.4111,
+      "step": 467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4199312714776634e-05,
+      "loss": 0.4384,
+      "step": 468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.418556701030928e-05,
+      "loss": 0.3892,
+      "step": 469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4171821305841926e-05,
+      "loss": 0.3991,
+      "step": 470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4158075601374572e-05,
+      "loss": 0.3946,
+      "step": 471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4144329896907218e-05,
+      "loss": 0.4113,
+      "step": 472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4130584192439864e-05,
+      "loss": 0.4246,
+      "step": 473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.411683848797251e-05,
+      "loss": 0.4082,
+      "step": 474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4103092783505156e-05,
+      "loss": 0.4136,
+      "step": 475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4089347079037802e-05,
+      "loss": 0.4156,
+      "step": 476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4075601374570448e-05,
+      "loss": 0.4118,
+      "step": 477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4061855670103093e-05,
+      "loss": 0.4023,
+      "step": 478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4048109965635739e-05,
+      "loss": 0.4109,
+      "step": 479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4034364261168385e-05,
+      "loss": 0.4113,
+      "step": 480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.402061855670103e-05,
+      "loss": 0.3979,
+      "step": 481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4006872852233677e-05,
+      "loss": 0.4,
+      "step": 482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3993127147766323e-05,
+      "loss": 0.4062,
+      "step": 483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3979381443298969e-05,
+      "loss": 0.4081,
+      "step": 484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3965635738831615e-05,
+      "loss": 0.4271,
+      "step": 485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3951890034364261e-05,
+      "loss": 0.3965,
+      "step": 486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3938144329896907e-05,
+      "loss": 0.4165,
+      "step": 487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3924398625429555e-05,
+      "loss": 0.4135,
+      "step": 488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.39106529209622e-05,
+      "loss": 0.4288,
+      "step": 489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3896907216494847e-05,
+      "loss": 0.4098,
+      "step": 490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3883161512027493e-05,
+      "loss": 0.3886,
+      "step": 491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3869415807560139e-05,
+      "loss": 0.4091,
+      "step": 492
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3855670103092785e-05,
+      "loss": 0.4095,
+      "step": 493
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.384192439862543e-05,
+      "loss": 0.3753,
+      "step": 494
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3828178694158077e-05,
+      "loss": 0.3702,
+      "step": 495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3814432989690723e-05,
+      "loss": 0.3959,
+      "step": 496
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3800687285223369e-05,
+      "loss": 0.4138,
+      "step": 497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3786941580756015e-05,
+      "loss": 0.407,
+      "step": 498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.377319587628866e-05,
+      "loss": 0.4074,
+      "step": 499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3759450171821307e-05,
+      "loss": 0.3735,
+      "step": 500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3745704467353953e-05,
+      "loss": 0.3048,
+      "step": 501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3731958762886599e-05,
+      "loss": 0.3089,
+      "step": 502
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3718213058419245e-05,
+      "loss": 0.2987,
+      "step": 503
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.370446735395189e-05,
+      "loss": 0.302,
+      "step": 504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3690721649484537e-05,
+      "loss": 0.3222,
+      "step": 505
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3676975945017183e-05,
+      "loss": 0.3141,
+      "step": 506
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3663230240549829e-05,
+      "loss": 0.3035,
+      "step": 507
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3649484536082475e-05,
+      "loss": 0.3047,
+      "step": 508
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3635738831615123e-05,
+      "loss": 0.3209,
+      "step": 509
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3621993127147769e-05,
+      "loss": 0.3083,
+      "step": 510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3608247422680415e-05,
+      "loss": 0.2953,
+      "step": 511
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.359450171821306e-05,
+      "loss": 0.2921,
+      "step": 512
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3580756013745707e-05,
+      "loss": 0.3138,
+      "step": 513
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3567010309278353e-05,
+      "loss": 0.3101,
+      "step": 514
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3553264604810999e-05,
+      "loss": 0.3131,
+      "step": 515
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3539518900343645e-05,
+      "loss": 0.3005,
+      "step": 516
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.352577319587629e-05,
+      "loss": 0.3165,
+      "step": 517
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3512027491408937e-05,
+      "loss": 0.319,
+      "step": 518
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3498281786941583e-05,
+      "loss": 0.271,
+      "step": 519
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3484536082474229e-05,
+      "loss": 0.3016,
+      "step": 520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3470790378006875e-05,
+      "loss": 0.3189,
+      "step": 521
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.345704467353952e-05,
+      "loss": 0.2969,
+      "step": 522
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3443298969072167e-05,
+      "loss": 0.3164,
+      "step": 523
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3429553264604811e-05,
+      "loss": 0.3167,
+      "step": 524
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3415807560137457e-05,
+      "loss": 0.3121,
+      "step": 525
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3402061855670103e-05,
+      "loss": 0.3242,
+      "step": 526
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.338831615120275e-05,
+      "loss": 0.3011,
+      "step": 527
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3374570446735395e-05,
+      "loss": 0.3088,
+      "step": 528
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3360824742268041e-05,
+      "loss": 0.3021,
+      "step": 529
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3347079037800687e-05,
+      "loss": 0.3109,
+      "step": 530
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.32,
+      "step": 531
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.331958762886598e-05,
+      "loss": 0.3233,
+      "step": 532
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3305841924398625e-05,
+      "loss": 0.2987,
+      "step": 533
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3292096219931271e-05,
+      "loss": 0.32,
+      "step": 534
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3278350515463917e-05,
+      "loss": 0.2843,
+      "step": 535
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3264604810996563e-05,
+      "loss": 0.3412,
+      "step": 536
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3250859106529211e-05,
+      "loss": 0.3414,
+      "step": 537
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3237113402061857e-05,
+      "loss": 0.3137,
+      "step": 538
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3223367697594503e-05,
+      "loss": 0.2939,
+      "step": 539
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3209621993127149e-05,
+      "loss": 0.3125,
+      "step": 540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3195876288659795e-05,
+      "loss": 0.2995,
+      "step": 541
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3182130584192441e-05,
+      "loss": 0.3042,
+      "step": 542
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3168384879725087e-05,
+      "loss": 0.3253,
+      "step": 543
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3154639175257733e-05,
+      "loss": 0.2872,
+      "step": 544
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.314089347079038e-05,
+      "loss": 0.297,
+      "step": 545
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3127147766323025e-05,
+      "loss": 0.3021,
+      "step": 546
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3113402061855671e-05,
+      "loss": 0.3207,
+      "step": 547
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3099656357388317e-05,
+      "loss": 0.3019,
+      "step": 548
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3085910652920963e-05,
+      "loss": 0.3137,
+      "step": 549
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.307216494845361e-05,
+      "loss": 0.3058,
+      "step": 550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3058419243986255e-05,
+      "loss": 0.3103,
+      "step": 551
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3044673539518901e-05,
+      "loss": 0.3148,
+      "step": 552
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3030927835051547e-05,
+      "loss": 0.289,
+      "step": 553
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3017182130584193e-05,
+      "loss": 0.3201,
+      "step": 554
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.300343642611684e-05,
+      "loss": 0.3238,
+      "step": 555
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2989690721649485e-05,
+      "loss": 0.2845,
+      "step": 556
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2975945017182131e-05,
+      "loss": 0.3098,
+      "step": 557
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2962199312714777e-05,
+      "loss": 0.2914,
+      "step": 558
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2948453608247425e-05,
+      "loss": 0.3027,
+      "step": 559
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2934707903780071e-05,
+      "loss": 0.311,
+      "step": 560
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2920962199312717e-05,
+      "loss": 0.318,
+      "step": 561
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2907216494845363e-05,
+      "loss": 0.3164,
+      "step": 562
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.289347079037801e-05,
+      "loss": 0.3041,
+      "step": 563
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2879725085910655e-05,
+      "loss": 0.3039,
+      "step": 564
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2865979381443301e-05,
+      "loss": 0.3079,
+      "step": 565
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2852233676975947e-05,
+      "loss": 0.2934,
+      "step": 566
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2838487972508593e-05,
+      "loss": 0.3026,
+      "step": 567
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.282474226804124e-05,
+      "loss": 0.3324,
+      "step": 568
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2810996563573884e-05,
+      "loss": 0.3,
+      "step": 569
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.279725085910653e-05,
+      "loss": 0.3157,
+      "step": 570
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2783505154639176e-05,
+      "loss": 0.2861,
+      "step": 571
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2769759450171822e-05,
+      "loss": 0.321,
+      "step": 572
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2756013745704468e-05,
+      "loss": 0.3189,
+      "step": 573
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2742268041237114e-05,
+      "loss": 0.3049,
+      "step": 574
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.272852233676976e-05,
+      "loss": 0.3131,
+      "step": 575
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2714776632302406e-05,
+      "loss": 0.2974,
+      "step": 576
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2701030927835052e-05,
+      "loss": 0.3304,
+      "step": 577
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2687285223367698e-05,
+      "loss": 0.2986,
+      "step": 578
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2673539518900344e-05,
+      "loss": 0.3398,
+      "step": 579
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.265979381443299e-05,
+      "loss": 0.3135,
+      "step": 580
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2646048109965636e-05,
+      "loss": 0.3251,
+      "step": 581
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2632302405498282e-05,
+      "loss": 0.3017,
+      "step": 582
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2618556701030928e-05,
+      "loss": 0.3019,
+      "step": 583
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2604810996563574e-05,
+      "loss": 0.2909,
+      "step": 584
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.259106529209622e-05,
+      "loss": 0.2904,
+      "step": 585
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2577319587628866e-05,
+      "loss": 0.329,
+      "step": 586
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2563573883161514e-05,
+      "loss": 0.3224,
+      "step": 587
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.254982817869416e-05,
+      "loss": 0.2814,
+      "step": 588
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2536082474226806e-05,
+      "loss": 0.3032,
+      "step": 589
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2522336769759452e-05,
+      "loss": 0.3024,
+      "step": 590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2508591065292098e-05,
+      "loss": 0.3078,
+      "step": 591
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2494845360824744e-05,
+      "loss": 0.2789,
+      "step": 592
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.248109965635739e-05,
+      "loss": 0.3101,
+      "step": 593
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2467353951890036e-05,
+      "loss": 0.3091,
+      "step": 594
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2453608247422682e-05,
+      "loss": 0.3005,
+      "step": 595
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2439862542955328e-05,
+      "loss": 0.3174,
+      "step": 596
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2426116838487974e-05,
+      "loss": 0.282,
+      "step": 597
