diff --git "a/lora-checkpoints-4/trainer_state.json" "b/lora-checkpoints-4/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/lora-checkpoints-4/trainer_state.json"
@@ -0,0 +1,12613 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 2098,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5873015873015873e-06,
+      "loss": 0.9864,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1746031746031746e-06,
+      "loss": 0.9814,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7619047619047615e-06,
+      "loss": 1.212,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.349206349206349e-06,
+      "loss": 1.2985,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.936507936507936e-06,
+      "loss": 0.9754,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 1.3065,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 1.0629,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2698412698412699e-05,
+      "loss": 1.1731,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 1.1186,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5873015873015872e-05,
+      "loss": 1.0272,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.746031746031746e-05,
+      "loss": 1.0833,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 1.0603,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0634920634920636e-05,
+      "loss": 0.904,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 1.1626,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 1.0183,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5396825396825397e-05,
+      "loss": 0.9557,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6984126984126984e-05,
+      "loss": 1.1194,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 1.2706,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0158730158730158e-05,
+      "loss": 0.9393,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1746031746031745e-05,
+      "loss": 1.2263,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.3404,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.492063492063492e-05,
+      "loss": 0.972,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.650793650793651e-05,
+      "loss": 1.2443,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 1.1523,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.968253968253968e-05,
+      "loss": 0.8782,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.126984126984127e-05,
+      "loss": 0.981,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.8137,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 1.0785,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.603174603174603e-05,
+      "loss": 0.9338,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.7816,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9206349206349204e-05,
+      "loss": 0.9182,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0793650793650794e-05,
+      "loss": 0.9208,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2380952380952384e-05,
+      "loss": 1.0022,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.396825396825397e-05,
+      "loss": 1.2013,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.555555555555556e-05,
+      "loss": 0.9994,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 0.947,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.873015873015873e-05,
+      "loss": 0.9728,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0317460317460316e-05,
+      "loss": 0.9321,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.19047619047619e-05,
+      "loss": 1.0283,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.349206349206349e-05,
+      "loss": 0.9364,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.507936507936509e-05,
+      "loss": 0.8004,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.7607,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.825396825396825e-05,
+      "loss": 0.8123,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.984126984126984e-05,
+      "loss": 0.8297,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 1.4468,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.301587301587302e-05,
+      "loss": 0.8044,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.460317460317461e-05,
+      "loss": 0.8763,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.619047619047618e-05,
+      "loss": 0.7772,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.777777777777778e-05,
+      "loss": 0.8264,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.936507936507937e-05,
+      "loss": 1.0836,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.095238095238096e-05,
+      "loss": 0.8482,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.253968253968255e-05,
+      "loss": 0.8366,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.412698412698413e-05,
+      "loss": 0.8408,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 1.3089,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.730158730158731e-05,
+      "loss": 1.0691,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 1.0904,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.047619047619048e-05,
+      "loss": 0.8917,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.206349206349206e-05,
+      "loss": 0.8136,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.365079365079366e-05,
+      "loss": 0.969,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.523809523809524e-05,
+      "loss": 0.9857,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.682539682539682e-05,
+      "loss": 0.925,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.841269841269841e-05,
+      "loss": 0.7801,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 0.9295,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999994041858121e-05,
+      "loss": 0.7702,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999976167446685e-05,
+      "loss": 0.8665,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999946376808289e-05,
+      "loss": 0.8157,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999904670013934e-05,
+      "loss": 0.8258,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999851047163016e-05,
+      "loss": 0.6479,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999785508383334e-05,
+      "loss": 0.8128,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999708053831083e-05,
+      "loss": 0.7533,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999618683690857e-05,
+      "loss": 0.8607,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999517398175646e-05,
+      "loss": 0.8066,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999404197526844e-05,
+      "loss": 0.8805,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999279082014232e-05,
+      "loss": 0.8003,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999142051935997e-05,
+      "loss": 0.7545,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998993107618713e-05,
+      "loss": 0.9053,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998832249417356e-05,
+      "loss": 0.7245,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99865947771529e-05,
+      "loss": 0.6596,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998474792924275e-05,
+      "loss": 0.7798,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998278195484463e-05,
+      "loss": 0.76,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998069685864394e-05,
+      "loss": 1.0431,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997849264561003e-05,
+      "loss": 0.8009,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997616932099607e-05,
+      "loss": 1.0364,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997372689033918e-05,
+      "loss": 0.9926,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997116535946028e-05,
+      "loss": 0.6677,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996848473446415e-05,
+      "loss": 0.8025,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99656850217394e-05,
+      "loss": 0.7313,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996276622795847e-05,
+      "loss": 0.8072,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995972836007761e-05,
+      "loss": 0.7171,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995657142533683e-05,
+      "loss": 0.9891,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995329543125992e-05,
+      "loss": 0.7316,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994990038565441e-05,
+      "loss": 0.8834,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994638629661157e-05,
+      "loss": 0.815,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994275317250638e-05,
+      "loss": 0.7481,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993900102199749e-05,
+      "loss": 0.8258,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993512985402724e-05,
+      "loss": 0.9606,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993113967782164e-05,
+      "loss": 0.8906,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99270305028903e-05,
+      "loss": 0.9602,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99228023390264e-05,
+      "loss": 0.7468,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991845519630678e-05,
+      "loss": 0.7769,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99139890850918e-05,
+      "loss": 0.8197,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990940401602533e-05,
+      "loss": 0.9714,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990470000003477e-05,
+      "loss": 1.2905,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989987704833099e-05,
+      "loss": 0.6985,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989493517240836e-05,
+      "loss": 0.9597,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988987438404459e-05,
+      "loss": 0.849,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988469469530086e-05,
+      "loss": 0.975,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98793961185217e-05,
+      "loss": 0.8148,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9873978666335e-05,
+      "loss": 1.4551,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986844235165189e-05,
+      "loss": 0.7676,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986278718766686e-05,
+      "loss": 0.7771,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985701318785763e-05,
+      "loss": 0.7579,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985112036598509e-05,
+      "loss": 0.7409,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984510873609336e-05,
+      "loss": 0.7512,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983897831250973e-05,
+      "loss": 0.8047,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983272910984452e-05,
+      "loss": 0.8763,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982636114299122e-05,
+      "loss": 1.0189,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981987442712633e-05,
+      "loss": 0.7481,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981326897770936e-05,
+      "loss": 0.7329,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980654481048275e-05,
+      "loss": 0.9694,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979970194147197e-05,
+      "loss": 0.7846,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979274038698529e-05,
+      "loss": 0.7246,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978566016361393e-05,
+      "loss": 0.9336,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977846128823185e-05,
+      "loss": 0.7859,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97711437779958e-05,
+      "loss": 0.8002,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976370765034535e-05,
+      "loss": 1.4592,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975615292300263e-05,
+      "loss": 1.0476,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974847961397253e-05,
+      "loss": 0.7912,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974068774154251e-05,
+      "loss": 0.6402,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97327773242826e-05,
+      "loss": 0.796,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972474838104538e-05,
+      "loss": 1.0377,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971660093096584e-05,
+      "loss": 0.7482,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970833499346148e-05,
+      "loss": 0.8123,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969995058823213e-05,
+      "loss": 0.8606,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969144773526e-05,
+      "loss": 0.8501,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968282645480955e-05,
+      "loss": 0.8582,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967408676742751e-05,
+      "loss": 0.6582,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966522869394281e-05,
+      "loss": 0.7341,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965625225546652e-05,
+      "loss": 0.8274,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964715747339178e-05,
+      "loss": 0.8257,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963794436939378e-05,
+      "loss": 1.1387,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962861296542975e-05,
+      "loss": 0.7025,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96191632837388e-05,
+      "loss": 0.7909,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960959534684192e-05,
+      "loss": 0.7712,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9599909177542e-05,
+      "loss": 0.6706,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959010479892368e-05,
+      "loss": 0.7575,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958018223435326e-05,
+      "loss": 0.8344,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95701415074788e-05,
+      "loss": 0.8509,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955998264222991e-05,
+      "loss": 0.7116,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954970566281778e-05,
+      "loss": 0.8225,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953931059373511e-05,
+      "loss": 0.6891,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952879745975599e-05,
+      "loss": 0.6876,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951816628593593e-05,
+      "loss": 0.8323,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950741709761173e-05,
+      "loss": 0.6962,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949654992040149e-05,
+      "loss": 1.3401,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948556478020447e-05,
+      "loss": 0.6951,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947446170320109e-05,
+      "loss": 1.4149,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946324071585283e-05,
+      "loss": 0.7371,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945190184490216e-05,
+      "loss": 0.8923,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944044511737256e-05,
+      "loss": 0.8058,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942887056056833e-05,
+      "loss": 0.922,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941717820207461e-05,
+      "loss": 1.0589,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94053680697573e-05,
+      "loss": 1.1125,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939344019176299e-05,
+      "loss": 0.7234,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938139459651885e-05,
+      "loss": 0.6668,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936923131273263e-05,
+      "loss": 0.6537,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935695036939256e-05,
+      "loss": 0.7629,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93445517957673e-05,
+      "loss": 0.7583,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933203562140583e-05,
+      "loss": 0.9143,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931940187613738e-05,
+      "loss": 0.8237,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930665059007143e-05,
+      "loss": 0.752,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929378179359756e-05,
+      "loss": 0.8145,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928079551738543e-05,
+      "loss": 0.9256,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926769179238466e-05,
+      "loss": 0.7313,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925447064982479e-05,
+      "loss": 0.853,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92411321212152e-05,
+      "loss": 0.6711,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922767623834503e-05,
+      "loss": 0.8925,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92141030332831e-05,
+      "loss": 0.854,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920041253837785e-05,
+      "loss": 0.6927,
+      "step": 179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918660478625722e-05,
+      "loss": 0.9022,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917267980982866e-05,
+      "loss": 0.85,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915863764227895e-05,
+      "loss": 0.8043,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914447831707418e-05,
+      "loss": 0.7239,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913020186795967e-05,
+      "loss": 0.6653,
+      "step": 184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911580832895985e-05,
+      "loss": 0.8466,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910129773437822e-05,
+      "loss": 0.7565,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908667011879724e-05,
+      "loss": 0.8126,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907192551707831e-05,
+      "loss": 0.8889,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905706396436157e-05,
+      "loss": 0.811,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904208549606593e-05,
+      "loss": 0.7936,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902699014788892e-05,
+      "loss": 0.8266,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901177795580663e-05,
+      "loss": 0.6377,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899644895607362e-05,
+      "loss": 0.8315,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898100318522283e-05,