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.241237113402062e-05,
+      "loss": 0.3066,
+      "step": 598
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2398625429553266e-05,
+      "loss": 0.3081,
+      "step": 599
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2384879725085912e-05,
+      "loss": 0.3032,
+      "step": 600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2371134020618558e-05,
+      "loss": 0.3082,
+      "step": 601
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2357388316151204e-05,
+      "loss": 0.3148,
+      "step": 602
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.234364261168385e-05,
+      "loss": 0.3077,
+      "step": 603
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2329896907216496e-05,
+      "loss": 0.3135,
+      "step": 604
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2316151202749142e-05,
+      "loss": 0.2988,
+      "step": 605
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2302405498281788e-05,
+      "loss": 0.2941,
+      "step": 606
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2288659793814434e-05,
+      "loss": 0.3033,
+      "step": 607
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2274914089347082e-05,
+      "loss": 0.3073,
+      "step": 608
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2261168384879728e-05,
+      "loss": 0.3036,
+      "step": 609
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2247422680412374e-05,
+      "loss": 0.3068,
+      "step": 610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.223367697594502e-05,
+      "loss": 0.3126,
+      "step": 611
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2219931271477666e-05,
+      "loss": 0.3098,
+      "step": 612
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2206185567010312e-05,
+      "loss": 0.3132,
+      "step": 613
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2192439862542958e-05,
+      "loss": 0.2992,
+      "step": 614
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2178694158075602e-05,
+      "loss": 0.3198,
+      "step": 615
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2164948453608248e-05,
+      "loss": 0.3032,
+      "step": 616
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2151202749140894e-05,
+      "loss": 0.3052,
+      "step": 617
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.213745704467354e-05,
+      "loss": 0.3202,
+      "step": 618
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2123711340206186e-05,
+      "loss": 0.3092,
+      "step": 619
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2109965635738832e-05,
+      "loss": 0.3189,
+      "step": 620
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2096219931271478e-05,
+      "loss": 0.3199,
+      "step": 621
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2082474226804124e-05,
+      "loss": 0.2942,
+      "step": 622
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.206872852233677e-05,
+      "loss": 0.2943,
+      "step": 623
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2054982817869416e-05,
+      "loss": 0.3185,
+      "step": 624
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2041237113402062e-05,
+      "loss": 0.3142,
+      "step": 625
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2027491408934708e-05,
+      "loss": 0.3098,
+      "step": 626
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2013745704467354e-05,
+      "loss": 0.3229,
+      "step": 627
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2e-05,
+      "loss": 0.3104,
+      "step": 628
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1986254295532646e-05,
+      "loss": 0.3129,
+      "step": 629
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1972508591065292e-05,
+      "loss": 0.2859,
+      "step": 630
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1958762886597938e-05,
+      "loss": 0.2955,
+      "step": 631
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1945017182130584e-05,
+      "loss": 0.3025,
+      "step": 632
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.193127147766323e-05,
+      "loss": 0.308,
+      "step": 633
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1917525773195876e-05,
+      "loss": 0.315,
+      "step": 634
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1903780068728522e-05,
+      "loss": 0.3054,
+      "step": 635
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.189003436426117e-05,
+      "loss": 0.2839,
+      "step": 636
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1876288659793816e-05,
+      "loss": 0.283,
+      "step": 637
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1862542955326462e-05,
+      "loss": 0.2994,
+      "step": 638
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1848797250859108e-05,
+      "loss": 0.3191,
+      "step": 639
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1835051546391754e-05,
+      "loss": 0.2914,
+      "step": 640
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.18213058419244e-05,
+      "loss": 0.3247,
+      "step": 641
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1807560137457046e-05,
+      "loss": 0.3051,
+      "step": 642
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1793814432989692e-05,
+      "loss": 0.3209,
+      "step": 643
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1780068728522338e-05,
+      "loss": 0.2745,
+      "step": 644
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1766323024054984e-05,
+      "loss": 0.2799,
+      "step": 645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.175257731958763e-05,
+      "loss": 0.3082,
+      "step": 646
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1738831615120276e-05,
+      "loss": 0.3104,
+      "step": 647
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1725085910652922e-05,
+      "loss": 0.3179,
+      "step": 648
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1711340206185568e-05,
+      "loss": 0.3346,
+      "step": 649
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1697594501718214e-05,
+      "loss": 0.2843,
+      "step": 650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.168384879725086e-05,
+      "loss": 0.2854,
+      "step": 651
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1670103092783506e-05,
+      "loss": 0.3262,
+      "step": 652
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1656357388316152e-05,
+      "loss": 0.2977,
+      "step": 653
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1642611683848798e-05,
+      "loss": 0.3183,
+      "step": 654
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1628865979381444e-05,
+      "loss": 0.3152,
+      "step": 655
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.161512027491409e-05,
+      "loss": 0.3201,
+      "step": 656
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1601374570446736e-05,
+      "loss": 0.3079,
+      "step": 657
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1587628865979384e-05,
+      "loss": 0.3209,
+      "step": 658
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.157388316151203e-05,
+      "loss": 0.304,
+      "step": 659
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1560137457044673e-05,
+      "loss": 0.2986,
+      "step": 660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1546391752577319e-05,
+      "loss": 0.3218,
+      "step": 661
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1532646048109965e-05,
+      "loss": 0.3074,
+      "step": 662
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1518900343642611e-05,
+      "loss": 0.3063,
+      "step": 663
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1505154639175259e-05,
+      "loss": 0.3174,
+      "step": 664
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1491408934707905e-05,
+      "loss": 0.3014,
+      "step": 665
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.147766323024055e-05,
+      "loss": 0.3121,
+      "step": 666
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1463917525773197e-05,
+      "loss": 0.2967,
+      "step": 667
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1450171821305843e-05,
+      "loss": 0.3094,
+      "step": 668
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1436426116838489e-05,
+      "loss": 0.3079,
+      "step": 669
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1422680412371135e-05,
+      "loss": 0.2932,
+      "step": 670
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.140893470790378e-05,
+      "loss": 0.3065,
+      "step": 671
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1395189003436427e-05,
+      "loss": 0.2877,
+      "step": 672
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1381443298969073e-05,
+      "loss": 0.3085,
+      "step": 673
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1367697594501719e-05,
+      "loss": 0.2842,
+      "step": 674
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1353951890034365e-05,
+      "loss": 0.3294,
+      "step": 675
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.134020618556701e-05,
+      "loss": 0.2893,
+      "step": 676
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1326460481099657e-05,
+      "loss": 0.2976,
+      "step": 677
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1312714776632303e-05,
+      "loss": 0.2892,
+      "step": 678
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1298969072164949e-05,
+      "loss": 0.3114,
+      "step": 679
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1285223367697595e-05,
+      "loss": 0.3056,
+      "step": 680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1271477663230241e-05,
+      "loss": 0.2982,
+      "step": 681
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1257731958762887e-05,
+      "loss": 0.3075,
+      "step": 682
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1243986254295533e-05,
+      "loss": 0.2844,
+      "step": 683
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1230240549828179e-05,
+      "loss": 0.2907,
+      "step": 684
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1216494845360825e-05,
+      "loss": 0.2832,
+      "step": 685
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1202749140893473e-05,
+      "loss": 0.2931,
+      "step": 686
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1189003436426119e-05,
+      "loss": 0.3021,
+      "step": 687
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1175257731958765e-05,
+      "loss": 0.2978,
+      "step": 688
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.116151202749141e-05,
+      "loss": 0.3002,
+      "step": 689
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1147766323024057e-05,
+      "loss": 0.3008,
+      "step": 690
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1134020618556703e-05,
+      "loss": 0.3069,
+      "step": 691
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1120274914089349e-05,
+      "loss": 0.2891,
+      "step": 692
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1106529209621995e-05,
+      "loss": 0.2965,
+      "step": 693
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.109278350515464e-05,
+      "loss": 0.2922,
+      "step": 694
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1079037800687287e-05,
+      "loss": 0.3129,
+      "step": 695
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1065292096219933e-05,
+      "loss": 0.286,
+      "step": 696
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1051546391752579e-05,
+      "loss": 0.2899,
+      "step": 697
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1037800687285225e-05,
+      "loss": 0.2951,
+      "step": 698
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1024054982817871e-05,
+      "loss": 0.2918,
+      "step": 699
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1010309278350517e-05,
+      "loss": 0.2957,
+      "step": 700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0996563573883163e-05,
+      "loss": 0.292,
+      "step": 701
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0982817869415809e-05,
+      "loss": 0.3235,
+      "step": 702
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0969072164948455e-05,
+      "loss": 0.3063,
+      "step": 703
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0955326460481101e-05,
+      "loss": 0.3238,
+      "step": 704
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0941580756013747e-05,
+      "loss": 0.296,
+      "step": 705
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0927835051546391e-05,
+      "loss": 0.2892,
+      "step": 706
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0914089347079037e-05,
+      "loss": 0.3137,
+      "step": 707
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0900343642611683e-05,
+      "loss": 0.3045,
+      "step": 708
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.088659793814433e-05,
+      "loss": 0.2945,
+      "step": 709
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0872852233676975e-05,
+      "loss": 0.3039,
+      "step": 710
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0859106529209621e-05,
+      "loss": 0.2927,
+      "step": 711
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0845360824742267e-05,
+      "loss": 0.2893,
+      "step": 712
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0831615120274913e-05,
+      "loss": 0.2865,
+      "step": 713
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0817869415807561e-05,
+      "loss": 0.3033,
+      "step": 714
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0804123711340207e-05,
+      "loss": 0.2932,
+      "step": 715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0790378006872853e-05,
+      "loss": 0.2858,
+      "step": 716
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.07766323024055e-05,
+      "loss": 0.2665,
+      "step": 717
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0762886597938145e-05,
+      "loss": 0.3067,
+      "step": 718
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0749140893470791e-05,
+      "loss": 0.3025,
+      "step": 719
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0735395189003437e-05,
+      "loss": 0.3188,
+      "step": 720
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0721649484536083e-05,
+      "loss": 0.2834,
+      "step": 721
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.070790378006873e-05,
+      "loss": 0.2926,
+      "step": 722
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0694158075601375e-05,
+      "loss": 0.301,
+      "step": 723
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0680412371134021e-05,
+      "loss": 0.3215,
+      "step": 724
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.2908,
+      "step": 725
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0652920962199313e-05,
+      "loss": 0.3027,
+      "step": 726
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.063917525773196e-05,
+      "loss": 0.2907,
+      "step": 727
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0625429553264605e-05,
+      "loss": 0.3135,
+      "step": 728
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0611683848797251e-05,
+      "loss": 0.3014,
+      "step": 729
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0597938144329897e-05,
+      "loss": 0.297,
+      "step": 730
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0584192439862543e-05,
+      "loss": 0.3189,
+      "step": 731
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.057044673539519e-05,
+      "loss": 0.2903,
+      "step": 732
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0556701030927835e-05,
+      "loss": 0.2928,
+      "step": 733
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0542955326460481e-05,
+      "loss": 0.2952,
+      "step": 734
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.052920962199313e-05,
+      "loss": 0.2798,
+      "step": 735
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0515463917525775e-05,
+      "loss": 0.2814,
+      "step": 736
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0501718213058421e-05,
+      "loss": 0.3165,
+      "step": 737
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0487972508591067e-05,
+      "loss": 0.2987,
+      "step": 738
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0474226804123713e-05,
+      "loss": 0.3231,
+      "step": 739
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.046048109965636e-05,
+      "loss": 0.2987,
+      "step": 740
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0446735395189005e-05,
+      "loss": 0.2831,
+      "step": 741
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0432989690721651e-05,
+      "loss": 0.3128,
+      "step": 742
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0419243986254297e-05,
+      "loss": 0.2915,
+      "step": 743
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0405498281786943e-05,
+      "loss": 0.3251,
+      "step": 744
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.039175257731959e-05,
+      "loss": 0.3193,
+      "step": 745
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0378006872852235e-05,
+      "loss": 0.3012,
+      "step": 746
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0364261168384881e-05,
+      "loss": 0.3139,
+      "step": 747
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0350515463917527e-05,
+      "loss": 0.3289,
+      "step": 748
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0336769759450173e-05,
+      "loss": 0.3141,
+      "step": 749
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.032302405498282e-05,
+      "loss": 0.3171,
+      "step": 750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0309278350515464e-05,
+      "loss": 0.2882,
+      "step": 751
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.029553264604811e-05,
+      "loss": 0.2867,
+      "step": 752
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0281786941580756e-05,
+      "loss": 0.2988,
+      "step": 753
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0268041237113402e-05,
+      "loss": 0.3036,
+      "step": 754
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0254295532646048e-05,
+      "loss": 0.3122,
+      "step": 755
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0240549828178694e-05,
+      "loss": 0.2901,
+      "step": 756
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.022680412371134e-05,
+      "loss": 0.2926,
+      "step": 757
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0213058419243986e-05,
+      "loss": 0.3024,
+      "step": 758
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0199312714776632e-05,
+      "loss": 0.3097,
+      "step": 759
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0185567010309278e-05,
+      "loss": 0.2955,
+      "step": 760
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0171821305841924e-05,
+      "loss": 0.2826,
+      "step": 761
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.015807560137457e-05,
+      "loss": 0.2865,
+      "step": 762
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144329896907218e-05,
+      "loss": 0.2959,
+      "step": 763
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0130584192439864e-05,
+      "loss": 0.311,
+      "step": 764
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011683848797251e-05,
+      "loss": 0.3038,
+      "step": 765
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0103092783505156e-05,
+      "loss": 0.2843,
+      "step": 766
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0089347079037802e-05,
+      "loss": 0.2902,
+      "step": 767
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0075601374570448e-05,
+      "loss": 0.3095,
+      "step": 768
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0061855670103094e-05,
+      "loss": 0.2868,
+      "step": 769
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004810996563574e-05,
+      "loss": 0.2948,
+      "step": 770
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0034364261168386e-05,
+      "loss": 0.3027,
+      "step": 771
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0020618556701032e-05,
+      "loss": 0.2833,
+      "step": 772
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0006872852233678e-05,
+      "loss": 0.3163,
+      "step": 773
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.993127147766324e-06,
+      "loss": 0.3088,
+      "step": 774
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97938144329897e-06,
+      "loss": 0.2981,
+      "step": 775
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965635738831616e-06,
+      "loss": 0.2814,
+      "step": 776
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.951890034364262e-06,
+      "loss": 0.3211,
+      "step": 777
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.938144329896908e-06,
+      "loss": 0.3138,
+      "step": 778
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.924398625429554e-06,
+      "loss": 0.3039,
+      "step": 779
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.9106529209622e-06,
+      "loss": 0.3093,
+      "step": 780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.896907216494846e-06,
+      "loss": 0.2711,
+      "step": 781
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.883161512027492e-06,
+      "loss": 0.2703,
+      "step": 782
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.869415807560138e-06,
+      "loss": 0.2932,
+      "step": 783
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.855670103092784e-06,
+      "loss": 0.3135,
+      "step": 784
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.84192439862543e-06,
+      "loss": 0.3132,
+      "step": 785
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.828178694158076e-06,
+      "loss": 0.2907,
+      "step": 786
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.814432989690722e-06,
+      "loss": 0.2927,
+      "step": 787
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.800687285223368e-06,
+      "loss": 0.2941,
+      "step": 788
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.786941580756014e-06,
+      "loss": 0.3037,
+      "step": 789
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.77319587628866e-06,
+      "loss": 0.3084,
+      "step": 790
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.759450171821306e-06,
+      "loss": 0.2987,
+      "step": 791
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.745704467353952e-06,
+      "loss": 0.2896,
+      "step": 792
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.731958762886598e-06,
+      "loss": 0.296,
+      "step": 793
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.718213058419244e-06,
+      "loss": 0.2925,
+      "step": 794
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.70446735395189e-06,
+      "loss": 0.2981,
+      "step": 795
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.690721649484536e-06,
+      "loss": 0.301,
+      "step": 796
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.676975945017182e-06,
+      "loss": 0.3101,
+      "step": 797
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.663230240549828e-06,
+      "loss": 0.2939,
+      "step": 798
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.649484536082476e-06,
+      "loss": 0.2997,
+      "step": 799
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.635738831615122e-06,
+      "loss": 0.2837,
+      "step": 800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.621993127147768e-06,
+      "loss": 0.2822,
+      "step": 801
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.608247422680414e-06,
+      "loss": 0.2877,
+      "step": 802
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.59450171821306e-06,
+      "loss": 0.2919,
+      "step": 803
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.580756013745706e-06,
+      "loss": 0.2934,
+      "step": 804
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.567010309278352e-06,
+      "loss": 0.2604,
+      "step": 805
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.553264604810998e-06,
+      "loss": 0.2968,
+      "step": 806
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.539518900343644e-06,
+      "loss": 0.2994,
+      "step": 807
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.525773195876288e-06,
+      "loss": 0.2806,
+      "step": 808
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.512027491408934e-06,
+      "loss": 0.31,
+      "step": 809
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.49828178694158e-06,
+      "loss": 0.2672,
+      "step": 810
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.484536082474226e-06,
+      "loss": 0.2963,
+      "step": 811
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.470790378006872e-06,
+      "loss": 0.2724,
+      "step": 812
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.45704467353952e-06,
+      "loss": 0.2782,
+      "step": 813
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.443298969072166e-06,
+      "loss": 0.3032,
+      "step": 814
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.429553264604812e-06,
+      "loss": 0.3017,
+      "step": 815
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.415807560137458e-06,
+      "loss": 0.2905,
+      "step": 816
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.402061855670104e-06,
+      "loss": 0.2877,
+      "step": 817
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.38831615120275e-06,
+      "loss": 0.3207,
+      "step": 818
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.374570446735396e-06,
+      "loss": 0.2991,
+      "step": 819
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.360824742268042e-06,
+      "loss": 0.3008,
+      "step": 820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.347079037800688e-06,
+      "loss": 0.2794,
+      "step": 821
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.2953,
+      "step": 822
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.31958762886598e-06,
+      "loss": 0.2958,
+      "step": 823
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.305841924398626e-06,
+      "loss": 0.2884,
+      "step": 824
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.292096219931272e-06,
+      "loss": 0.2944,
+      "step": 825
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.278350515463918e-06,
+      "loss": 0.2805,
+      "step": 826
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.264604810996564e-06,
+      "loss": 0.2896,
+      "step": 827
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.25085910652921e-06,
+      "loss": 0.2947,
+      "step": 828
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.237113402061856e-06,
+      "loss": 0.278,
+      "step": 829
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.223367697594502e-06,
+      "loss": 0.2983,
+      "step": 830
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.209621993127148e-06,
+      "loss": 0.291,
+      "step": 831
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.195876288659794e-06,
+      "loss": 0.2753,
+      "step": 832
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.18213058419244e-06,
+      "loss": 0.2974,
+      "step": 833
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.168384879725087e-06,
+      "loss": 0.2958,
+      "step": 834
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.154639175257733e-06,
+      "loss": 0.2813,
+      "step": 835
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.140893470790379e-06,
+      "loss": 0.2941,
+      "step": 836
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.127147766323025e-06,
+      "loss": 0.2854,
+      "step": 837
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.11340206185567e-06,
+      "loss": 0.2886,
+      "step": 838
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.099656357388317e-06,
+      "loss": 0.2921,
+      "step": 839
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.085910652920963e-06,
+      "loss": 0.2782,
+      "step": 840
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.072164948453609e-06,
+      "loss": 0.2675,
+      "step": 841
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.058419243986255e-06,
+      "loss": 0.283,
+      "step": 842
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.0446735395189e-06,
+      "loss": 0.2853,
+      "step": 843
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.030927835051547e-06,
+      "loss": 0.2843,
+      "step": 844
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.017182130584193e-06,
+      "loss": 0.2862,
+      "step": 845
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.003436426116839e-06,
+      "loss": 0.297,
+      "step": 846
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.989690721649485e-06,
+      "loss": 0.2762,
+      "step": 847
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.975945017182132e-06,
+      "loss": 0.3105,
+      "step": 848
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.962199312714778e-06,
+      "loss": 0.2957,
+      "step": 849
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.948453608247424e-06,
+      "loss": 0.3049,
+      "step": 850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.93470790378007e-06,
+      "loss": 0.2982,
+      "step": 851
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.920962199312717e-06,
+      "loss": 0.2891,
+      "step": 852
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.907216494845363e-06,
+      "loss": 0.3057,
+      "step": 853
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.893470790378007e-06,
+      "loss": 0.3066,
+      "step": 854
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.879725085910653e-06,
+      "loss": 0.3129,
+      "step": 855
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.865979381443299e-06,
+      "loss": 0.2898,
+      "step": 856
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.852233676975945e-06,
+      "loss": 0.3087,
+      "step": 857