+      "loss": 0.8798,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89654406800655e-05,
+      "loss": 0.9238,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894976147769109e-05,
+      "loss": 0.706,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893396561546715e-05,
+      "loss": 0.8136,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891805313103927e-05,
+      "loss": 0.7271,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8902024062331e-05,
+      "loss": 0.8482,
+      "step": 199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88858784475437e-05,
+      "loss": 0.8153,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886961632515654e-05,
+      "loss": 0.7877,
+      "step": 201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885323773392635e-05,
+      "loss": 1.0376,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883674271288748e-05,
+      "loss": 0.7293,
+      "step": 203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882013130135181e-05,
+      "loss": 0.7212,
+      "step": 204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880340353890861e-05,
+      "loss": 1.0005,
+      "step": 205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878655946542443e-05,
+      "loss": 0.8288,
+      "step": 206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876959912104302e-05,
+      "loss": 1.1313,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875252254618523e-05,
+      "loss": 0.7617,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873532978154892e-05,
+      "loss": 0.8041,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871802086810889e-05,
+      "loss": 0.908,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870059584711668e-05,
+      "loss": 0.6777,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868305476010062e-05,
+      "loss": 0.7266,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866539764886562e-05,
+      "loss": 0.7436,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864762455549312e-05,
+      "loss": 0.7493,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862973552234094e-05,
+      "loss": 0.8705,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861173059204324e-05,
+      "loss": 0.794,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85936098075104e-05,
+      "loss": 1.2424,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857537321192893e-05,
+      "loss": 0.7713,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855702084876127e-05,
+      "loss": 0.8444,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853855276174584e-05,
+      "loss": 0.7383,
+      "step": 220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851996899489684e-05,
+      "loss": 1.24,
+      "step": 221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850126959250415e-05,
+      "loss": 0.816,
+      "step": 222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848245459913324e-05,
+      "loss": 0.7151,
+      "step": 223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846352405962506e-05,
+      "loss": 1.1845,
+      "step": 224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844447801909598e-05,
+      "loss": 0.9991,
+      "step": 225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842531652293759e-05,
+      "loss": 0.9591,
+      "step": 226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840603961681663e-05,
+      "loss": 0.5932,
+      "step": 227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838664734667495e-05,
+      "loss": 1.1259,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836713975872929e-05,
+      "loss": 0.9345,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834751689947126e-05,
+      "loss": 0.7277,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832777881566714e-05,
+      "loss": 0.6633,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830792555435788e-05,
+      "loss": 1.114,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828795716285889e-05,
+      "loss": 0.836,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826787368875997e-05,
+      "loss": 0.7808,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82476751799252e-05,
+      "loss": 0.7045,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822736168449281e-05,
+      "loss": 0.747,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820693325087507e-05,
+      "loss": 0.7416,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81863899277582e-05,
+      "loss": 0.7905,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81657317641022e-05,
+      "loss": 0.8005,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814495880914078e-05,
+      "loss": 0.6512,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812407111238123e-05,
+      "loss": 0.8115,
+      "step": 241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810306872360426e-05,
+      "loss": 0.706,
+      "step": 242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808195169286401e-05,
+      "loss": 0.8306,
+      "step": 243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806072007048777e-05,
+      "loss": 1.0005,
+      "step": 244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803937390707592e-05,
+      "loss": 0.8646,
+      "step": 245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801791325350187e-05,
+      "loss": 0.8124,
+      "step": 246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799633816091184e-05,
+      "loss": 0.7156,
+      "step": 247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797464868072488e-05,
+      "loss": 0.8569,
+      "step": 248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795284486463252e-05,
+      "loss": 0.7937,
+      "step": 249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79309267645989e-05,
+      "loss": 0.7418,
+      "step": 250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790889443286044e-05,
+      "loss": 0.7027,
+      "step": 251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788674792192587e-05,
+      "loss": 0.7942,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786448728457601e-05,
+      "loss": 0.7306,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784211257386366e-05,
+      "loss": 0.7645,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781962384311353e-05,
+      "loss": 0.7658,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7797021145922e-05,
+      "loss": 0.966,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777430453615712e-05,
+      "loss": 0.7002,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775147406795841e-05,
+      "loss": 0.6624,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772852979573675e-05,
+      "loss": 1.3444,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77054717741742e-05,
+      "loss": 0.9381,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768230005822395e-05,
+      "loss": 0.8334,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765901470311016e-05,
+      "loss": 0.7214,
+      "step": 262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76356157643278e-05,
+      "loss": 0.8753,
+      "step": 263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761210329764256e-05,
+      "loss": 0.7779,
+      "step": 264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758847735909068e-05,
+      "loss": 0.7526,
+      "step": 265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756473800497884e-05,
+      "loss": 0.8678,
+      "step": 266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754088529188401e-05,
+      "loss": 0.76,
+      "step": 267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751691927665334e-05,
+      "loss": 0.8424,
+      "step": 268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749284001640399e-05,
+      "loss": 0.7711,
+      "step": 269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746864756852302e-05,
+      "loss": 0.7512,
+      "step": 270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744434199066724e-05,
+      "loss": 0.7142,
+      "step": 271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741992334076308e-05,
+      "loss": 0.9043,
+      "step": 272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739539167700649e-05,
+      "loss": 0.8009,
+      "step": 273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737074705786267e-05,
+      "loss": 0.8928,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734598954206611e-05,
+      "loss": 1.0732,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73211191886203e-05,
+      "loss": 0.5632,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72961360567977e-05,
+      "loss": 0.6959,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727104020613954e-05,
+      "loss": 1.3221,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724583169645562e-05,
+      "loss": 0.7092,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722051058782435e-05,
+      "loss": 0.6504,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71950769405924e-05,
+      "loss": 0.668,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71695308153747e-05,
+      "loss": 0.6686,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714387227305422e-05,
+      "loss": 0.6445,
+      "step": 283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711810137478183e-05,
+      "loss": 0.6916,
+      "step": 284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709221818197624e-05,
+      "loss": 0.6506,
+      "step": 285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70662227563237e-05,
+      "loss": 0.7425,
+      "step": 286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704011515977803e-05,
+      "loss": 0.712,
+      "step": 287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70138954545603e-05,
+      "loss": 0.7212,
+      "step": 288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698756370315883e-05,
+      "loss": 0.7064,
+      "step": 289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69611199683289e-05,
+      "loss": 0.7965,
+      "step": 290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693456431309276e-05,
+      "loss": 0.9281,
+      "step": 291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690789680073933e-05,
+      "loss": 0.7838,
+      "step": 292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688111749482416e-05,
+      "loss": 0.7758,
+      "step": 293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685422645916918e-05,
+      "loss": 0.8731,
+      "step": 294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682722375786266e-05,
+      "loss": 0.7409,
+      "step": 295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680010945525897e-05,
+      "loss": 0.9665,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677288361597844e-05,
+      "loss": 0.9525,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674554630490726e-05,
+      "loss": 0.7285,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671809758719723e-05,
+      "loss": 0.7294,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669053752826571e-05,
+      "loss": 0.8716,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666286619379539e-05,
+      "loss": 0.6891,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663508364973417e-05,
+      "loss": 0.7587,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660718996229498e-05,
+      "loss": 0.9461,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657918519795563e-05,
+      "loss": 0.7554,
+      "step": 304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65510694234587e-05,
+      "loss": 0.679,
+      "step": 305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652284270581126e-05,
+      "loss": 0.6841,
+      "step": 306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649450511228485e-05,
+      "loss": 0.7418,
+      "step": 307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64660567104152e-05,
+      "loss": 0.7678,
+      "step": 308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643749756800219e-05,
+      "loss": 0.7633,
+      "step": 309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640882775310958e-05,
+      "loss": 0.948,
+      "step": 310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63800473340649e-05,
+      "loss": 0.7967,
+      "step": 311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635115637945925e-05,
+      "loss": 0.8408,
+      "step": 312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632215495814723e-05,
+      "loss": 0.7278,
+      "step": 313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629304313924666e-05,
+      "loss": 0.6953,
+      "step": 314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626382099213849e-05,
+      "loss": 0.7089,
+      "step": 315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623448858646657e-05,
+      "loss": 0.5901,
+      "step": 316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620504599213758e-05,
+      "loss": 0.7492,
+      "step": 317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617549327932078e-05,
+      "loss": 0.661,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614583051844786e-05,
+      "loss": 0.7582,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61160577802128e-05,
+      "loss": 0.6707,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608617513557168e-05,
+      "loss": 0.7187,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60561826557425e-05,
+      "loss": 0.7014,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602608041220508e-05,
+      "loss": 0.6953,
+      "step": 323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599586847670077e-05,
+      "loss": 0.9609,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596554692123235e-05,
+      "loss": 1.1491,
+      "step": 325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59351158180639e-05,
+      "loss": 0.8599,
+      "step": 326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590457523972056e-05,
+      "loss": 0.6204,
+      "step": 327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587392525898833e-05,
+      "loss": 0.7407,
+      "step": 328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5843165948914e-05,
+      "loss": 0.617,
+      "step": 329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581229738280493e-05,
+      "loss": 0.9441,
+      "step": 330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578131963422882e-05,
+      "loss": 0.7137,
+      "step": 331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575023277701358e-05,
+      "loss": 0.6327,
+      "step": 332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57190368852472e-05,
+      "loss": 1.297,
+      "step": 333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56877320332775e-05,
+      "loss": 0.7465,
+      "step": 334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565631829571195e-05,
+      "loss": 0.884,
+      "step": 335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562479574741758e-05,
+      "loss": 0.7374,
+      "step": 336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55931644635207e-05,
+      "loss": 0.8372,
+      "step": 337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55614245194068e-05,
+      "loss": 0.6568,
+      "step": 338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552957599072029e-05,
+      "loss": 0.6376,
+      "step": 339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54976189533644e-05,
+      "loss": 0.7344,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546555348350097e-05,
+      "loss": 0.7372,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543337965755024e-05,
+      "loss": 0.9547,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54010975521907e-05,
+      "loss": 1.5404,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536870724435887e-05,
+      "loss": 0.5974,
+      "step": 344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533620881124921e-05,
+      "loss": 0.7945,
+      "step": 345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530360233031382e-05,
+      "loss": 0.7053,
+      "step": 346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527088787926229e-05,
+      "loss": 1.0492,
+      "step": 347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523806553606156e-05,
+      "loss": 0.979,
+      "step": 348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520513537893574e-05,
+      "loss": 0.7435,
+      "step": 349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517209748636583e-05,
+      "loss": 0.7421,
+      "step": 350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513895193708958e-05,
+      "loss": 0.9453,
+      "step": 351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510569881010136e-05,
+      "loss": 0.7089,
+      "step": 352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507233818465193e-05,
+      "loss": 1.2344,
+      "step": 353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503887014024822e-05,
+      "loss": 0.8915,
+      "step": 354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500529475665316e-05,
+      "loss": 0.7107,
+      "step": 355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497161211388552e-05,
+      "loss": 0.8573,
+      "step": 356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493782229221968e-05,
+      "loss": 0.7406,
+      "step": 357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490392537218546e-05,
+      "loss": 0.6063,
+      "step": 358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486992143456792e-05,
+      "loss": 0.7611,
+      "step": 359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483581056040719e-05,
+      "loss": 0.791,
+      "step": 360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480159283099823e-05,
+      "loss": 0.6867,
+      "step": 361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476726832789066e-05,
+      "loss": 0.7861,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473283713288862e-05,
+      "loss": 0.7511,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469829932805046e-05,