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.838487972508591e-06,
+      "loss": 0.288,
+      "step": 858
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.824742268041237e-06,
+      "loss": 0.2792,
+      "step": 859
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.810996563573883e-06,
+      "loss": 0.2854,
+      "step": 860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.797250859106529e-06,
+      "loss": 0.3028,
+      "step": 861
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.783505154639177e-06,
+      "loss": 0.2694,
+      "step": 862
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.769759450171823e-06,
+      "loss": 0.2884,
+      "step": 863
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.756013745704469e-06,
+      "loss": 0.3002,
+      "step": 864
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.742268041237115e-06,
+      "loss": 0.2718,
+      "step": 865
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.72852233676976e-06,
+      "loss": 0.2806,
+      "step": 866
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.714776632302407e-06,
+      "loss": 0.3045,
+      "step": 867
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.701030927835053e-06,
+      "loss": 0.2956,
+      "step": 868
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.687285223367699e-06,
+      "loss": 0.2723,
+      "step": 869
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.673539518900345e-06,
+      "loss": 0.2936,
+      "step": 870
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.65979381443299e-06,
+      "loss": 0.2869,
+      "step": 871
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.646048109965637e-06,
+      "loss": 0.2735,
+      "step": 872
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.632302405498283e-06,
+      "loss": 0.2896,
+      "step": 873
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.618556701030929e-06,
+      "loss": 0.3225,
+      "step": 874
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.604810996563575e-06,
+      "loss": 0.269,
+      "step": 875
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.591065292096221e-06,
+      "loss": 0.2713,
+      "step": 876
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.577319587628867e-06,
+      "loss": 0.2889,
+      "step": 877
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.563573883161513e-06,
+      "loss": 0.2766,
+      "step": 878
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.549828178694159e-06,
+      "loss": 0.2857,
+      "step": 879
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.536082474226805e-06,
+      "loss": 0.2806,
+      "step": 880
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.522336769759451e-06,
+      "loss": 0.2903,
+      "step": 881
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.508591065292097e-06,
+      "loss": 0.2797,
+      "step": 882
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.494845360824743e-06,
+      "loss": 0.2625,
+      "step": 883
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.481099656357389e-06,
+      "loss": 0.2994,
+      "step": 884
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.467353951890035e-06,
+      "loss": 0.2812,
+      "step": 885
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.453608247422681e-06,
+      "loss": 0.2909,
+      "step": 886
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.439862542955327e-06,
+      "loss": 0.2888,
+      "step": 887
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.426116838487973e-06,
+      "loss": 0.2949,
+      "step": 888
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.412371134020619e-06,
+      "loss": 0.2794,
+      "step": 889
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.398625429553265e-06,
+      "loss": 0.2979,
+      "step": 890
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.384879725085911e-06,
+      "loss": 0.2906,
+      "step": 891
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.371134020618557e-06,
+      "loss": 0.3013,
+      "step": 892
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.357388316151203e-06,
+      "loss": 0.2774,
+      "step": 893
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.34364261168385e-06,
+      "loss": 0.303,
+      "step": 894
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.329896907216495e-06,
+      "loss": 0.2754,
+      "step": 895
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.316151202749141e-06,
+      "loss": 0.2942,
+      "step": 896
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.302405498281787e-06,
+      "loss": 0.2791,
+      "step": 897
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.288659793814435e-06,
+      "loss": 0.2822,
+      "step": 898
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.27491408934708e-06,
+      "loss": 0.2825,
+      "step": 899
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.261168384879725e-06,
+      "loss": 0.2978,
+      "step": 900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.247422680412371e-06,
+      "loss": 0.2764,
+      "step": 901
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.233676975945017e-06,
+      "loss": 0.2636,
+      "step": 902
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.219931271477663e-06,
+      "loss": 0.2659,
+      "step": 903
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.20618556701031e-06,
+      "loss": 0.2905,
+      "step": 904
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.192439862542955e-06,
+      "loss": 0.2952,
+      "step": 905
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.178694158075601e-06,
+      "loss": 0.2711,
+      "step": 906
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.164948453608247e-06,
+      "loss": 0.2543,
+      "step": 907
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.151202749140893e-06,
+      "loss": 0.2923,
+      "step": 908
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.13745704467354e-06,
+      "loss": 0.302,
+      "step": 909
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.123711340206185e-06,
+      "loss": 0.2835,
+      "step": 910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.109965635738832e-06,
+      "loss": 0.2969,
+      "step": 911
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.09621993127148e-06,
+      "loss": 0.2927,
+      "step": 912
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.082474226804125e-06,
+      "loss": 0.3285,
+      "step": 913
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.068728522336771e-06,
+      "loss": 0.2965,
+      "step": 914
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.054982817869417e-06,
+      "loss": 0.2886,
+      "step": 915
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.041237113402063e-06,
+      "loss": 0.2816,
+      "step": 916
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.02749140893471e-06,
+      "loss": 0.2863,
+      "step": 917
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.013745704467355e-06,
+      "loss": 0.3111,
+      "step": 918
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.2972,
+      "step": 919
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.986254295532647e-06,
+      "loss": 0.2933,
+      "step": 920
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.972508591065293e-06,
+      "loss": 0.2837,
+      "step": 921
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.958762886597938e-06,
+      "loss": 0.3022,
+      "step": 922
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.945017182130584e-06,
+      "loss": 0.2893,
+      "step": 923
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.93127147766323e-06,
+      "loss": 0.2955,
+      "step": 924
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.917525773195876e-06,
+      "loss": 0.3019,
+      "step": 925
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.903780068728523e-06,
+      "loss": 0.2831,
+      "step": 926
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.89003436426117e-06,
+      "loss": 0.2847,
+      "step": 927
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.876288659793815e-06,
+      "loss": 0.2898,
+      "step": 928
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.862542955326461e-06,
+      "loss": 0.2775,
+      "step": 929
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.848797250859108e-06,
+      "loss": 0.2881,
+      "step": 930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.835051546391754e-06,
+      "loss": 0.2839,
+      "step": 931
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.8213058419244e-06,
+      "loss": 0.2966,
+      "step": 932
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.807560137457046e-06,
+      "loss": 0.2859,
+      "step": 933
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.793814432989692e-06,
+      "loss": 0.2843,
+      "step": 934
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.780068728522338e-06,
+      "loss": 0.2777,
+      "step": 935
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.766323024054984e-06,
+      "loss": 0.282,
+      "step": 936
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.75257731958763e-06,
+      "loss": 0.2998,
+      "step": 937
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.738831615120276e-06,
+      "loss": 0.2737,
+      "step": 938
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.725085910652922e-06,
+      "loss": 0.2845,
+      "step": 939
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.711340206185568e-06,
+      "loss": 0.2674,
+      "step": 940
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.697594501718214e-06,
+      "loss": 0.3024,
+      "step": 941
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.68384879725086e-06,
+      "loss": 0.2794,
+      "step": 942
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.670103092783506e-06,
+      "loss": 0.2761,
+      "step": 943
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.656357388316152e-06,
+      "loss": 0.2796,
+      "step": 944
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.642611683848798e-06,
+      "loss": 0.2989,
+      "step": 945
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.628865979381444e-06,
+      "loss": 0.2801,
+      "step": 946
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.61512027491409e-06,
+      "loss": 0.2722,
+      "step": 947
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.601374570446736e-06,
+      "loss": 0.2876,
+      "step": 948
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.587628865979382e-06,
+      "loss": 0.298,
+      "step": 949
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.573883161512028e-06,
+      "loss": 0.2808,
+      "step": 950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.560137457044674e-06,
+      "loss": 0.2645,
+      "step": 951
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.54639175257732e-06,
+      "loss": 0.2944,
+      "step": 952
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.532646048109966e-06,
+      "loss": 0.2955,
+      "step": 953
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.518900343642612e-06,
+      "loss": 0.2845,
+      "step": 954
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.505154639175258e-06,
+      "loss": 0.2885,
+      "step": 955
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.491408934707905e-06,
+      "loss": 0.281,
+      "step": 956
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.477663230240551e-06,
+      "loss": 0.2727,
+      "step": 957
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.463917525773197e-06,
+      "loss": 0.2818,
+      "step": 958
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.450171821305843e-06,
+      "loss": 0.2675,
+      "step": 959
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.436426116838489e-06,
+      "loss": 0.2743,
+      "step": 960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.422680412371135e-06,
+      "loss": 0.271,
+      "step": 961
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.408934707903781e-06,
+      "loss": 0.2865,
+      "step": 962
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.395189003436427e-06,
+      "loss": 0.2863,
+      "step": 963
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.381443298969073e-06,
+      "loss": 0.2916,
+      "step": 964
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.367697594501719e-06,
+      "loss": 0.2865,
+      "step": 965
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.353951890034365e-06,
+      "loss": 0.2917,
+      "step": 966
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.34020618556701e-06,
+      "loss": 0.2803,
+      "step": 967
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.326460481099656e-06,
+      "loss": 0.2799,
+      "step": 968
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.312714776632302e-06,
+      "loss": 0.2778,
+      "step": 969
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.298969072164949e-06,
+      "loss": 0.2698,
+      "step": 970
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.285223367697595e-06,
+      "loss": 0.2791,
+      "step": 971
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.271477663230241e-06,
+      "loss": 0.2923,
+      "step": 972
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.257731958762887e-06,
+      "loss": 0.2645,
+      "step": 973
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.243986254295533e-06,
+      "loss": 0.2706,
+      "step": 974
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.230240549828179e-06,
+      "loss": 0.274,
+      "step": 975
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.216494845360825e-06,
+      "loss": 0.2905,
+      "step": 976
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.202749140893471e-06,
+      "loss": 0.2957,
+      "step": 977
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.189003436426117e-06,
+      "loss": 0.3161,
+      "step": 978
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.175257731958763e-06,
+      "loss": 0.305,
+      "step": 979
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.161512027491409e-06,
+      "loss": 0.2804,
+      "step": 980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.147766323024056e-06,
+      "loss": 0.2914,
+      "step": 981
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.134020618556702e-06,
+      "loss": 0.2854,
+      "step": 982
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.120274914089348e-06,
+      "loss": 0.2684,
+      "step": 983
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.106529209621994e-06,
+      "loss": 0.3057,
+      "step": 984
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.09278350515464e-06,
+      "loss": 0.2832,
+      "step": 985
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.079037800687286e-06,
+      "loss": 0.2738,
+      "step": 986
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.065292096219932e-06,
+      "loss": 0.2687,
+      "step": 987
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.051546391752578e-06,
+      "loss": 0.2789,
+      "step": 988
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.037800687285224e-06,
+      "loss": 0.2892,
+      "step": 989
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.024054982817869e-06,
+      "loss": 0.2956,
+      "step": 990
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.010309278350515e-06,
+      "loss": 0.2617,
+      "step": 991
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.996563573883161e-06,
+      "loss": 0.28,
+      "step": 992
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.982817869415807e-06,
+      "loss": 0.2701,
+      "step": 993
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9690721649484534e-06,
+      "loss": 0.2623,
+      "step": 994
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9553264604811e-06,
+      "loss": 0.2674,
+      "step": 995
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.941580756013746e-06,
+      "loss": 0.283,
+      "step": 996
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.927835051546392e-06,
+      "loss": 0.2673,
+      "step": 997
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.914089347079038e-06,
+      "loss": 0.2836,
+      "step": 998
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.900343642611684e-06,
+      "loss": 0.281,
+      "step": 999
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.88659793814433e-06,
+      "loss": 0.267,
+      "step": 1000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.872852233676976e-06,
+      "loss": 0.1981,
+      "step": 1001
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.859106529209622e-06,
+      "loss": 0.183,
+      "step": 1002
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.8453608247422684e-06,
+      "loss": 0.1764,
+      "step": 1003
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.8316151202749144e-06,
+      "loss": 0.1794,
+      "step": 1004
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.817869415807561e-06,
+      "loss": 0.1695,
+      "step": 1005
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.804123711340207e-06,
+      "loss": 0.1596,
+      "step": 1006
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.790378006872853e-06,
+      "loss": 0.1861,
+      "step": 1007
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.776632302405499e-06,
+      "loss": 0.1833,
+      "step": 1008
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.762886597938145e-06,
+      "loss": 0.1758,
+      "step": 1009
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.749140893470791e-06,
+      "loss": 0.1736,
+      "step": 1010
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.735395189003437e-06,
+      "loss": 0.1699,
+      "step": 1011
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7216494845360834e-06,
+      "loss": 0.1801,
+      "step": 1012
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.707903780068729e-06,
+      "loss": 0.1759,
+      "step": 1013
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.694158075601375e-06,
+      "loss": 0.1803,
+      "step": 1014
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.680412371134021e-06,
+      "loss": 0.1693,
+      "step": 1015
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.1713,
+      "step": 1016
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.652920962199313e-06,
+      "loss": 0.1616,
+      "step": 1017
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.639175257731959e-06,
+      "loss": 0.1754,
+      "step": 1018
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6254295532646055e-06,
+      "loss": 0.1687,
+      "step": 1019
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6116838487972516e-06,
+      "loss": 0.1733,
+      "step": 1020
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.597938144329898e-06,
+      "loss": 0.1709,
+      "step": 1021
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.584192439862544e-06,
+      "loss": 0.1646,
+      "step": 1022
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.57044673539519e-06,
+      "loss": 0.1691,
+      "step": 1023
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.556701030927836e-06,
+      "loss": 0.1649,
+      "step": 1024
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.542955326460482e-06,
+      "loss": 0.1773,
+      "step": 1025
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.529209621993128e-06,
+      "loss": 0.1617,
+      "step": 1026
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.515463917525774e-06,
+      "loss": 0.1579,
+      "step": 1027
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.50171821305842e-06,
+      "loss": 0.164,
+      "step": 1028
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.487972508591066e-06,
+      "loss": 0.1768,
+      "step": 1029
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4742268041237126e-06,
+      "loss": 0.1798,
+      "step": 1030
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.460481099656359e-06,
+      "loss": 0.1633,
+      "step": 1031
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.446735395189005e-06,
+      "loss": 0.1795,
+      "step": 1032
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.432989690721651e-06,
+      "loss": 0.1803,
+      "step": 1033
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.419243986254297e-06,
+      "loss": 0.175,
+      "step": 1034
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.405498281786942e-06,
+      "loss": 0.1867,
+      "step": 1035
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.391752577319588e-06,
+      "loss": 0.1577,
+      "step": 1036
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.378006872852234e-06,
+      "loss": 0.1738,
+      "step": 1037
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.36426116838488e-06,
+      "loss": 0.1718,
+      "step": 1038
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.350515463917526e-06,
+      "loss": 0.1761,
+      "step": 1039
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.336769759450172e-06,
+      "loss": 0.1793,
+      "step": 1040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.323024054982818e-06,
+      "loss": 0.1811,
+      "step": 1041
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.309278350515464e-06,
+      "loss": 0.177,
+      "step": 1042
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.29553264604811e-06,
+      "loss": 0.1643,
+      "step": 1043
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.281786941580757e-06,
+      "loss": 0.1681,
+      "step": 1044
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.268041237113403e-06,
+      "loss": 0.1717,
+      "step": 1045
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.254295532646049e-06,
+      "loss": 0.1594,
+      "step": 1046
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.240549828178695e-06,
+      "loss": 0.1779,
+      "step": 1047
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.226804123711341e-06,
+      "loss": 0.1798,
+      "step": 1048
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.213058419243987e-06,
+      "loss": 0.157,
+      "step": 1049
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.199312714776633e-06,
+      "loss": 0.1758,
+      "step": 1050
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.185567010309279e-06,
+      "loss": 0.1676,
+      "step": 1051
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.171821305841925e-06,
+      "loss": 0.1718,
+      "step": 1052
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.158075601374571e-06,
+      "loss": 0.1594,
+      "step": 1053
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.144329896907217e-06,
+      "loss": 0.1658,
+      "step": 1054
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.130584192439864e-06,
+      "loss": 0.1711,
+      "step": 1055
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.11683848797251e-06,
+      "loss": 0.1621,
+      "step": 1056
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.103092783505156e-06,
+      "loss": 0.1642,
+      "step": 1057
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.089347079037801e-06,
+      "loss": 0.1713,
+      "step": 1058
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.075601374570447e-06,
+      "loss": 0.1744,
+      "step": 1059
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.061855670103093e-06,
+      "loss": 0.1693,
+      "step": 1060
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.048109965635739e-06,
+      "loss": 0.1619,
+      "step": 1061
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.034364261168385e-06,
+      "loss": 0.1754,
+      "step": 1062
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.020618556701031e-06,
+      "loss": 0.1731,
+      "step": 1063
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.006872852233677e-06,
+      "loss": 0.178,
+      "step": 1064
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.993127147766323e-06,
+      "loss": 0.1647,
+      "step": 1065
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.979381443298969e-06,
+      "loss": 0.1813,
+      "step": 1066
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.965635738831615e-06,
+      "loss": 0.172,
+      "step": 1067
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.951890034364261e-06,
+      "loss": 0.1608,
+      "step": 1068
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.938144329896908e-06,
+      "loss": 0.1611,
+      "step": 1069
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.924398625429554e-06,
+      "loss": 0.1618,
+      "step": 1070
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.9106529209622e-06,
+      "loss": 0.1689,
+      "step": 1071
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.896907216494846e-06,
+      "loss": 0.1801,
+      "step": 1072
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.883161512027492e-06,
+      "loss": 0.1694,
+      "step": 1073
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.869415807560138e-06,
+      "loss": 0.1788,
+      "step": 1074
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.855670103092784e-06,
+      "loss": 0.1732,
+      "step": 1075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.84192439862543e-06,
+      "loss": 0.1771,
+      "step": 1076
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.828178694158076e-06,
+      "loss": 0.1626,
+      "step": 1077
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.814432989690722e-06,
+      "loss": 0.1715,
+      "step": 1078
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.800687285223368e-06,
+      "loss": 0.1801,
+      "step": 1079
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.786941580756015e-06,
+      "loss": 0.1589,
+      "step": 1080
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7731958762886594e-06,
+      "loss": 0.1667,
+      "step": 1081
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7594501718213054e-06,
+      "loss": 0.184,
+      "step": 1082
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.745704467353952e-06,
+      "loss": 0.1662,
+      "step": 1083
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.731958762886598e-06,
+      "loss": 0.1718,
+      "step": 1084
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.718213058419244e-06,
+      "loss": 0.1981,
+      "step": 1085
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.70446735395189e-06,
+      "loss": 0.1625,
+      "step": 1086
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.690721649484536e-06,
+      "loss": 0.161,
+      "step": 1087
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.676975945017182e-06,
+      "loss": 0.1829,
+      "step": 1088
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.663230240549828e-06,
+      "loss": 0.1736,
+      "step": 1089
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6494845360824744e-06,
+      "loss": 0.1646,
+      "step": 1090
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6357388316151204e-06,
+      "loss": 0.1561,
+      "step": 1091
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6219931271477665e-06,
+      "loss": 0.1611,
+      "step": 1092
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.6082474226804125e-06,
+      "loss": 0.1624,
+      "step": 1093
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.594501718213059e-06,
+      "loss": 0.1671,
+      "step": 1094
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.580756013745705e-06,
+      "loss": 0.1724,
+      "step": 1095
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.567010309278351e-06,
+      "loss": 0.1742,
+      "step": 1096
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.553264604810997e-06,
+      "loss": 0.171,
+      "step": 1097
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.539518900343643e-06,
+      "loss": 0.1709,
+      "step": 1098
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.525773195876289e-06,
+      "loss": 0.1618,
+      "step": 1099
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.5120274914089354e-06,
+      "loss": 0.1689,
+      "step": 1100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.4982817869415815e-06,
+      "loss": 0.1684,
+      "step": 1101
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.4845360824742275e-06,
+      "loss": 0.1637,
+      "step": 1102
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.4707903780068735e-06,
+      "loss": 0.1656,
+      "step": 1103
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.457044673539519e-06,
+      "loss": 0.1733,
+      "step": 1104
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.443298969072165e-06,
+      "loss": 0.173,
+      "step": 1105
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.429553264604811e-06,
+      "loss": 0.1721,
+      "step": 1106
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.415807560137457e-06,
+      "loss": 0.1727,
+      "step": 1107
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.4020618556701036e-06,
+      "loss": 0.1785,
+      "step": 1108
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.38831615120275e-06,
+      "loss": 0.1634,
+      "step": 1109
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.374570446735396e-06,
+      "loss": 0.1819,
+      "step": 1110
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.360824742268042e-06,
+      "loss": 0.1757,
+      "step": 1111
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.347079037800688e-06,
+      "loss": 0.1656,
+      "step": 1112
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.1751,
+      "step": 1113
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.31958762886598e-06,
+      "loss": 0.1922,
+      "step": 1114
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.305841924398626e-06,
+      "loss": 0.1737,
+      "step": 1115
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.292096219931272e-06,
+      "loss": 0.176,
+      "step": 1116
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.278350515463918e-06,