+      "loss": 0.6765,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466365499568864e-05,
+      "loss": 0.879,
+      "step": 365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462890421836951e-05,
+      "loss": 0.7466,
+      "step": 366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459404707891308e-05,
+      "loss": 0.7553,
+      "step": 367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455908366039288e-05,
+      "loss": 0.65,
+      "step": 368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452401404613569e-05,
+      "loss": 0.6588,
+      "step": 369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448883831972142e-05,
+      "loss": 0.6732,
+      "step": 370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445355656498285e-05,
+      "loss": 0.7163,
+      "step": 371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441816886600549e-05,
+      "loss": 0.6913,
+      "step": 372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438267530712727e-05,
+      "loss": 0.7606,
+      "step": 373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434707597293848e-05,
+      "loss": 0.6302,
+      "step": 374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431137094828146e-05,
+      "loss": 0.6273,
+      "step": 375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427556031825046e-05,
+      "loss": 0.8073,
+      "step": 376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42396441681914e-05,
+      "loss": 0.7282,
+      "step": 377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42036225837017e-05,
+      "loss": 0.636,
+      "step": 378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416749565063002e-05,
+      "loss": 1.0353,
+      "step": 379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413126345507614e-05,
+      "loss": 0.6264,
+      "step": 380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409492608339068e-05,
+      "loss": 0.7519,
+      "step": 381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405848362217491e-05,
+      "loss": 0.781,
+      "step": 382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402193615828059e-05,
+      "loss": 0.6213,
+      "step": 383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398528377880971e-05,
+      "loss": 0.7763,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39485265711143e-05,
+      "loss": 0.7924,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391166462279619e-05,
+      "loss": 0.8281,
+      "step": 386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387469802170693e-05,
+      "loss": 0.9522,
+      "step": 387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.383762685594737e-05,
+      "loss": 0.7059,
+      "step": 388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380045121386764e-05,
+      "loss": 0.9309,
+      "step": 389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376317118406683e-05,
+      "loss": 0.5627,
+      "step": 390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372578685539283e-05,
+      "loss": 0.9492,
+      "step": 391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368829831694209e-05,
+      "loss": 0.9037,
+      "step": 392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365070565805941e-05,
+      "loss": 0.7957,
+      "step": 393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361300896833778e-05,
+      "loss": 0.6831,
+      "step": 394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357520833761806e-05,
+      "loss": 0.8914,
+      "step": 395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353730385598887e-05,
+      "loss": 0.725,
+      "step": 396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349929561378632e-05,
+      "loss": 1.1919,
+      "step": 397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34611837015938e-05,
+      "loss": 0.8337,
+      "step": 398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342296821024181e-05,
+      "loss": 0.8971,
+      "step": 399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338464923080765e-05,
+      "loss": 0.6721,
+      "step": 400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334622685461532e-05,
+      "loss": 0.6036,
+      "step": 401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330770117323515e-05,
+      "loss": 1.011,
+      "step": 402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326907227848378e-05,
+      "loss": 0.8626,
+      "step": 403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323034026242377e-05,
+      "loss": 1.0827,
+      "step": 404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319150521736346e-05,
+      "loss": 0.8646,
+      "step": 405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315256723585673e-05,
+      "loss": 0.8135,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311352641070279e-05,
+      "loss": 0.7269,
+      "step": 407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307438283494595e-05,
+      "loss": 0.7963,
+      "step": 408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303513660187539e-05,
+      "loss": 0.7098,
+      "step": 409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299578780502498e-05,
+      "loss": 0.6408,
+      "step": 410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295633653817299e-05,
+      "loss": 0.6409,
+      "step": 411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291678289534194e-05,
+      "loss": 0.642,
+      "step": 412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287712697079827e-05,
+      "loss": 0.7966,
+      "step": 413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283736885905228e-05,
+      "loss": 0.7932,
+      "step": 414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279750865485772e-05,
+      "loss": 0.6876,
+      "step": 415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275754645321173e-05,
+      "loss": 0.7723,
+      "step": 416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271748234935447e-05,
+      "loss": 0.694,
+      "step": 417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267731643876898e-05,
+      "loss": 0.6828,
+      "step": 418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263704881718096e-05,
+      "loss": 1.126,
+      "step": 419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259667958055849e-05,
+      "loss": 0.7013,
+      "step": 420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25562088251118e-05,
+      "loss": 0.882,
+      "step": 421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25156366472931e-05,
+      "loss": 1.2019,
+      "step": 422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247496314379634e-05,
+      "loss": 0.9464,
+      "step": 423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243418841155687e-05,
+      "loss": 0.8431,
+      "step": 424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239331254775139e-05,
+      "loss": 0.746,
+      "step": 425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235233564979755e-05,
+      "loss": 0.8789,
+      "step": 426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231125781535384e-05,
+      "loss": 0.9749,
+      "step": 427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227007914231925e-05,
+      "loss": 0.6313,
+      "step": 428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222879972883318e-05,
+      "loss": 0.878,
+      "step": 429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.218741967327504e-05,
+      "loss": 0.706,
+      "step": 430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214593907426412e-05,
+      "loss": 0.8917,
+      "step": 431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210435803065937e-05,
+      "loss": 0.7374,
+      "step": 432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206267664155907e-05,
+      "loss": 0.7562,
+      "step": 433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202089500630067e-05,
+      "loss": 0.6311,
+      "step": 434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197901322446054e-05,
+      "loss": 0.7302,
+      "step": 435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193703139585373e-05,
+      "loss": 0.7587,
+      "step": 436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189494962053368e-05,
+      "loss": 0.7872,
+      "step": 437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185276799879211e-05,
+      "loss": 0.777,
+      "step": 438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181048663115863e-05,
+      "loss": 0.8436,
+      "step": 439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176810561840061e-05,
+      "loss": 0.7092,
+      "step": 440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172562506152287e-05,
+      "loss": 0.8175,
+      "step": 441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168304506176748e-05,
+      "loss": 1.1903,
+      "step": 442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164036572061354e-05,
+      "loss": 0.806,
+      "step": 443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159758713977684e-05,
+      "loss": 1.1096,
+      "step": 444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155470942120976e-05,
+      "loss": 0.749,
+      "step": 445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151173266710088e-05,
+      "loss": 0.7202,
+      "step": 446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146865697987485e-05,
+      "loss": 0.7457,
+      "step": 447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142548246219212e-05,
+      "loss": 0.5466,
+      "step": 448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13822092169486e-05,
+      "loss": 0.6082,
+      "step": 449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133883734727557e-05,
+      "loss": 0.7618,
+      "step": 450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.129536695653935e-05,
+      "loss": 0.6834,
+      "step": 451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125179814834101e-05,
+      "loss": 0.6531,
+      "step": 452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120813102651621e-05,
+      "loss": 0.6935,
+      "step": 453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116436569513494e-05,
+      "loss": 0.7582,
+      "step": 454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11205022585012e-05,
+      "loss": 0.7578,
+      "step": 455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107654082115282e-05,
+      "loss": 1.1111,
+      "step": 456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103248148786118e-05,
+      "loss": 0.8363,
+      "step": 457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.098832436363102e-05,
+      "loss": 0.6569,
+      "step": 458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09440695537001e-05,
+      "loss": 1.214,
+      "step": 459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089971716353896e-05,
+      "loss": 0.7246,
+      "step": 460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085526729885076e-05,
+      "loss": 0.7039,
+      "step": 461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081072006557094e-05,
+      "loss": 0.7472,
+      "step": 462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076607556986698e-05,
+      "loss": 0.8867,
+      "step": 463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072133391813819e-05,
+      "loss": 0.6915,
+      "step": 464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06764952170154e-05,
+      "loss": 1.0172,
+      "step": 465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063155957336076e-05,
+      "loss": 0.705,
+      "step": 466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058652709426743e-05,
+      "loss": 0.7757,
+      "step": 467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05413978870594e-05,
+      "loss": 0.6618,
+      "step": 468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049617205929113e-05,
+      "loss": 0.8362,
+      "step": 469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 0.5726,
+      "step": 470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040543097344294e-05,
+      "loss": 0.7681,
+      "step": 471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.035991593162232e-05,
+      "loss": 0.7239,
+      "step": 472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031430470175958e-05,
+      "loss": 0.7934,
+      "step": 473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026859739255797e-05,
+      "loss": 0.684,
+      "step": 474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.022279411294973e-05,
+      "loss": 0.8491,
+      "step": 475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017689497209589e-05,
+      "loss": 0.8955,
+      "step": 476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013090007938583e-05,
+      "loss": 0.7783,
+      "step": 477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008480954443721e-05,
+      "loss": 0.6185,
+      "step": 478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003862347709562e-05,
+      "loss": 0.6826,
+      "step": 479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999234198743429e-05,
+      "loss": 0.936,
+      "step": 480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994596518575392e-05,
+      "loss": 0.879,
+      "step": 481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989949318258233e-05,
+      "loss": 0.7053,
+      "step": 482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985292608867422e-05,
+      "loss": 0.7397,
+      "step": 483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980626401501093e-05,
+      "loss": 0.7135,
+      "step": 484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975950707280019e-05,
+      "loss": 0.697,
+      "step": 485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971265537347577e-05,
+      "loss": 0.5872,
+      "step": 486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96657090286973e-05,
+      "loss": 0.7644,
+      "step": 487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961866815035001e-05,
+      "loss": 0.8238,
+      "step": 488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957153285054434e-05,
+      "loss": 0.7092,
+      "step": 489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952430324161582e-05,
+      "loss": 0.9511,
+      "step": 490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947697943612478e-05,
+      "loss": 1.0517,
+      "step": 491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942956154685596e-05,
+      "loss": 0.724,
+      "step": 492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.938204968681839e-05,
+      "loss": 0.7028,
+      "step": 493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.933444396924498e-05,
+      "loss": 0.8329,
+      "step": 494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.928674450759243e-05,
+      "loss": 0.8414,
+      "step": 495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923895141554077e-05,
+      "loss": 1.0448,
+      "step": 496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919106480699323e-05,
+      "loss": 0.7895,
+      "step": 497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914308479607592e-05,
+      "loss": 0.6673,
+      "step": 498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.909501149713746e-05,
+      "loss": 0.7117,
+      "step": 499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904684502474892e-05,
+      "loss": 0.84,
+      "step": 500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899858549370333e-05,
+      "loss": 0.7953,
+      "step": 501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895023301901556e-05,
+      "loss": 0.7783,
+      "step": 502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.890178771592199e-05,
+      "loss": 1.2516,
+      "step": 503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885324969988018e-05,
+      "loss": 0.6819,
+      "step": 504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880461908656872e-05,
+      "loss": 0.7047,
+      "step": 505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87558959918868e-05,
+      "loss": 0.7853,
+      "step": 506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870708053195413e-05,
+      "loss": 0.707,
+      "step": 507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865817282311043e-05,
+      "loss": 0.7987,
+      "step": 508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860917298191535e-05,
+      "loss": 0.6466,
+      "step": 509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85600811251481e-05,
+      "loss": 0.7314,
+      "step": 510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851089736980716e-05,
+      "loss": 0.7873,
+      "step": 511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846162183311005e-05,
+      "loss": 0.6105,
+      "step": 512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841225463249304e-05,
+      "loss": 0.7682,
+      "step": 513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.836279588561083e-05,
+      "loss": 0.9181,
+      "step": 514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831324571033632e-05,
+      "loss": 0.9699,
+      "step": 515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82636042247603e-05,
+      "loss": 0.7504,
+      "step": 516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.821387154719116e-05,
+      "loss": 0.662,
+      "step": 517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816404779615465e-05,
+      "loss": 0.7567,
+      "step": 518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.811413309039358e-05,
+      "loss": 0.9182,
+      "step": 519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80641275488675e-05,
+      "loss": 0.719,
+      "step": 520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801403129075242e-05,
+      "loss": 0.7329,
+      "step": 521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.796384443544064e-05,
+      "loss": 0.6133,
+      "step": 522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791356710254026e-05,
+      "loss": 0.6413,
+      "step": 523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786319941187512e-05,