+      "loss": 0.1773,
+      "step": 1117
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.264604810996565e-06,
+      "loss": 0.1609,
+      "step": 1118
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.250859106529211e-06,
+      "loss": 0.1706,
+      "step": 1119
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.237113402061857e-06,
+      "loss": 0.179,
+      "step": 1120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.223367697594503e-06,
+      "loss": 0.1645,
+      "step": 1121
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.209621993127149e-06,
+      "loss": 0.162,
+      "step": 1122
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.195876288659795e-06,
+      "loss": 0.1669,
+      "step": 1123
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.182130584192441e-06,
+      "loss": 0.175,
+      "step": 1124
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.168384879725087e-06,
+      "loss": 0.1667,
+      "step": 1125
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.154639175257732e-06,
+      "loss": 0.1609,
+      "step": 1126
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.140893470790378e-06,
+      "loss": 0.162,
+      "step": 1127
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.127147766323024e-06,
+      "loss": 0.1683,
+      "step": 1128
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.11340206185567e-06,
+      "loss": 0.1681,
+      "step": 1129
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.099656357388316e-06,
+      "loss": 0.1544,
+      "step": 1130
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.085910652920962e-06,
+      "loss": 0.1786,
+      "step": 1131
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.072164948453609e-06,
+      "loss": 0.1554,
+      "step": 1132
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.058419243986255e-06,
+      "loss": 0.1738,
+      "step": 1133
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.044673539518901e-06,
+      "loss": 0.165,
+      "step": 1134
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.030927835051547e-06,
+      "loss": 0.1545,
+      "step": 1135
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.017182130584193e-06,
+      "loss": 0.1636,
+      "step": 1136
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.003436426116839e-06,
+      "loss": 0.1746,
+      "step": 1137
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.989690721649485e-06,
+      "loss": 0.1714,
+      "step": 1138
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.975945017182131e-06,
+      "loss": 0.1422,
+      "step": 1139
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.962199312714777e-06,
+      "loss": 0.1649,
+      "step": 1140
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.948453608247423e-06,
+      "loss": 0.154,
+      "step": 1141
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.934707903780069e-06,
+      "loss": 0.168,
+      "step": 1142
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.920962199312715e-06,
+      "loss": 0.1736,
+      "step": 1143
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.907216494845361e-06,
+      "loss": 0.1682,
+      "step": 1144
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.893470790378007e-06,
+      "loss": 0.151,
+      "step": 1145
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.879725085910653e-06,
+      "loss": 0.1644,
+      "step": 1146
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.865979381443299e-06,
+      "loss": 0.1832,
+      "step": 1147
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.852233676975945e-06,
+      "loss": 0.184,
+      "step": 1148
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.838487972508591e-06,
+      "loss": 0.1721,
+      "step": 1149
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.824742268041238e-06,
+      "loss": 0.1685,
+      "step": 1150
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.810996563573884e-06,
+      "loss": 0.1726,
+      "step": 1151
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.79725085910653e-06,
+      "loss": 0.1653,
+      "step": 1152
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.783505154639176e-06,
+      "loss": 0.1681,
+      "step": 1153
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.769759450171822e-06,
+      "loss": 0.1691,
+      "step": 1154
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.756013745704467e-06,
+      "loss": 0.1539,
+      "step": 1155
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.742268041237113e-06,
+      "loss": 0.1734,
+      "step": 1156
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.72852233676976e-06,
+      "loss": 0.1714,
+      "step": 1157
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.714776632302406e-06,
+      "loss": 0.1794,
+      "step": 1158
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.701030927835052e-06,
+      "loss": 0.1612,
+      "step": 1159
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.687285223367698e-06,
+      "loss": 0.1833,
+      "step": 1160
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.673539518900344e-06,
+      "loss": 0.1616,
+      "step": 1161
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.65979381443299e-06,
+      "loss": 0.1684,
+      "step": 1162
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.646048109965636e-06,
+      "loss": 0.1927,
+      "step": 1163
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.632302405498282e-06,
+      "loss": 0.1735,
+      "step": 1164
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.618556701030928e-06,
+      "loss": 0.1735,
+      "step": 1165
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.604810996563574e-06,
+      "loss": 0.1724,
+      "step": 1166
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.59106529209622e-06,
+      "loss": 0.1614,
+      "step": 1167
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.577319587628866e-06,
+      "loss": 0.1541,
+      "step": 1168
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.563573883161512e-06,
+      "loss": 0.1704,
+      "step": 1169
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.549828178694158e-06,
+      "loss": 0.1704,
+      "step": 1170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.536082474226804e-06,
+      "loss": 0.16,
+      "step": 1171
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.52233676975945e-06,
+      "loss": 0.1652,
+      "step": 1172
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.508591065292096e-06,
+      "loss": 0.1763,
+      "step": 1173
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.494845360824742e-06,
+      "loss": 0.1625,
+      "step": 1174
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.481099656357389e-06,
+      "loss": 0.1662,
+      "step": 1175
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.467353951890035e-06,
+      "loss": 0.1578,
+      "step": 1176
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.453608247422681e-06,
+      "loss": 0.1619,
+      "step": 1177
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4398625429553264e-06,
+      "loss": 0.1645,
+      "step": 1178
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4261168384879725e-06,
+      "loss": 0.1644,
+      "step": 1179
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4123711340206185e-06,
+      "loss": 0.1653,
+      "step": 1180
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.3986254295532645e-06,
+      "loss": 0.1685,
+      "step": 1181
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.384879725085911e-06,
+      "loss": 0.1626,
+      "step": 1182
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.371134020618557e-06,
+      "loss": 0.1717,
+      "step": 1183
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.357388316151203e-06,
+      "loss": 0.1712,
+      "step": 1184
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.343642611683849e-06,
+      "loss": 0.1644,
+      "step": 1185
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.329896907216495e-06,
+      "loss": 0.1631,
+      "step": 1186
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3161512027491414e-06,
+      "loss": 0.1549,
+      "step": 1187
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.3024054982817875e-06,
+      "loss": 0.1459,
+      "step": 1188
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2886597938144335e-06,
+      "loss": 0.1533,
+      "step": 1189
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2749140893470795e-06,
+      "loss": 0.1683,
+      "step": 1190
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2611683848797255e-06,
+      "loss": 0.1662,
+      "step": 1191
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2474226804123715e-06,
+      "loss": 0.1568,
+      "step": 1192
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2336769759450175e-06,
+      "loss": 0.1643,
+      "step": 1193
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2199312714776636e-06,
+      "loss": 0.1605,
+      "step": 1194
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2061855670103096e-06,
+      "loss": 0.1659,
+      "step": 1195
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.192439862542956e-06,
+      "loss": 0.1604,
+      "step": 1196
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.178694158075602e-06,
+      "loss": 0.176,
+      "step": 1197
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.164948453608248e-06,
+      "loss": 0.1618,
+      "step": 1198
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.151202749140894e-06,
+      "loss": 0.1801,
+      "step": 1199
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.13745704467354e-06,
+      "loss": 0.1639,
+      "step": 1200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.123711340206186e-06,
+      "loss": 0.1532,
+      "step": 1201
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.109965635738832e-06,
+      "loss": 0.171,
+      "step": 1202
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.096219931271478e-06,
+      "loss": 0.1627,
+      "step": 1203
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.082474226804124e-06,
+      "loss": 0.1674,
+      "step": 1204
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.06872852233677e-06,
+      "loss": 0.176,
+      "step": 1205
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.054982817869416e-06,
+      "loss": 0.1593,
+      "step": 1206
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.041237113402063e-06,
+      "loss": 0.1529,
+      "step": 1207
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.027491408934709e-06,
+      "loss": 0.1657,
+      "step": 1208
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.013745704467355e-06,
+      "loss": 0.1673,
+      "step": 1209
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.1413,
+      "step": 1210
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.986254295532647e-06,
+      "loss": 0.1721,
+      "step": 1211
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.972508591065292e-06,
+      "loss": 0.1727,
+      "step": 1212
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.958762886597938e-06,
+      "loss": 0.1666,
+      "step": 1213
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.945017182130585e-06,
+      "loss": 0.176,
+      "step": 1214
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.931271477663231e-06,
+      "loss": 0.1777,
+      "step": 1215
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.917525773195877e-06,
+      "loss": 0.1607,
+      "step": 1216
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.903780068728523e-06,
+      "loss": 0.1809,
+      "step": 1217
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.890034364261169e-06,
+      "loss": 0.1663,
+      "step": 1218
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.876288659793815e-06,
+      "loss": 0.1697,
+      "step": 1219
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.862542955326461e-06,
+      "loss": 0.1518,
+      "step": 1220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.848797250859107e-06,
+      "loss": 0.1647,
+      "step": 1221
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.835051546391753e-06,
+      "loss": 0.1552,
+      "step": 1222
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.821305841924399e-06,
+      "loss": 0.1689,
+      "step": 1223
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.807560137457045e-06,
+      "loss": 0.1635,
+      "step": 1224
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.793814432989691e-06,
+      "loss": 0.1616,
+      "step": 1225
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.780068728522337e-06,
+      "loss": 0.1704,
+      "step": 1226
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.766323024054983e-06,
+      "loss": 0.1687,
+      "step": 1227
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.752577319587629e-06,
+      "loss": 0.1564,
+      "step": 1228
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7388316151202754e-06,
+      "loss": 0.1636,
+      "step": 1229
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7250859106529214e-06,
+      "loss": 0.1678,
+      "step": 1230
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7113402061855674e-06,
+      "loss": 0.1668,
+      "step": 1231
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6975945017182135e-06,
+      "loss": 0.1669,
+      "step": 1232
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6838487972508595e-06,
+      "loss": 0.1788,
+      "step": 1233
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.670103092783505e-06,
+      "loss": 0.1637,
+      "step": 1234
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.656357388316151e-06,
+      "loss": 0.1767,
+      "step": 1235
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6426116838487975e-06,
+      "loss": 0.1527,
+      "step": 1236
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6288659793814435e-06,
+      "loss": 0.1548,
+      "step": 1237
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6151202749140896e-06,
+      "loss": 0.1522,
+      "step": 1238
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6013745704467356e-06,
+      "loss": 0.1679,
+      "step": 1239
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5876288659793816e-06,
+      "loss": 0.1711,
+      "step": 1240
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.573883161512028e-06,
+      "loss": 0.1557,
+      "step": 1241
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.560137457044674e-06,
+      "loss": 0.1551,
+      "step": 1242
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.54639175257732e-06,
+      "loss": 0.1538,
+      "step": 1243
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.532646048109966e-06,
+      "loss": 0.1594,
+      "step": 1244
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.518900343642612e-06,
+      "loss": 0.1757,
+      "step": 1245
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5051546391752577e-06,