+      "loss": 0.7226,
+      "step": 524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781274148348437e-05,
+      "loss": 0.795,
+      "step": 525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776219343762218e-05,
+      "loss": 0.7786,
+      "step": 526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77115553947575e-05,
+      "loss": 0.6591,
+      "step": 527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766082747557383e-05,
+      "loss": 0.7197,
+      "step": 528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761000980096882e-05,
+      "loss": 0.8501,
+      "step": 529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755910249205401e-05,
+      "loss": 0.6619,
+      "step": 530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750810567015462e-05,
+      "loss": 0.6742,
+      "step": 531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745701945680913e-05,
+      "loss": 0.6762,
+      "step": 532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740584397376915e-05,
+      "loss": 0.5886,
+      "step": 533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735457934299895e-05,
+      "loss": 0.7911,
+      "step": 534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.730322568667535e-05,
+      "loss": 0.6279,
+      "step": 535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725178312718725e-05,
+      "loss": 0.5888,
+      "step": 536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720025178713554e-05,
+      "loss": 0.8389,
+      "step": 537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714863178933257e-05,
+      "loss": 0.602,
+      "step": 538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70969232568021e-05,
+      "loss": 0.6934,
+      "step": 539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70451263127788e-05,
+      "loss": 0.6534,
+      "step": 540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699324108070812e-05,
+      "loss": 0.737,
+      "step": 541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694126768424586e-05,
+      "loss": 0.9521,
+      "step": 542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688920624725798e-05,
+      "loss": 0.7492,
+      "step": 543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683705689382024e-05,
+      "loss": 0.7433,
+      "step": 544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678481974821797e-05,
+      "loss": 0.6906,
+      "step": 545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673249493494568e-05,
+      "loss": 0.9128,
+      "step": 546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.668008257870683e-05,
+      "loss": 0.7578,
+      "step": 547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662758280441352e-05,
+      "loss": 1.1134,
+      "step": 548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657499573718621e-05,
+      "loss": 0.6489,
+      "step": 549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.652232150235338e-05,
+      "loss": 0.6827,
+      "step": 550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646956022545124e-05,
+      "loss": 0.9065,
+      "step": 551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641671203222345e-05,
+      "loss": 0.9572,
+      "step": 552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636377704862084e-05,
+      "loss": 0.9293,
+      "step": 553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63107554008011e-05,
+      "loss": 0.681,
+      "step": 554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625764721512837e-05,
+      "loss": 0.677,
+      "step": 555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.620445261817312e-05,
+      "loss": 0.8547,
+      "step": 556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.615117173671172e-05,
+      "loss": 0.7937,
+      "step": 557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609780469772623e-05,
+      "loss": 0.6268,
+      "step": 558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604435162840396e-05,
+      "loss": 0.8382,
+      "step": 559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599081265613734e-05,
+      "loss": 0.8652,
+      "step": 560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.593718790852343e-05,
+      "loss": 0.6253,
+      "step": 561
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588347751336383e-05,
+      "loss": 0.683,
+      "step": 562
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582968159866415e-05,
+      "loss": 0.6382,
+      "step": 563
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577580029263392e-05,
+      "loss": 0.8269,
+      "step": 564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.572183372368609e-05,
+      "loss": 0.8728,
+      "step": 565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.566778202043687e-05,
+      "loss": 0.8533,
+      "step": 566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561364531170533e-05,
+      "loss": 0.7561,
+      "step": 567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.555942372651316e-05,
+      "loss": 0.6532,
+      "step": 568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550511739408428e-05,
+      "loss": 0.7159,
+      "step": 569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545072644384471e-05,
+      "loss": 0.6788,
+      "step": 570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539625100542195e-05,
+      "loss": 0.7798,
+      "step": 571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534169120864503e-05,
+      "loss": 0.7487,
+      "step": 572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528704718354391e-05,
+      "loss": 0.7727,
+      "step": 573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523231906034935e-05,
+      "loss": 0.8485,
+      "step": 574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.517750696949251e-05,
+      "loss": 0.7878,
+      "step": 575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51226110416047e-05,
+      "loss": 0.7538,
+      "step": 576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.506763140751699e-05,
+      "loss": 0.7035,
+      "step": 577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501256819825996e-05,
+      "loss": 0.7211,
+      "step": 578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495742154506339e-05,
+      "loss": 0.7228,
+      "step": 579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490219157935589e-05,
+      "loss": 0.6673,
+      "step": 580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.484687843276469e-05,
+      "loss": 0.7855,
+      "step": 581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479148223711517e-05,
+      "loss": 0.7726,
+      "step": 582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473600312443072e-05,
+      "loss": 0.7747,
+      "step": 583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46804412269323e-05,
+      "loss": 0.9048,
+      "step": 584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462479667703818e-05,
+      "loss": 0.7518,
+      "step": 585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456906960736361e-05,
+      "loss": 0.7584,
+      "step": 586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451326015072051e-05,
+      "loss": 0.8652,
+      "step": 587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.445736844011713e-05,
+      "loss": 0.693,
+      "step": 588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440139460875777e-05,
+      "loss": 0.676,
+      "step": 589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434533879004245e-05,
+      "loss": 0.7211,
+      "step": 590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.428920111756658e-05,
+      "loss": 0.6976,
+      "step": 591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423298172512065e-05,
+      "loss": 0.7326,
+      "step": 592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.417668074668987e-05,
+      "loss": 0.6965,
+      "step": 593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.412029831645398e-05,
+      "loss": 0.7691,
+      "step": 594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406383456878674e-05,
+      "loss": 0.7239,
+      "step": 595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40072896382558e-05,
+      "loss": 1.0907,
+      "step": 596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395066365962223e-05,
+      "loss": 0.5705,
+      "step": 597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.389395676784025e-05,
+      "loss": 0.8456,
+      "step": 598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.383716909805699e-05,
+      "loss": 0.7757,
+      "step": 599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.378030078561201e-05,
+      "loss": 0.693,
+      "step": 600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372335196603711e-05,
+      "loss": 0.7764,
+      "step": 601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.366632277505597e-05,
+      "loss": 0.7693,
+      "step": 602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360921334858376e-05,
+      "loss": 0.9362,
+      "step": 603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.355202382272693e-05,
+      "loss": 0.8721,
+      "step": 604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34947543337828e-05,
+      "loss": 0.6766,
+      "step": 605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343740501823925e-05,
+      "loss": 0.7539,
+      "step": 606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337997601277446e-05,
+      "loss": 0.6601,
+      "step": 607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.332246745425644e-05,
+      "loss": 0.6202,
+      "step": 608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32648794797429e-05,
+      "loss": 0.7045,
+      "step": 609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320721222648075e-05,
+      "loss": 0.644,
+      "step": 610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.314946583190586e-05,
+      "loss": 0.6984,
+      "step": 611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30916404336427e-05,
+      "loss": 0.7338,
+      "step": 612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.303373616950408e-05,
+      "loss": 0.914,
+      "step": 613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.297575317749068e-05,
+      "loss": 0.6585,
+      "step": 614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.291769159579091e-05,
+      "loss": 0.8963,
+      "step": 615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285955156278038e-05,
+      "loss": 0.7145,
+      "step": 616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280133321702175e-05,
+      "loss": 0.6484,
+      "step": 617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274303669726426e-05,
+      "loss": 0.7442,
+      "step": 618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.26846621424435e-05,
+      "loss": 0.6347,
+      "step": 619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2626209691681e-05,
+      "loss": 0.6468,
+      "step": 620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2567679484284e-05,
+      "loss": 0.604,
+      "step": 621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250907165974497e-05,
+      "loss": 0.6487,
+      "step": 622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245038635774143e-05,
+      "loss": 0.7686,
+      "step": 623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.239162371813551e-05,
+      "loss": 0.7782,
+      "step": 624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233278388097368e-05,
+      "loss": 0.7012,
+      "step": 625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.227386698648634e-05,
+      "loss": 0.6393,
+      "step": 626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.221487317508762e-05,
+      "loss": 0.6421,
+      "step": 627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.215580258737493e-05,
+      "loss": 0.5991,
+      "step": 628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209665536412859e-05,
+      "loss": 0.7817,
+      "step": 629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.203743164631166e-05,
+      "loss": 0.8408,
+      "step": 630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197813157506946e-05,
+      "loss": 0.798,
+      "step": 631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191875529172928e-05,
+      "loss": 1.1555,
+      "step": 632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185930293780004e-05,
+      "loss": 0.7304,
+      "step": 633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.1799774654972e-05,
+      "loss": 0.6785,
+      "step": 634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174017058511629e-05,
+      "loss": 1.0836,
+      "step": 635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168049087028476e-05,
+      "loss": 0.6625,
+      "step": 636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.162073565270945e-05,
+      "loss": 0.6435,
+      "step": 637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15609050748024e-05,
+      "loss": 0.729,
+      "step": 638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.150099927915526e-05,
+      "loss": 0.7903,
+      "step": 639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.144101840853892e-05,
+      "loss": 0.6815,
+      "step": 640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138096260590316e-05,
+      "loss": 0.7434,
+      "step": 641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.132083201437641e-05,
+      "loss": 0.8381,
+      "step": 642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12606267772653e-05,
+      "loss": 0.83,
+      "step": 643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120034703805438e-05,
+      "loss": 0.7178,
+      "step": 644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.113999294040569e-05,
+      "loss": 0.9256,
+      "step": 645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.107956462815861e-05,
+      "loss": 0.644,
+      "step": 646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10190622453293e-05,
+      "loss": 0.8365,
+      "step": 647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.095848593611044e-05,
+      "loss": 0.732,
+      "step": 648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089783584487099e-05,
+      "loss": 0.6557,
+      "step": 649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.083711211615562e-05,
+      "loss": 0.698,
+      "step": 650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.077631489468463e-05,
+      "loss": 0.6846,
+      "step": 651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.071544432535337e-05,
+      "loss": 0.7541,
+      "step": 652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065450055323203e-05,
+      "loss": 0.7837,
+      "step": 653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059348372356529e-05,
+      "loss": 0.6558,
+      "step": 654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.053239398177191e-05,
+      "loss": 0.919,
+      "step": 655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.047123147344444e-05,
+      "loss": 0.7884,
+      "step": 656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.040999634434883e-05,
+      "loss": 0.7046,
+      "step": 657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.034868874042412e-05,
+      "loss": 0.6682,
+      "step": 658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028730880778205e-05,
+      "loss": 0.7855,
+      "step": 659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02258566927068e-05,
+      "loss": 0.6832,
+      "step": 660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01643325416545e-05,
+      "loss": 0.8519,
+      "step": 661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010273650125304e-05,
+      "loss": 0.755,
+      "step": 662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.004106871830154e-05,
+      "loss": 0.6387,
+      "step": 663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.997932933977022e-05,
+      "loss": 0.8311,
+      "step": 664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.991751851279982e-05,
+      "loss": 0.9165,
+      "step": 665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985563638470146e-05,
+      "loss": 0.8222,
+      "step": 666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.979368310295609e-05,
+      "loss": 0.6193,
+      "step": 667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973165881521434e-05,
+      "loss": 0.7507,
+      "step": 668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966956366929597e-05,
+      "loss": 0.6921,
+      "step": 669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.960739781318967e-05,
+      "loss": 0.8417,
+      "step": 670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.954516139505262e-05,
+      "loss": 0.8091,
+      "step": 671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94828545632102e-05,
+      "loss": 0.6761,
+      "step": 672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942047746615562e-05,
+      "loss": 0.7384,
+      "step": 673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.935803025254947e-05,
+      "loss": 0.7527,
+      "step": 674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92955130712195e-05,
+      "loss": 0.6387,
+      "step": 675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92329260711602e-05,
+      "loss": 0.8265,
+      "step": 676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91702694015325e-05,
+      "loss": 0.7055,
+      "step": 677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910754321166329e-05,
+      "loss": 0.53,
+      "step": 678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90447476510452e-05,
+      "loss": 0.7138,
+      "step": 679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898188286933619e-05,
+      "loss": 0.6613,
+      "step": 680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.891894901635914e-05,
+      "loss": 0.8926,
+      "step": 681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88559462421016e-05,
+      "loss": 0.8322,