+      "loss": 0.1716,
+      "step": 1246
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4914089347079037e-06,
+      "loss": 0.1598,
+      "step": 1247
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.47766323024055e-06,
+      "loss": 0.1541,
+      "step": 1248
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.463917525773196e-06,
+      "loss": 0.1589,
+      "step": 1249
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.450171821305842e-06,
+      "loss": 0.15,
+      "step": 1250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.436426116838488e-06,
+      "loss": 0.159,
+      "step": 1251
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4226804123711342e-06,
+      "loss": 0.1759,
+      "step": 1252
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.4089347079037807e-06,
+      "loss": 0.1502,
+      "step": 1253
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3951890034364267e-06,
+      "loss": 0.1628,
+      "step": 1254
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3814432989690727e-06,
+      "loss": 0.1585,
+      "step": 1255
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3676975945017187e-06,
+      "loss": 0.1587,
+      "step": 1256
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3539518900343643e-06,
+      "loss": 0.1583,
+      "step": 1257
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3402061855670103e-06,
+      "loss": 0.1637,
+      "step": 1258
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3264604810996563e-06,
+      "loss": 0.1629,
+      "step": 1259
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3127147766323028e-06,
+      "loss": 0.1636,
+      "step": 1260
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.298969072164949e-06,
+      "loss": 0.1775,
+      "step": 1261
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.285223367697595e-06,
+      "loss": 0.1611,
+      "step": 1262
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.271477663230241e-06,
+      "loss": 0.1485,
+      "step": 1263
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.257731958762887e-06,
+      "loss": 0.1566,
+      "step": 1264
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.243986254295533e-06,
+      "loss": 0.166,
+      "step": 1265
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2302405498281793e-06,
+      "loss": 0.1693,
+      "step": 1266
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2164948453608253e-06,
+      "loss": 0.15,
+      "step": 1267
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.202749140893471e-06,
+      "loss": 0.1675,
+      "step": 1268
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.189003436426117e-06,
+      "loss": 0.1564,
+      "step": 1269
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.175257731958763e-06,
+      "loss": 0.148,
+      "step": 1270
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.161512027491409e-06,
+      "loss": 0.1591,
+      "step": 1271
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.147766323024055e-06,
+      "loss": 0.1683,
+      "step": 1272
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1340206185567014e-06,
+      "loss": 0.1651,
+      "step": 1273
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1202749140893474e-06,
+      "loss": 0.1708,
+      "step": 1274
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1065292096219934e-06,
+      "loss": 0.1756,
+      "step": 1275
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0927835051546395e-06,
+      "loss": 0.1655,
+      "step": 1276
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0790378006872855e-06,
+      "loss": 0.167,
+      "step": 1277
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.065292096219932e-06,
+      "loss": 0.1713,
+      "step": 1278
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.051546391752578e-06,
+      "loss": 0.1647,
+      "step": 1279
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0378006872852235e-06,
+      "loss": 0.168,
+      "step": 1280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0240549828178695e-06,
+      "loss": 0.1604,
+      "step": 1281
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0103092783505156e-06,
+      "loss": 0.1751,
+      "step": 1282
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9965635738831616e-06,
+      "loss": 0.1684,
+      "step": 1283
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9828178694158076e-06,
+      "loss": 0.1622,
+      "step": 1284
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.969072164948454e-06,
+      "loss": 0.1719,
+      "step": 1285
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9553264604811e-06,
+      "loss": 0.1537,
+      "step": 1286
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.941580756013746e-06,
+      "loss": 0.1624,
+      "step": 1287
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.927835051546392e-06,
+      "loss": 0.1663,
+      "step": 1288
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.914089347079038e-06,
+      "loss": 0.1468,
+      "step": 1289
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.900343642611684e-06,
+      "loss": 0.1677,
+      "step": 1290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8865979381443297e-06,
+      "loss": 0.1641,
+      "step": 1291
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.872852233676976e-06,
+      "loss": 0.1678,
+      "step": 1292
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.859106529209622e-06,
+      "loss": 0.1651,
+      "step": 1293
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.845360824742268e-06,
+      "loss": 0.1485,
+      "step": 1294
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.831615120274914e-06,
+      "loss": 0.1477,
+      "step": 1295
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8178694158075602e-06,
+      "loss": 0.1666,
+      "step": 1296
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8041237113402062e-06,
+      "loss": 0.167,
+      "step": 1297
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7903780068728527e-06,
+      "loss": 0.1512,
+      "step": 1298
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7766323024054987e-06,
+      "loss": 0.1468,
+      "step": 1299
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7628865979381447e-06,
+      "loss": 0.1576,
+      "step": 1300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7491408934707907e-06,
+      "loss": 0.1655,
+      "step": 1301
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7353951890034367e-06,
+      "loss": 0.152,
+      "step": 1302
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7216494845360823e-06,
+      "loss": 0.162,
+      "step": 1303
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7079037800687284e-06,
+      "loss": 0.1587,
+      "step": 1304
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.694158075601375e-06,
+      "loss": 0.1836,
+      "step": 1305
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.680412371134021e-06,
+      "loss": 0.1494,
+      "step": 1306
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.1527,
+      "step": 1307
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.652920962199313e-06,
+      "loss": 0.1624,
+      "step": 1308
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.639175257731959e-06,
+      "loss": 0.156,
+      "step": 1309
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6254295532646053e-06,
+      "loss": 0.1493,
+      "step": 1310
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6116838487972513e-06,
+      "loss": 0.1612,
+      "step": 1311
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5979381443298973e-06,
+      "loss": 0.1655,
+      "step": 1312
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5841924398625434e-06,
+      "loss": 0.1456,
+      "step": 1313
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.570446735395189e-06,
+      "loss": 0.1514,
+      "step": 1314
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.556701030927835e-06,
+      "loss": 0.1689,
+      "step": 1315
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.542955326460481e-06,
+      "loss": 0.1529,
+      "step": 1316
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5292096219931274e-06,
+      "loss": 0.1515,
+      "step": 1317
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.5154639175257734e-06,
+      "loss": 0.1448,
+      "step": 1318
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.5017182130584195e-06,
+      "loss": 0.1599,
+      "step": 1319
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4879725085910655e-06,
+      "loss": 0.1614,
+      "step": 1320
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4742268041237115e-06,
+      "loss": 0.1489,
+      "step": 1321
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4604810996563575e-06,
+      "loss": 0.1611,
+      "step": 1322
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4467353951890035e-06,
+      "loss": 0.1605,
+      "step": 1323
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4329896907216495e-06,
+      "loss": 0.1544,
+      "step": 1324
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4192439862542956e-06,
+      "loss": 0.1569,
+      "step": 1325
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.405498281786942e-06,
+      "loss": 0.1522,
+      "step": 1326
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.391752577319588e-06,
+      "loss": 0.1587,
+      "step": 1327
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3780068728522336e-06,
+      "loss": 0.1714,
+      "step": 1328
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.36426116838488e-06,
+      "loss": 0.1546,
+      "step": 1329
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.350515463917526e-06,
+      "loss": 0.1702,
+      "step": 1330
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.336769759450172e-06,
+      "loss": 0.1482,
+      "step": 1331
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.323024054982818e-06,
+      "loss": 0.1676,
+      "step": 1332
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.309278350515464e-06,
+      "loss": 0.1603,
+      "step": 1333
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.29553264604811e-06,
+      "loss": 0.1581,
+      "step": 1334
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.281786941580756e-06,
+      "loss": 0.1544,
+      "step": 1335
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.268041237113402e-06,
+      "loss": 0.1636,
+      "step": 1336
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.254295532646048e-06,
+      "loss": 0.1534,
+      "step": 1337
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2405498281786946e-06,
+      "loss": 0.1447,
+      "step": 1338
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2268041237113406e-06,
+      "loss": 0.1392,
+      "step": 1339
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2130584192439862e-06,
+      "loss": 0.1566,
+      "step": 1340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1993127147766322e-06,
+      "loss": 0.1516,
+      "step": 1341
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1855670103092787e-06,
+      "loss": 0.1535,
+      "step": 1342
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1718213058419247e-06,
+      "loss": 0.1583,
+      "step": 1343
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1580756013745707e-06,
+      "loss": 0.153,
+      "step": 1344
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1443298969072167e-06,
+      "loss": 0.1468,
+      "step": 1345
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1305841924398628e-06,
+      "loss": 0.1608,
+      "step": 1346
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1168384879725088e-06,
+      "loss": 0.1535,
+      "step": 1347
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1030927835051548e-06,
+      "loss": 0.1649,
+      "step": 1348
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.089347079037801e-06,
+      "loss": 0.1741,
+      "step": 1349
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.075601374570447e-06,
+      "loss": 0.1597,
+      "step": 1350
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.061855670103093e-06,
+      "loss": 0.1604,
+      "step": 1351
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.048109965635739e-06,
+      "loss": 0.1652,
+      "step": 1352
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.034364261168385e-06,
+      "loss": 0.1514,
+      "step": 1353
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0206185567010313e-06,
+      "loss": 0.1459,
+      "step": 1354
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0068728522336773e-06,
+      "loss": 0.1452,
+      "step": 1355
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9931271477663233e-06,
+      "loss": 0.1631,
+      "step": 1356
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.979381443298969e-06,
+      "loss": 0.1591,
+      "step": 1357
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9656357388316154e-06,
+      "loss": 0.1591,
+      "step": 1358
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9518900343642614e-06,
+      "loss": 0.1499,
+      "step": 1359
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9381443298969074e-06,
+      "loss": 0.1686,
+      "step": 1360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9243986254295534e-06,
+      "loss": 0.1714,
+      "step": 1361
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9106529209621994e-06,
+      "loss": 0.1471,
+      "step": 1362
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8969072164948455e-06,
+      "loss": 0.1549,
+      "step": 1363
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8831615120274915e-06,
+      "loss": 0.1533,
+      "step": 1364
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8694158075601377e-06,
+      "loss": 0.1592,
+      "step": 1365
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8556701030927837e-06,
+      "loss": 0.1611,
+      "step": 1366
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8419243986254297e-06,
+      "loss": 0.1574,
+      "step": 1367
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8281786941580755e-06,
+      "loss": 0.1412,
+      "step": 1368
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8144329896907218e-06,
+      "loss": 0.147,
+      "step": 1369
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8006872852233678e-06,
+      "loss": 0.1512,
+      "step": 1370
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.786941580756014e-06,
+      "loss": 0.1648,
+      "step": 1371
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.77319587628866e-06,
+      "loss": 0.1843,
+      "step": 1372
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.759450171821306e-06,
+      "loss": 0.1396,
+      "step": 1373
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7457044673539519e-06,
+      "loss": 0.1411,
+      "step": 1374
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.731958762886598e-06,