+      "step": 682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879287469671537e-05,
+      "loss": 0.7166,
+      "step": 683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872973453051611e-05,
+      "loss": 0.7396,
+      "step": 684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.866652589398307e-05,
+      "loss": 0.6558,
+      "step": 685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.860324893775864e-05,
+      "loss": 0.7986,
+      "step": 686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.853990381264807e-05,
+      "loss": 0.7084,
+      "step": 687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847649066961904e-05,
+      "loss": 0.7386,
+      "step": 688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.841300965980136e-05,
+      "loss": 0.7377,
+      "step": 689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834946093448659e-05,
+      "loss": 0.8041,
+      "step": 690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.828584464512764e-05,
+      "loss": 0.7888,
+      "step": 691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.822216094333847e-05,
+      "loss": 0.8032,
+      "step": 692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.815840998089368e-05,
+      "loss": 0.6934,
+      "step": 693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.809459190972818e-05,
+      "loss": 0.7912,
+      "step": 694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.803070688193685e-05,
+      "loss": 0.6019,
+      "step": 695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.796675504977409e-05,
+      "loss": 0.6249,
+      "step": 696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790273656565352e-05,
+      "loss": 0.7558,
+      "step": 697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783865158214768e-05,
+      "loss": 0.6832,
+      "step": 698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777450025198748e-05,
+      "loss": 0.7176,
+      "step": 699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.771028272806202e-05,
+      "loss": 0.7973,
+      "step": 700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.764599916341817e-05,
+      "loss": 0.7801,
+      "step": 701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.758164971126019e-05,
+      "loss": 0.8816,
+      "step": 702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.751723452494928e-05,
+      "loss": 0.7272,
+      "step": 703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74527537580034e-05,
+      "loss": 0.7961,
+      "step": 704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73882075640968e-05,
+      "loss": 0.8567,
+      "step": 705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.732359609705962e-05,
+      "loss": 0.6106,
+      "step": 706
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.725891951087754e-05,
+      "loss": 0.6125,
+      "step": 707
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71941779596915e-05,
+      "loss": 0.6624,
+      "step": 708
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.712937159779724e-05,
+      "loss": 0.6531,
+      "step": 709
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.706450057964496e-05,
+      "loss": 1.3041,
+      "step": 710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699956505983894e-05,
+      "loss": 0.6118,
+      "step": 711
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69345651931372e-05,
+      "loss": 0.7409,
+      "step": 712
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68695011344511e-05,
+      "loss": 0.7003,
+      "step": 713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.680437303884505e-05,
+      "loss": 0.8441,
+      "step": 714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.673918106153595e-05,
+      "loss": 0.7282,
+      "step": 715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.667392535789306e-05,
+      "loss": 0.76,
+      "step": 716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.660860608343748e-05,
+      "loss": 0.622,
+      "step": 717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654322339384178e-05,
+      "loss": 0.6948,
+      "step": 718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.647777744492974e-05,
+      "loss": 0.6038,
+      "step": 719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641226839267584e-05,
+      "loss": 0.7677,
+      "step": 720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634669639320496e-05,
+      "loss": 0.5842,
+      "step": 721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.628106160279202e-05,
+      "loss": 0.6853,
+      "step": 722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621536417786159e-05,
+      "loss": 0.6826,
+      "step": 723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.614960427498748e-05,
+      "loss": 0.8668,
+      "step": 724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608378205089244e-05,
+      "loss": 0.6256,
+      "step": 725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.601789766244768e-05,
+      "loss": 0.6625,
+      "step": 726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595195126667271e-05,
+      "loss": 0.5848,
+      "step": 727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588594302073464e-05,
+      "loss": 0.608,
+      "step": 728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.58198730819481e-05,
+      "loss": 0.6574,
+      "step": 729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.57537416077747e-05,
+      "loss": 0.8285,
+      "step": 730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568754875582277e-05,
+      "loss": 0.774,
+      "step": 731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.562129468384681e-05,
+      "loss": 0.7347,
+      "step": 732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.555497954974732e-05,
+      "loss": 1.0299,
+      "step": 733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.548860351157027e-05,
+      "loss": 0.5531,
+      "step": 734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.542216672750682e-05,
+      "loss": 0.8055,
+      "step": 735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.535566935589288e-05,
+      "loss": 0.7668,
+      "step": 736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528911155520877e-05,
+      "loss": 0.6194,
+      "step": 737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.522249348407879e-05,
+      "loss": 0.9377,
+      "step": 738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.515581530127093e-05,
+      "loss": 0.723,
+      "step": 739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50890771656964e-05,
+      "loss": 1.2398,
+      "step": 740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502227923640935e-05,
+      "loss": 0.7544,
+      "step": 741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.495542167260634e-05,
+      "loss": 0.8236,
+      "step": 742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488850463362615e-05,
+      "loss": 0.655,
+      "step": 743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482152827894927e-05,
+      "loss": 0.7255,
+      "step": 744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.475449276819753e-05,
+      "loss": 0.747,
+      "step": 745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.468739826113376e-05,
+      "loss": 0.8632,
+      "step": 746
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.462024491766142e-05,
+      "loss": 0.7063,
+      "step": 747
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.455303289782414e-05,
+      "loss": 0.6912,
+      "step": 748
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.448576236180544e-05,
+      "loss": 0.745,
+      "step": 749
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.441843346992828e-05,
+      "loss": 0.6089,
+      "step": 750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.435104638265468e-05,
+      "loss": 0.7792,
+      "step": 751
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42836012605854e-05,
+      "loss": 0.7493,
+      "step": 752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.421609826445944e-05,
+      "loss": 0.6814,
+      "step": 753
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.41485375551538e-05,
+      "loss": 0.8042,
+      "step": 754
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.408091929368302e-05,
+      "loss": 0.6681,
+      "step": 755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401324364119871e-05,
+      "loss": 0.728,
+      "step": 756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39455107589894e-05,
+      "loss": 0.5984,
+      "step": 757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.387772080847988e-05,
+      "loss": 0.7803,
+      "step": 758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380987395123103e-05,
+      "loss": 0.6874,
+      "step": 759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374197034893934e-05,
+      "loss": 0.8061,
+      "step": 760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.367401016343654e-05,
+      "loss": 1.0835,
+      "step": 761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.360599355668916e-05,
+      "loss": 0.7241,
+      "step": 762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.353792069079826e-05,
+      "loss": 1.0123,
+      "step": 763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.346979172799895e-05,
+      "loss": 0.7277,
+      "step": 764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.340160683066006e-05,
+      "loss": 0.6885,
+      "step": 765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333336616128369e-05,
+      "loss": 0.9051,
+      "step": 766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.326506988250488e-05,
+      "loss": 0.6802,
+      "step": 767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319671815709119e-05,
+      "loss": 0.6828,
+      "step": 768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312831114794235e-05,
+      "loss": 0.6499,
+      "step": 769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.305984901808981e-05,
+      "loss": 0.787,
+      "step": 770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.299133193069641e-05,
+      "loss": 1.2216,
+      "step": 771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292276004905596e-05,
+      "loss": 0.9868,
+      "step": 772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.285413353659285e-05,
+      "loss": 0.7208,
+      "step": 773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27854525568617e-05,
+      "loss": 0.6928,
+      "step": 774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27167172735469e-05,
+      "loss": 0.7955,
+      "step": 775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264792785046229e-05,
+      "loss": 0.6255,
+      "step": 776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.25790844515507e-05,
+      "loss": 0.5929,
+      "step": 777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251018724088367e-05,
+      "loss": 0.6606,
+      "step": 778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.244123638266092e-05,
+      "loss": 0.7013,
+      "step": 779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.237223204121004e-05,
+      "loss": 0.7239,
+      "step": 780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23031743809861e-05,
+      "loss": 0.6786,
+      "step": 781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223406356657124e-05,
+      "loss": 0.5829,
+      "step": 782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.216489976267426e-05,
+      "loss": 0.6691,
+      "step": 783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.209568313413028e-05,
+      "loss": 0.7932,
+      "step": 784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.202641384590029e-05,
+      "loss": 0.839,
+      "step": 785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.195709206307079e-05,
+      "loss": 0.5508,
+      "step": 786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.188771795085339e-05,
+      "loss": 0.8429,
+      "step": 787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.181829167458441e-05,
+      "loss": 0.6012,
+      "step": 788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.174881339972448e-05,
+      "loss": 0.6465,
+      "step": 789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167928329185818e-05,
+      "loss": 0.6309,
+      "step": 790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.160970151669361e-05,
+      "loss": 0.7413,
+      "step": 791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1540068240062e-05,
+      "loss": 0.6107,
+      "step": 792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.147038362791732e-05,
+      "loss": 0.6966,
+      "step": 793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14006478463359e-05,
+      "loss": 0.6138,
+      "step": 794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.133086106151603e-05,
+      "loss": 0.784,
+      "step": 795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12610234397775e-05,
+      "loss": 0.6981,
+      "step": 796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.119113514756133e-05,
+      "loss": 0.637,
+      "step": 797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112119635142923e-05,
+      "loss": 0.6219,
+      "step": 798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.105120721806333e-05,
+      "loss": 0.7648,
+      "step": 799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09811679142657e-05,
+      "loss": 0.6052,
+      "step": 800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091107860695799e-05,
+      "loss": 0.657,
+      "step": 801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0840939463181e-05,
+      "loss": 0.5916,
+      "step": 802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077075065009433e-05,
+      "loss": 0.6539,
+      "step": 803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.070051233497593e-05,
+      "loss": 0.6651,
+      "step": 804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063022468522176e-05,
+      "loss": 0.7274,
+      "step": 805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05598878683453e-05,
+      "loss": 0.7491,
+      "step": 806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.048950205197728e-05,
+      "loss": 0.8121,
+      "step": 807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.041906740386518e-05,
+      "loss": 0.6014,
+      "step": 808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.034858409187281e-05,
+      "loss": 0.9482,
+      "step": 809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.027805228398001e-05,
+      "loss": 0.645,
+      "step": 810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.020747214828221e-05,
+      "loss": 0.9933,
+      "step": 811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.013684385298997e-05,
+      "loss": 0.7181,
+      "step": 812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.006616756642866e-05,
+      "loss": 0.6261,
+      "step": 813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.999544345703804e-05,
+      "loss": 0.804,
+      "step": 814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.992467169337178e-05,
+      "loss": 0.6846,
+      "step": 815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985385244409721e-05,
+      "loss": 0.7428,
+      "step": 816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978298587799474e-05,
+      "loss": 0.7957,
+      "step": 817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.97120721639576e-05,
+      "loss": 0.6206,
+      "step": 818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.964111147099142e-05,
+      "loss": 0.7215,
+      "step": 819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.957010396821369e-05,
+      "loss": 0.6271,
+      "step": 820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.949904982485355e-05,
+      "loss": 0.5902,
+      "step": 821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.942794921025126e-05,
+      "loss": 0.7049,
+      "step": 822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.935680229385787e-05,
+      "loss": 0.5858,
+      "step": 823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.92856092452347e-05,
+      "loss": 0.7596,
+      "step": 824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921437023405311e-05,
+      "loss": 0.6633,
+      "step": 825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.914308543009392e-05,
+      "loss": 0.651,
+      "step": 826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.907175500324717e-05,
+      "loss": 0.6259,
+      "step": 827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90003791235115e-05,
+      "loss": 0.6896,
+      "step": 828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8928957960994e-05,
+      "loss": 0.6347,
+      "step": 829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885749168590967e-05,
+      "loss": 0.6312,
+      "step": 830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.878598046858094e-05,
+      "loss": 0.5652,
+      "step": 831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.871442447943743e-05,
+      "loss": 0.7961,
+      "step": 832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864282388901544e-05,
+      "loss": 0.6983,
+      "step": 833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.857117886795755e-05,
+      "loss": 1.1647,
+      "step": 834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.84994895870122e-05,
+      "loss": 0.7556,
+      "step": 835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.842775621703344e-05,
+      "loss": 0.783,
+      "step": 836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.835597892898023e-05,
+      "loss": 0.6589,
+      "step": 837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.828415789391631e-05,
+      "loss": 0.792,
+      "step": 838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.821229328300964e-05,
+      "loss": 0.672,
+      "step": 839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814038526753205e-05,
+      "loss": 0.6891,
+      "step": 840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80684340188588e-05,
+      "loss": 0.6455,
+      "step": 841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.799643970846818e-05,
+      "loss": 0.57,
+      "step": 842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.792440250794112e-05,
+      "loss": 0.7421,
+      "step": 843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.785232258896077e-05,
+      "loss": 0.7143,
+      "step": 844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778020012331208e-05,
+      "loss": 0.7264,
+      "step": 845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.770803528288139e-05,
+      "loss": 1.1817,
+      "step": 846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.763582823965606e-05,
+      "loss": 0.8491,
+      "step": 847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7563579165724e-05,
+      "loss": 0.7562,
+      "step": 848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.749128823327333e-05,
+      "loss": 0.6372,
+      "step": 849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.741895561459188e-05,
+      "loss": 0.8228,
+      "step": 850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.734658148206684e-05,
+      "loss": 0.7723,
+      "step": 851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.727416600818439e-05,
+      "loss": 0.7047,
+      "step": 852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720170936552916e-05,
+      "loss": 0.663,
+      "step": 853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.712921172678395e-05,
+      "loss": 0.6644,
+      "step": 854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.705667326472925e-05,
+      "loss": 0.8451,
+      "step": 855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.698409415224283e-05,
+      "loss": 0.8686,
+      "step": 856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.691147456229934e-05,
+      "loss": 0.8598,
+      "step": 857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.683881466796994e-05,
+      "loss": 0.7546,
+      "step": 858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.676611464242177e-05,
+      "loss": 0.7632,
+      "step": 859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669337465891771e-05,
+      "loss": 0.7343,
+      "step": 860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.662059489081576e-05,
+      "loss": 0.7782,
+      "step": 861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.654777551156885e-05,
+      "loss": 0.7053,
+      "step": 862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.647491669472421e-05,
+      "loss": 0.5633,
+      "step": 863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640201861392314e-05,
+      "loss": 0.7411,
+      "step": 864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.632908144290045e-05,
+      "loss": 1.0784,
+      "step": 865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.625610535548418e-05,
+      "loss": 0.712,
+      "step": 866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618309052559506e-05,
+      "loss": 0.7535,
+      "step": 867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.611003712724617e-05,
+      "loss": 0.7274,
+      "step": 868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.603694533454256e-05,
+      "loss": 1.1312,
+      "step": 869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596381532168067e-05,
+      "loss": 1.2743,
+      "step": 870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.589064726294816e-05,
+      "loss": 0.904,
+      "step": 871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581744133272328e-05,
+      "loss": 0.7413,
+      "step": 872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.574419770547454e-05,
+      "loss": 0.6283,
+      "step": 873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.567091655576032e-05,
+      "loss": 0.6421,
+      "step": 874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.559759805822843e-05,
+      "loss": 0.7597,
+      "step": 875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.552424238761564e-05,
+      "loss": 0.6995,
+      "step": 876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545084971874738e-05,
+      "loss": 0.6478,
+      "step": 877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.537742022653721e-05,
+      "loss": 0.7234,
+      "step": 878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.530395408598644e-05,
+      "loss": 0.823,
+      "step": 879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52304514721838e-05,
+      "loss": 0.9184,
+      "step": 880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.515691256030484e-05,
+      "loss": 0.6796,
+      "step": 881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50833375256117e-05,
+      "loss": 1.1385,
+      "step": 882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.500972654345257e-05,
+      "loss": 0.6503,
+      "step": 883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.49360797892613e-05,
+      "loss": 1.3562,
+      "step": 884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.486239743855704e-05,
+      "loss": 0.7337,
+      "step": 885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.478867966694374e-05,
+      "loss": 0.6553,
+      "step": 886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.471492665010977e-05,
+      "loss": 0.8389,
+      "step": 887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.464113856382752e-05,
+      "loss": 1.0518,
+      "step": 888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456731558395295e-05,
+      "loss": 0.7554,
+      "step": 889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.449345788642514e-05,
+      "loss": 0.7698,
+      "step": 890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.441956564726597e-05,
+      "loss": 0.6948,
+      "step": 891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.434563904257962e-05,
+      "loss": 0.6906,
+      "step": 892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.427167824855218e-05,
+      "loss": 0.8378,
+      "step": 893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41976834414512e-05,
+      "loss": 0.672,
+      "step": 894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.412365479762528e-05,
+      "loss": 0.7556,
+      "step": 895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.404959249350372e-05,
+      "loss": 0.6444,
+      "step": 896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.397549670559599e-05,
+      "loss": 0.6575,
+      "step": 897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.390136761049137e-05,
+      "loss": 0.6981,
+      "step": 898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.382720538485856e-05,
+      "loss": 0.8274,
+      "step": 899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.375301020544516e-05,
+      "loss": 0.6302,
+      "step": 900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.36787822490773e-05,
+      "loss": 0.6375,
+      "step": 901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.360452169265934e-05,
+      "loss": 0.6292,
+      "step": 902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353022871317318e-05,
+      "loss": 0.7007,
+      "step": 903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.345590348767807e-05,
+      "loss": 0.9572,
+      "step": 904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.338154619331015e-05,
+      "loss": 0.6055,
+      "step": 905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.33071570072819e-05,
+      "loss": 0.6304,
+      "step": 906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.323273610688189e-05,
+      "loss": 0.7673,
+      "step": 907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31582836694742e-05,
+      "loss": 0.6223,
+      "step": 908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.308379987249812e-05,
+      "loss": 0.5899,
+      "step": 909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.300928489346766e-05,
+      "loss": 0.8128,
+      "step": 910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.293473890997115e-05,
+      "loss": 0.6495,
+      "step": 911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.286016209967078e-05,
+      "loss": 0.6667,
+      "step": 912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.278555464030228e-05,
+      "loss": 0.6578,
+      "step": 913
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.271091670967436e-05,
+      "loss": 0.856,
+      "step": 914
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263624848566836e-05,
+      "loss": 0.6822,
+      "step": 915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.256155014623786e-05,
+      "loss": 0.729,
+      "step": 916
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.248682186940814e-05,
+      "loss": 0.6701,
+      "step": 917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.241206383327592e-05,
+      "loss": 0.6578,
+      "step": 918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.233727621600876e-05,
+      "loss": 0.6709,
+      "step": 919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.226245919584474e-05,
+      "loss": 0.7921,
+      "step": 920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218761295109208e-05,
+      "loss": 0.6655,
+      "step": 921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.211273766012857e-05,
+      "loss": 0.6593,
+      "step": 922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.203783350140123e-05,
+      "loss": 0.57,
+      "step": 923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.196290065342594e-05,
+      "loss": 1.0254,
+      "step": 924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.18879392947869e-05,
+      "loss": 0.8179,
+      "step": 925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.181294960413626e-05,
+      "loss": 0.6284,
+      "step": 926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.173793176019374e-05,
+      "loss": 0.6222,
+      "step": 927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.166288594174608e-05,
+      "loss": 0.5973,
+      "step": 928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.158781232764678e-05,
+      "loss": 0.8294,
+      "step": 929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.15127110968155e-05,
+      "loss": 0.6886,
+      "step": 930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.143758242823777e-05,
+      "loss": 0.7352,
+      "step": 931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.136242650096451e-05,
+      "loss": 0.7225,
+      "step": 932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.128724349411156e-05,
+      "loss": 0.6916,
+      "step": 933
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.121203358685935e-05,
+      "loss": 0.6152,
+      "step": 934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11367969584524e-05,
+      "loss": 0.8789,
+      "step": 935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.106153378819892e-05,
+      "loss": 0.6732,
+      "step": 936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.098624425547034e-05,
+      "loss": 0.7927,
+      "step": 937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.091092853970097e-05,
+      "loss": 0.6878,
+      "step": 938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.083558682038748e-05,
+      "loss": 0.8736,
+      "step": 939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.076021927708856e-05,
+      "loss": 0.6765,
+      "step": 940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0684826089424395e-05,
+      "loss": 0.5856,
+      "step": 941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.060940743707631e-05,
+      "loss": 0.6337,
+      "step": 942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0533963499786314e-05,
+      "loss": 0.7569,
+      "step": 943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.045849445735671e-05,
+      "loss": 0.7069,
+      "step": 944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0383000489649555e-05,
+      "loss": 0.7575,
+      "step": 945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0307481776586396e-05,
+      "loss": 0.5923,
+      "step": 946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.02319384981477e-05,
+      "loss": 1.2071,
+      "step": 947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.015637083437249e-05,
+      "loss": 0.6403,
+      "step": 948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.008077896535792e-05,
+      "loss": 0.7852,
+      "step": 949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.000516307125882e-05,
+      "loss": 0.7695,
+      "step": 950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.992952333228728e-05,
+      "loss": 0.7626,
+      "step": 951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985385992871222e-05,
+      "loss": 0.7351,
+      "step": 952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9778173040858954e-05,
+      "loss": 0.6072,
+      "step": 953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.970246284910876e-05,
+      "loss": 0.9917,
+      "step": 954
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.962672953389851e-05,
+      "loss": 0.6903,
+      "step": 955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9550973275720065e-05,
+      "loss": 0.6589,
+      "step": 956
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.947519425512009e-05,
+      "loss": 0.6848,
+      "step": 957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.939939265269945e-05,
+      "loss": 0.6723,
+      "step": 958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9323568649112784e-05,
+      "loss": 0.8012,
+      "step": 959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9247722425068196e-05,
+      "loss": 0.6694,
+      "step": 960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9171854161326704e-05,
+      "loss": 0.6843,
+      "step": 961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9095964038701856e-05,
+      "loss": 0.6726,
+      "step": 962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.902005223805931e-05,
+      "loss": 0.5939,
+      "step": 963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.894411894031635e-05,
+      "loss": 0.7266,
+      "step": 964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8868164326441546e-05,
+      "loss": 0.7922,
+      "step": 965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.879218857745425e-05,
+      "loss": 0.6347,
+      "step": 966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.871619187442415e-05,
+      "loss": 0.7527,
+      "step": 967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8640174398470926e-05,
+      "loss": 0.6767,
+      "step": 968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.856413633076374e-05,
+      "loss": 0.7656,
+      "step": 969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84880778525208e-05,
+      "loss": 0.7361,
+      "step": 970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.841199914500903e-05,
+      "loss": 0.6798,
+      "step": 971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.83359003895435e-05,
+      "loss": 0.6126,
+      "step": 972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.825978176748709e-05,
+      "loss": 0.8001,
+      "step": 973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.818364346025e-05,
+      "loss": 0.6413,
+      "step": 974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.81074856492894e-05,
+      "loss": 0.8012,
+      "step": 975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.803130851610886e-05,
+      "loss": 0.6632,
+      "step": 976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7955112242258114e-05,
+      "loss": 1.0737,
+      "step": 977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7878897009332376e-05,
+      "loss": 0.6765,
+      "step": 978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7802662998972156e-05,
+      "loss": 0.8329,
+      "step": 979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7726410392862674e-05,
+      "loss": 1.2231,
+      "step": 980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7650139372733446e-05,
+      "loss": 0.595,
+      "step": 981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.75738501203579e-05,
+      "loss": 0.6472,
+      "step": 982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.749754281755293e-05,
+      "loss": 0.7215,
+      "step": 983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.742121764617843e-05,
+      "loss": 1.0619,
+      "step": 984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.734487478813686e-05,
+      "loss": 0.6493,
+      "step": 985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.726851442537285e-05,
+      "loss": 0.597,
+      "step": 986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.719213673987277e-05,
+      "loss": 0.7099,
+      "step": 987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7115741913664264e-05,
+      "loss": 0.6555,
+      "step": 988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7039330128815785e-05,
+      "loss": 1.0621,
+      "step": 989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.696290156743625e-05,
+      "loss": 0.9931,
+      "step": 990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.688645641167456e-05,
+      "loss": 1.1268,
+      "step": 991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.680999484371911e-05,
+      "loss": 0.6801,
+      "step": 992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.673351704579747e-05,
+      "loss": 0.6368,
+      "step": 993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6657023200175874e-05,
+      "loss": 0.652,
+      "step": 994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.65805134891588e-05,
+      "loss": 0.6321,
+      "step": 995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.650398809508852e-05,
+      "loss": 0.6154,
+      "step": 996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6427447200344695e-05,
+      "loss": 0.6091,