+      "loss": 0.1517,
+      "step": 1375
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.718213058419244e-06,
+      "loss": 0.1564,
+      "step": 1376
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7044673539518903e-06,
+      "loss": 0.1476,
+      "step": 1377
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6907216494845363e-06,
+      "loss": 0.1453,
+      "step": 1378
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6769759450171821e-06,
+      "loss": 0.1664,
+      "step": 1379
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6632302405498282e-06,
+      "loss": 0.1503,
+      "step": 1380
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6494845360824744e-06,
+      "loss": 0.1572,
+      "step": 1381
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6357388316151204e-06,
+      "loss": 0.1584,
+      "step": 1382
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6219931271477664e-06,
+      "loss": 0.1564,
+      "step": 1383
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6082474226804127e-06,
+      "loss": 0.1517,
+      "step": 1384
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5945017182130585e-06,
+      "loss": 0.1608,
+      "step": 1385
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5807560137457045e-06,
+      "loss": 0.1662,
+      "step": 1386
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5670103092783507e-06,
+      "loss": 0.1663,
+      "step": 1387
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5532646048109967e-06,
+      "loss": 0.1417,
+      "step": 1388
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5395189003436427e-06,
+      "loss": 0.1598,
+      "step": 1389
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.525773195876289e-06,
+      "loss": 0.1685,
+      "step": 1390
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5120274914089348e-06,
+      "loss": 0.1355,
+      "step": 1391
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4982817869415808e-06,
+      "loss": 0.1602,
+      "step": 1392
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.484536082474227e-06,
+      "loss": 0.1486,
+      "step": 1393
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.470790378006873e-06,
+      "loss": 0.1544,
+      "step": 1394
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.457044673539519e-06,
+      "loss": 0.1461,
+      "step": 1395
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4432989690721649e-06,
+      "loss": 0.1489,
+      "step": 1396
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.429553264604811e-06,
+      "loss": 0.1279,
+      "step": 1397
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.415807560137457e-06,
+      "loss": 0.1509,
+      "step": 1398
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4020618556701031e-06,
+      "loss": 0.1536,
+      "step": 1399
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3883161512027493e-06,
+      "loss": 0.1526,
+      "step": 1400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3745704467353954e-06,
+      "loss": 0.1628,
+      "step": 1401
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3608247422680412e-06,
+      "loss": 0.1611,
+      "step": 1402
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3470790378006874e-06,
+      "loss": 0.1717,
+      "step": 1403
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.1494,
+      "step": 1404
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3195876288659794e-06,
+      "loss": 0.1339,
+      "step": 1405
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3058419243986257e-06,
+      "loss": 0.1458,
+      "step": 1406
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2920962199312717e-06,
+      "loss": 0.1493,
+      "step": 1407
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2783505154639175e-06,
+      "loss": 0.1509,
+      "step": 1408
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2646048109965637e-06,
+      "loss": 0.143,
+      "step": 1409
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2508591065292097e-06,
+      "loss": 0.1559,
+      "step": 1410
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2371134020618557e-06,
+      "loss": 0.1549,
+      "step": 1411
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2233676975945018e-06,
+      "loss": 0.1416,
+      "step": 1412
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.2096219931271478e-06,
+      "loss": 0.1529,
+      "step": 1413
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.195876288659794e-06,
+      "loss": 0.1553,
+      "step": 1414
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.18213058419244e-06,
+      "loss": 0.1568,
+      "step": 1415
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.168384879725086e-06,
+      "loss": 0.1689,
+      "step": 1416
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.154639175257732e-06,
+      "loss": 0.1571,
+      "step": 1417
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.140893470790378e-06,
+      "loss": 0.1427,
+      "step": 1418
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.127147766323024e-06,
+      "loss": 0.1425,
+      "step": 1419
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1134020618556703e-06,
+      "loss": 0.1558,
+      "step": 1420
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0996563573883161e-06,
+      "loss": 0.1503,
+      "step": 1421
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0859106529209624e-06,
+      "loss": 0.1451,
+      "step": 1422
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0721649484536084e-06,
+      "loss": 0.1552,
+      "step": 1423
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0584192439862544e-06,
+      "loss": 0.146,
+      "step": 1424
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0446735395189004e-06,
+      "loss": 0.1551,
+      "step": 1425
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0309278350515464e-06,
+      "loss": 0.1583,
+      "step": 1426
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0171821305841924e-06,
+      "loss": 0.1513,
+      "step": 1427
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0034364261168387e-06,
+      "loss": 0.1405,
+      "step": 1428
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.896907216494845e-07,
+      "loss": 0.1577,
+      "step": 1429
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.759450171821307e-07,
+      "loss": 0.1626,
+      "step": 1430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.621993127147767e-07,
+      "loss": 0.146,
+      "step": 1431
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.484536082474227e-07,
+      "loss": 0.1583,
+      "step": 1432
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.347079037800689e-07,
+      "loss": 0.1474,
+      "step": 1433
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.209621993127149e-07,
+      "loss": 0.1641,
+      "step": 1434
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.072164948453609e-07,
+      "loss": 0.1553,
+      "step": 1435
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.93470790378007e-07,
+      "loss": 0.1603,
+      "step": 1436
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.79725085910653e-07,
+      "loss": 0.1718,
+      "step": 1437
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.65979381443299e-07,
+      "loss": 0.1409,
+      "step": 1438
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.522336769759452e-07,
+      "loss": 0.1512,
+      "step": 1439
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.384879725085911e-07,
+      "loss": 0.1615,
+      "step": 1440
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.247422680412372e-07,
+      "loss": 0.1617,
+      "step": 1441
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.109965635738832e-07,
+      "loss": 0.1599,
+      "step": 1442
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.972508591065292e-07,
+      "loss": 0.1572,
+      "step": 1443
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.835051546391754e-07,
+      "loss": 0.1476,
+      "step": 1444
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.697594501718214e-07,
+      "loss": 0.1457,
+      "step": 1445
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.560137457044674e-07,
+      "loss": 0.1476,
+      "step": 1446
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.422680412371135e-07,
+      "loss": 0.1554,
+      "step": 1447
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.285223367697595e-07,
+      "loss": 0.152,
+      "step": 1448
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.147766323024055e-07,
+      "loss": 0.1568,
+      "step": 1449
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.010309278350516e-07,
+      "loss": 0.1607,
+      "step": 1450
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.872852233676977e-07,
+      "loss": 0.1527,
+      "step": 1451
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.735395189003437e-07,
+      "loss": 0.1482,
+      "step": 1452
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.597938144329897e-07,
+      "loss": 0.1517,
+      "step": 1453
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.460481099656358e-07,
+      "loss": 0.1677,
+      "step": 1454
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.323024054982819e-07,
+      "loss": 0.1518,
+      "step": 1455
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.185567010309279e-07,
+      "loss": 0.1429,
+      "step": 1456
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.048109965635739e-07,
+      "loss": 0.1466,
+      "step": 1457
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.9106529209622e-07,
+      "loss": 0.1412,
+      "step": 1458
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.77319587628866e-07,
+      "loss": 0.1549,
+      "step": 1459
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.63573883161512e-07,
+      "loss": 0.1618,
+      "step": 1460
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.498281786941581e-07,
+      "loss": 0.1449,
+      "step": 1461
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.360824742268042e-07,
+      "loss": 0.1534,
+      "step": 1462
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.223367697594502e-07,
+      "loss": 0.1533,
+      "step": 1463
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.085910652920962e-07,
+      "loss": 0.1517,
+      "step": 1464
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.948453608247422e-07,
+      "loss": 0.1536,
+      "step": 1465
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.810996563573884e-07,
+      "loss": 0.1538,
+      "step": 1466
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.673539518900344e-07,
+      "loss": 0.1513,
+      "step": 1467
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.5360824742268044e-07,
+      "loss": 0.1558,
+      "step": 1468
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.398625429553265e-07,
+      "loss": 0.144,
+      "step": 1469
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.261168384879726e-07,
+      "loss": 0.1591,
+      "step": 1470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.123711340206186e-07,
+      "loss": 0.1543,
+      "step": 1471
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.986254295532646e-07,
+      "loss": 0.1667,
+      "step": 1472
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.848797250859107e-07,
+      "loss": 0.143,
+      "step": 1473
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.7113402061855675e-07,
+      "loss": 0.162,
+      "step": 1474
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5738831615120277e-07,
+      "loss": 0.1453,
+      "step": 1475
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4364261168384884e-07,
+      "loss": 0.1623,
+      "step": 1476
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2989690721649486e-07,
+      "loss": 0.1466,
+      "step": 1477
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.1615120274914093e-07,
+      "loss": 0.1553,
+      "step": 1478
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0240549828178694e-07,
+      "loss": 0.1407,
+      "step": 1479
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.88659793814433e-07,
+      "loss": 0.1539,
+      "step": 1480
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7491408934707903e-07,
+      "loss": 0.1599,
+      "step": 1481
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.611683848797251e-07,
+      "loss": 0.1635,
+      "step": 1482
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.474226804123711e-07,
+      "loss": 0.15,
+      "step": 1483
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.336769759450172e-07,
+      "loss": 0.1742,
+      "step": 1484
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1993127147766326e-07,
+      "loss": 0.1527,
+      "step": 1485
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.061855670103093e-07,
+      "loss": 0.1603,
+      "step": 1486
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9243986254295534e-07,
+      "loss": 0.1456,
+      "step": 1487
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7869415807560139e-07,
+      "loss": 0.163,
+      "step": 1488
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6494845360824743e-07,
+      "loss": 0.1527,
+      "step": 1489
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5120274914089347e-07,
+      "loss": 0.1487,
+      "step": 1490
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3745704467353952e-07,
+      "loss": 0.1463,
+      "step": 1491
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2371134020618556e-07,
+      "loss": 0.1443,
+      "step": 1492
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0996563573883163e-07,
+      "loss": 0.1543,
+      "step": 1493
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.621993127147767e-08,
+      "loss": 0.1418,
+      "step": 1494
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.247422680412371e-08,
+      "loss": 0.1582,
+      "step": 1495
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.872852233676976e-08,
+      "loss": 0.1607,
+      "step": 1496
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.4982817869415814e-08,
+      "loss": 0.1458,
+      "step": 1497
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.123711340206186e-08,
+      "loss": 0.1612,
+      "step": 1498
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7491408934707907e-08,
+      "loss": 0.1609,
+      "step": 1499
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3745704467353954e-08,
+      "loss": 0.1466,
+      "step": 1500
+    },
+    {
+      "epoch": 3.0,
+      "step": 1500,
+      "total_flos": 84600537415680.0,
+      "train_loss": 0.30807759308815,
+      "train_runtime": 25235.2912,
+      "train_samples_per_second": 7.608,
+      "train_steps_per_second": 0.059
+    }
+  ],
+  "max_steps": 1500,
+  "num_train_epochs": 3,
+  "total_flos": 84600537415680.0,
+  "trial_name": null,
+  "trial_params": null
+}