+      "step": 997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6350890987343944e-05,
+      "loss": 0.7881,
+      "step": 998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6274319638539375e-05,
+      "loss": 0.703,
+      "step": 999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6197733336420144e-05,
+      "loss": 1.1963,
+      "step": 1000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.61211322635111e-05,
+      "loss": 0.6774,
+      "step": 1001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.604451660237228e-05,
+      "loss": 0.7365,
+      "step": 1002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.596788653559843e-05,
+      "loss": 0.6513,
+      "step": 1003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.589124224581872e-05,
+      "loss": 0.6123,
+      "step": 1004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.581458391569615e-05,
+      "loss": 0.6221,
+      "step": 1005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.57379117279272e-05,
+      "loss": 0.607,
+      "step": 1006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5661225865241394e-05,
+      "loss": 0.6681,
+      "step": 1007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.558452651040081e-05,
+      "loss": 0.6139,
+      "step": 1008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.550781384619973e-05,
+      "loss": 0.7657,
+      "step": 1009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.543108805546412e-05,
+      "loss": 0.6163,
+      "step": 1010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.535434932105121e-05,
+      "loss": 0.7208,
+      "step": 1011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5277597825849134e-05,
+      "loss": 0.7214,
+      "step": 1012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5200833752776434e-05,
+      "loss": 0.6671,
+      "step": 1013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.512405728478156e-05,
+      "loss": 0.7602,
+      "step": 1014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.504726860484256e-05,
+      "loss": 0.7617,
+      "step": 1015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.497046789596659e-05,
+      "loss": 0.6738,
+      "step": 1016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4893655341189455e-05,
+      "loss": 1.2309,
+      "step": 1017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.481683112357517e-05,
+      "loss": 0.8226,
+      "step": 1018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.473999542621561e-05,
+      "loss": 0.6644,
+      "step": 1019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.466314843222993e-05,
+      "loss": 0.8669,
+      "step": 1020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4586290324764265e-05,
+      "loss": 0.5629,
+      "step": 1021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.450942128699124e-05,
+      "loss": 0.5082,
+      "step": 1022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.443254150210946e-05,
+      "loss": 0.6491,
+      "step": 1023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4355651153343245e-05,
+      "loss": 0.6105,
+      "step": 1024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.427875042394199e-05,
+      "loss": 0.6181,
+      "step": 1025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.420183949717991e-05,
+      "loss": 0.7397,
+      "step": 1026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4124918556355475e-05,
+      "loss": 0.8715,
+      "step": 1027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.404798778479104e-05,
+      "loss": 0.6037,
+      "step": 1028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3971047365832396e-05,
+      "loss": 0.5944,
+      "step": 1029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.389409748284831e-05,
+      "loss": 0.6484,
+      "step": 1030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3817138319230076e-05,
+      "loss": 0.7079,
+      "step": 1031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.37401700583912e-05,
+      "loss": 0.7914,
+      "step": 1032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.366319288376678e-05,
+      "loss": 0.5862,
+      "step": 1033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.358620697881317e-05,
+      "loss": 0.676,
+      "step": 1034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.350921252700758e-05,
+      "loss": 0.534,
+      "step": 1035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.343220971184754e-05,
+      "loss": 0.6891,
+      "step": 1036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.335519871685052e-05,
+      "loss": 0.6951,
+      "step": 1037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3278179725553525e-05,
+      "loss": 0.7029,
+      "step": 1038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.320115292151255e-05,
+      "loss": 0.7032,
+      "step": 1039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.312411848830227e-05,
+      "loss": 0.7595,
+      "step": 1040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3047076609515514e-05,
+      "loss": 0.8021,
+      "step": 1041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.297002746876284e-05,
+      "loss": 0.633,
+      "step": 1042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.289297124967216e-05,
+      "loss": 0.7147,
+      "step": 1043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.281590813588823e-05,
+      "loss": 0.6126,
+      "step": 1044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2738838311072194e-05,
+      "loss": 0.7433,
+      "step": 1045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.266176195890128e-05,
+      "loss": 0.6929,
+      "step": 1046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.258467926306821e-05,
+      "loss": 0.6378,
+      "step": 1047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2507590407280815e-05,
+      "loss": 0.7705,
+      "step": 1048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.243049557526167e-05,
+      "loss": 0.6384,
+      "step": 1049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2353394950747533e-05,
+      "loss": 0.9568,
+      "step": 1050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.227628871748899e-05,
+      "loss": 0.6022,
+      "step": 1051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.219917705924999e-05,
+      "loss": 0.6164,
+      "step": 1052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.212206015980742e-05,
+      "loss": 0.651,
+      "step": 1053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2044938202950644e-05,
+      "loss": 0.6527,
+      "step": 1054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.196781137248109e-05,
+      "loss": 0.7365,
+      "step": 1055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1890679852211785e-05,
+      "loss": 0.9223,
+      "step": 1056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181354382596698e-05,
+      "loss": 0.6033,
+      "step": 1057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1736403477581594e-05,
+      "loss": 0.645,
+      "step": 1058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.165925899090089e-05,
+      "loss": 0.8123,
+      "step": 1059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.158211054978002e-05,
+      "loss": 0.6923,
+      "step": 1060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.150495833808348e-05,
+      "loss": 0.6832,
+      "step": 1061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.142780253968481e-05,
+      "loss": 0.6111,
+      "step": 1062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.135064333846612e-05,
+      "loss": 0.6978,
+      "step": 1063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1273480918317554e-05,
+      "loss": 0.6906,
+      "step": 1064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.119631546313701e-05,
+      "loss": 0.9422,
+      "step": 1065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.111914715682956e-05,
+      "loss": 0.7139,
+      "step": 1066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.104197618330707e-05,
+      "loss": 0.6824,
+      "step": 1067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0964802726487835e-05,
+      "loss": 0.7506,
+      "step": 1068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0887626970295975e-05,
+      "loss": 0.6294,
+      "step": 1069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.081044909866114e-05,
+      "loss": 0.6997,
+      "step": 1070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0733269295518024e-05,
+      "loss": 0.8451,
+      "step": 1071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.065608774480591e-05,
+      "loss": 0.6913,
+      "step": 1072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.057890463046825e-05,
+      "loss": 0.8621,
+      "step": 1073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.050172013645222e-05,
+      "loss": 0.561,
+      "step": 1074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.042453444670828e-05,
+      "loss": 0.5773,
+      "step": 1075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0347347745189765e-05,
+      "loss": 0.7262,
+      "step": 1076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.027016021585239e-05,
+      "loss": 0.6705,
+      "step": 1077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0192972042653844e-05,
+      "loss": 0.6298,
+      "step": 1078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0115783409553385e-05,
+      "loss": 0.8693,
+      "step": 1079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.003859450051134e-05,
+      "loss": 0.628,
+      "step": 1080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9961405499488674e-05,
+      "loss": 1.1319,
+      "step": 1081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9884216590446626e-05,
+      "loss": 0.7223,
+      "step": 1082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9807027957346154e-05,
+      "loss": 0.6702,
+      "step": 1083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.972983978414761e-05,
+      "loss": 0.6163,
+      "step": 1084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.965265225481025e-05,
+      "loss": 0.7733,
+      "step": 1085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.957546555329173e-05,
+      "loss": 0.6689,
+      "step": 1086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.949827986354779e-05,
+      "loss": 0.7302,
+      "step": 1087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9421095369531764e-05,
+      "loss": 0.5691,
+      "step": 1088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.934391225519409e-05,
+      "loss": 0.5611,
+      "step": 1089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926673070448198e-05,
+      "loss": 0.6113,
+      "step": 1090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.918955090133886e-05,
+      "loss": 0.5998,
+      "step": 1091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.911237302970404e-05,
+      "loss": 0.7048,
+      "step": 1092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9035197273512176e-05,
+      "loss": 0.9286,
+      "step": 1093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8958023816692936e-05,
+      "loss": 0.79,
+      "step": 1094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8880852843170454e-05,
+      "loss": 0.6907,
+      "step": 1095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8803684536863005e-05,
+      "loss": 0.648,
+      "step": 1096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8726519081682444e-05,
+      "loss": 0.6059,
+      "step": 1097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.864935666153388e-05,
+      "loss": 0.8507,
+      "step": 1098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.85721974603152e-05,
+      "loss": 0.8391,
+      "step": 1099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849504166191654e-05,
+      "loss": 0.6664,
+      "step": 1100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.841788945022e-05,
+      "loss": 0.9959,
+      "step": 1101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8340741009099114e-05,
+      "loss": 0.7089,
+      "step": 1102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.826359652241841e-05,
+      "loss": 0.7382,
+      "step": 1103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.818645617403303e-05,
+      "loss": 0.6908,
+      "step": 1104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8109320147788206e-05,
+      "loss": 0.6664,
+      "step": 1105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.803218862751891e-05,
+      "loss": 0.7108,
+      "step": 1106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.795506179704937e-05,
+      "loss": 0.733,
+      "step": 1107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.78779398401926e-05,
+      "loss": 0.5717,
+      "step": 1108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.780082294075002e-05,
+      "loss": 0.704,
+      "step": 1109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.772371128251102e-05,
+      "loss": 0.6757,
+      "step": 1110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.764660504925247e-05,
+      "loss": 0.7298,
+      "step": 1111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7569504424738326e-05,
+      "loss": 0.6136,
+      "step": 1112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7492409592719176e-05,
+      "loss": 0.6608,
+      "step": 1113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741532073693181e-05,
+      "loss": 0.6404,
+      "step": 1114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7338238041098734e-05,
+      "loss": 0.7029,
+      "step": 1115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.726116168892782e-05,
+      "loss": 0.638,
+      "step": 1116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.718409186411179e-05,
+      "loss": 0.6051,
+      "step": 1117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.710702875032785e-05,
+      "loss": 0.7291,
+      "step": 1118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.702997253123716e-05,
+      "loss": 0.7049,
+      "step": 1119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.695292339048449e-05,
+      "loss": 0.6115,
+      "step": 1120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6875881511697744e-05,
+      "loss": 0.6525,
+      "step": 1121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.679884707848747e-05,
+      "loss": 0.6913,
+      "step": 1122
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.672182027444649e-05,
+      "loss": 0.6888,
+      "step": 1123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.664480128314949e-05,
+      "loss": 0.6021,
+      "step": 1124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.656779028815247e-05,
+      "loss": 0.8043,
+      "step": 1125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649078747299243e-05,
+      "loss": 0.6288,
+      "step": 1126
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.641379302118683e-05,
+      "loss": 0.7194,
+      "step": 1127
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.633680711623322e-05,
+      "loss": 0.6406,
+      "step": 1128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.625982994160881e-05,
+      "loss": 0.7763,
+      "step": 1129
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.618286168076993e-05,
+      "loss": 0.5809,
+      "step": 1130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.610590251715171e-05,
+      "loss": 0.7194,
+      "step": 1131
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.602895263416762e-05,
+      "loss": 0.6155,
+      "step": 1132
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5952012215208964e-05,
+      "loss": 0.6069,
+      "step": 1133
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.587508144364452e-05,
+      "loss": 0.6053,
+      "step": 1134
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.57981605028201e-05,
+      "loss": 0.6298,
+      "step": 1135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5721249576058027e-05,
+      "loss": 0.6974,
+      "step": 1136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5644348846656774e-05,
+      "loss": 0.6251,
+      "step": 1137
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5567458497890546e-05,
+      "loss": 0.6031,
+      "step": 1138
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5490578713008775e-05,
+      "loss": 0.7847,
+      "step": 1139
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5413709675235746e-05,
+      "loss": 0.6543,
+      "step": 1140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5336851567770076e-05,
+      "loss": 0.6451,
+      "step": 1141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5260004573784396e-05,
+      "loss": 0.859,
+      "step": 1142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.518316887642484e-05,
+      "loss": 0.7363,
+      "step": 1143
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.510634465881057e-05,
+      "loss": 0.6,
+      "step": 1144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.502953210403342e-05,
+      "loss": 0.6852,
+      "step": 1145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.495273139515746e-05,
+      "loss": 0.7222,
+      "step": 1146
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4875942715218456e-05,
+      "loss": 0.9857,
+      "step": 1147
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.479916624722357e-05,
+      "loss": 0.7181,
+      "step": 1148
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4722402174150864e-05,
+      "loss": 0.6123,
+      "step": 1149
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464565067894879e-05,
+      "loss": 0.5605,
+      "step": 1150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.45689119445359e-05,
+      "loss": 1.0358,
+      "step": 1151
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.449218615380029e-05,
+      "loss": 0.8563,
+      "step": 1152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4415473489599194e-05,
+      "loss": 0.6271,
+      "step": 1153
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.433877413475862e-05,
+      "loss": 0.747,
+      "step": 1154
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.42620882720728e-05,
+      "loss": 0.7017,
+      "step": 1155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.418541608430385e-05,
+      "loss": 0.5243,
+      "step": 1156
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4108757754181286e-05,
+      "loss": 0.6496,
+      "step": 1157
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4032113464401585e-05,
+      "loss": 0.7882,
+      "step": 1158
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.395548339762774e-05,
+      "loss": 0.6044,
+      "step": 1159
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3878867736488914e-05,
+      "loss": 0.641,
+      "step": 1160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.380226666357987e-05,
+      "loss": 0.585,
+      "step": 1161
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3725680361460643e-05,
+      "loss": 0.6056,
+      "step": 1162
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.364910901265606e-05,
+      "loss": 0.5707,
+      "step": 1163
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3572552799655303e-05,
+      "loss": 0.5797,
+      "step": 1164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3496011904911496e-05,
+      "loss": 0.6403,
+      "step": 1165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3419486510841215e-05,
+      "loss": 0.6957,
+      "step": 1166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.334297679982413e-05,
+      "loss": 0.7918,
+      "step": 1167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3266482954202545e-05,
+      "loss": 0.5861,
+      "step": 1168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3190005156280904e-05,
+      "loss": 0.5979,
+      "step": 1169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3113543588325454e-05,
+      "loss": 0.956,
+      "step": 1170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.303709843256376e-05,
+      "loss": 1.1336,
+      "step": 1171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.296066987118422e-05,
+      "loss": 0.7496,
+      "step": 1172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.288425808633575e-05,
+      "loss": 0.5645,
+      "step": 1173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.280786326012723e-05,
+      "loss": 0.9008,
+      "step": 1174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.273148557462715e-05,
+      "loss": 1.0176,
+      "step": 1175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2655125211863146e-05,
+      "loss": 0.6338,
+      "step": 1176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.257878235382158e-05,
+      "loss": 0.6515,
+      "step": 1177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2502457182447066e-05,
+      "loss": 0.6753,
+      "step": 1178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2426149879642094e-05,
+      "loss": 0.6439,
+      "step": 1179
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.234986062726658e-05,
+      "loss": 0.7139,
+      "step": 1180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2273589607137344e-05,
+      "loss": 0.6303,
+      "step": 1181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.219733700102786e-05,
+      "loss": 0.7059,
+      "step": 1182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2121102990667635e-05,
+      "loss": 0.5676,
+      "step": 1183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.20448877577419e-05,
+      "loss": 0.6216,
+      "step": 1184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.196869148389114e-05,
+      "loss": 0.6652,
+      "step": 1185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.189251435071061e-05,
+      "loss": 0.621,
+      "step": 1186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1816356539750015e-05,
+      "loss": 0.668,
+      "step": 1187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.174021823251294e-05,
+      "loss": 0.6722,
+      "step": 1188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.166409961045651e-05,
+      "loss": 0.961,
+      "step": 1189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.158800085499097e-05,
+      "loss": 1.0743,
+      "step": 1190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.15119221474792e-05,
+      "loss": 0.7896,
+      "step": 1191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1435863669236266e-05,
+      "loss": 0.5907,
+      "step": 1192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.135982560152907e-05,
+      "loss": 0.7663,
+      "step": 1193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.128380812557586e-05,
+      "loss": 0.7482,
+      "step": 1194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120781142254576e-05,
+      "loss": 0.6167,
+      "step": 1195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.113183567355846e-05,
+      "loss": 0.5973,
+      "step": 1196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.105588105968365e-05,
+      "loss": 0.7227,
+      "step": 1197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0979947761940694e-05,
+      "loss": 0.5297,
+      "step": 1198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.090403596129815e-05,
+      "loss": 0.5709,
+      "step": 1199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0828145838673294e-05,
+      "loss": 0.716,
+      "step": 1200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.075227757493181e-05,
+      "loss": 0.574,
+      "step": 1201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0676431350887234e-05,
+      "loss": 0.7416,
+      "step": 1202
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0600607347300576e-05,
+      "loss": 0.7825,
+      "step": 1203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0524805744879916e-05,
+      "loss": 0.7426,
+      "step": 1204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044902672427994e-05,
+      "loss": 0.6524,
+      "step": 1205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.037327046610151e-05,
+      "loss": 0.6105,
+      "step": 1206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0297537150891235e-05,
+      "loss": 0.5776,
+      "step": 1207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.022182695914105e-05,
+      "loss": 0.5024,
+      "step": 1208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.01461400712878e-05,
+      "loss": 0.7001,
+      "step": 1209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.007047666771274e-05,
+      "loss": 0.7361,
+      "step": 1210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.99948369287412e-05,
+      "loss": 0.6708,
+      "step": 1211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991922103464209e-05,
+      "loss": 0.6037,
+      "step": 1212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.984362916562753e-05,
+      "loss": 0.7037,
+      "step": 1213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.976806150185231e-05,
+      "loss": 0.6164,
+      "step": 1214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9692518223413616e-05,
+      "loss": 0.5796,
+      "step": 1215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9616999510350464e-05,
+      "loss": 0.5915,
+      "step": 1216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.954150554264332e-05,
+      "loss": 0.7212,
+      "step": 1217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.94660365002137e-05,
+      "loss": 0.6774,
+      "step": 1218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.93905925629237e-05,
+      "loss": 0.6339,
+      "step": 1219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9315173910575616e-05,
+      "loss": 0.6919,
+      "step": 1220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.923978072291145e-05,
+      "loss": 0.7086,
+      "step": 1221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.916441317961252e-05,
+      "loss": 0.7133,
+      "step": 1222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.908907146029904e-05,
+      "loss": 0.6521,
+      "step": 1223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.901375574452968e-05,
+      "loss": 0.5724,
+      "step": 1224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89384662118011e-05,
+      "loss": 0.6262,
+      "step": 1225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.88632030415476e-05,
+      "loss": 0.6409,
+      "step": 1226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8787966413140654e-05,
+      "loss": 0.5815,
+      "step": 1227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.871275650588844e-05,
+      "loss": 0.7456,
+      "step": 1228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.863757349903551e-05,
+      "loss": 0.5571,
+      "step": 1229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.856241757176223e-05,
+      "loss": 0.8758,
+      "step": 1230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.848728890318452e-05,
+      "loss": 0.677,
+      "step": 1231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.841218767235325e-05,
+      "loss": 0.6028,
+      "step": 1232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8337114058253926e-05,
+      "loss": 0.5919,
+      "step": 1233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.826206823980627e-05,
+      "loss": 0.7729,
+      "step": 1234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8187050395863744e-05,
+      "loss": 0.6941,
+      "step": 1235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8112060705213104e-05,
+      "loss": 0.6402,
+      "step": 1236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.803709934657407e-05,
+      "loss": 0.5874,
+      "step": 1237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.796216649859878e-05,
+      "loss": 0.7545,
+      "step": 1238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.788726233987146e-05,
+      "loss": 0.9992,
+      "step": 1239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.781238704890793e-05,
+      "loss": 0.6543,
+      "step": 1240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.773754080415526e-05,
+      "loss": 0.531,
+      "step": 1241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.766272378399126e-05,
+      "loss": 0.8358,
+      "step": 1242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7587936166724095e-05,
+      "loss": 0.6225,
+      "step": 1243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.751317813059186e-05,
+      "loss": 0.7485,
+      "step": 1244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.743844985376216e-05,
+      "loss": 0.6617,
+      "step": 1245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.736375151433166e-05,
+      "loss": 0.5649,
+      "step": 1246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.728908329032567e-05,
+      "loss": 0.6091,
+      "step": 1247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7214445359697735e-05,
+      "loss": 0.7876,
+      "step": 1248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.713983790032923e-05,
+      "loss": 0.6818,
+      "step": 1249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.706526109002887e-05,
+      "loss": 0.7346,
+      "step": 1250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6990715106532356e-05,
+      "loss": 0.5383,
+      "step": 1251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.691620012750188e-05,
+      "loss": 0.6747,
+      "step": 1252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.684171633052581e-05,
+      "loss": 0.6592,
+      "step": 1253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.676726389311812e-05,
+      "loss": 0.6214,
+      "step": 1254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6692842992718105e-05,
+      "loss": 0.6796,
+      "step": 1255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.661845380668986e-05,
+      "loss": 0.6326,
+      "step": 1256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.654409651232193e-05,
+      "loss": 0.5998,
+      "step": 1257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.646977128682684e-05,
+      "loss": 0.5962,
+      "step": 1258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.639547830734068e-05,
+      "loss": 0.613,
+      "step": 1259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.632121775092271e-05,
+      "loss": 1.0985,
+      "step": 1260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.624698979455486e-05,
+      "loss": 0.6819,
+      "step": 1261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6172794615141446e-05,
+      "loss": 0.5972,
+      "step": 1262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.609863238950864e-05,
+      "loss": 0.7665,
+      "step": 1263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6024503294404016e-05,
+      "loss": 0.7558,
+      "step": 1264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.595040750649629e-05,
+      "loss": 0.8987,
+      "step": 1265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.587634520237472e-05,
+      "loss": 0.6237,
+      "step": 1266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.580231655854881e-05,
+      "loss": 0.6238,
+      "step": 1267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.572832175144783e-05,
+      "loss": 0.6809,
+      "step": 1268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.565436095742039e-05,
+      "loss": 0.6978,
+      "step": 1269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.558043435273403e-05,
+      "loss": 0.6748,
+      "step": 1270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.550654211357487e-05,
+      "loss": 0.6855,
+      "step": 1271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5432684416047066e-05,
+      "loss": 0.6783,
+      "step": 1272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5358861436172485e-05,
+      "loss": 0.6285,
+      "step": 1273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.528507334989023e-05,
+      "loss": 0.6714,
+      "step": 1274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.521132033305628e-05,
+      "loss": 0.6406,
+      "step": 1275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5137602561442974e-05,
+      "loss": 0.7453,
+      "step": 1276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.506392021073871e-05,
+      "loss": 0.707,
+      "step": 1277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.499027345654745e-05,
+      "loss": 0.5974,
+      "step": 1278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.491666247438831e-05,
+      "loss": 0.6605,
+      "step": 1279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.484308743969516e-05,
+      "loss": 0.6947,
+      "step": 1280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.47695485278162e-05,
+      "loss": 0.7318,
+      "step": 1281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4696045914013566e-05,
+      "loss": 0.6801,
+      "step": 1282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4622579773462815e-05,
+      "loss": 0.7998,
+      "step": 1283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4549150281252636e-05,
+      "loss": 0.5796,
+      "step": 1284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.447575761238437e-05,
+      "loss": 0.5459,
+      "step": 1285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.440240194177158e-05,
+      "loss": 0.7937,
+      "step": 1286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.432908344423969e-05,
+      "loss": 0.662,
+      "step": 1287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.425580229452546e-05,
+      "loss": 0.6909,
+      "step": 1288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.418255866727672e-05,
+      "loss": 0.7096,
+      "step": 1289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4109352737051836e-05,
+      "loss": 0.6347,
+      "step": 1290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4036184678319326e-05,
+      "loss": 0.5264,
+      "step": 1291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.396305466545745e-05,
+      "loss": 0.6365,
+      "step": 1292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.388996287275383e-05,
+      "loss": 0.57,
+      "step": 1293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.381690947440495e-05,
+      "loss": 0.7997,
+      "step": 1294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.374389464451583e-05,
+      "loss": 0.8229,
+      "step": 1295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.367091855709955e-05,
+      "loss": 0.7417,
+      "step": 1296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3597981386076886e-05,
+      "loss": 0.6884,
+      "step": 1297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3525083305275806e-05,
+      "loss": 0.5164,
+      "step": 1298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.345222448843117e-05,
+      "loss": 0.5587,
+      "step": 1299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3379405109184236e-05,
+      "loss": 1.0155,
+      "step": 1300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.330662534108231e-05,
+      "loss": 0.6199,
+      "step": 1301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3233885357578224e-05,
+      "loss": 0.6606,
+      "step": 1302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.316118533203007e-05,
+      "loss": 0.5631,
+      "step": 1303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3088525437700666e-05,
+      "loss": 0.7189,
+      "step": 1304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.301590584775719e-05,
+      "loss": 0.5739,
+      "step": 1305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.294332673527076e-05,
+      "loss": 0.8186,
+      "step": 1306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.287078827321606e-05,
+      "loss": 0.7169,
+      "step": 1307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.279829063447084e-05,
+      "loss": 0.7423,
+      "step": 1308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.272583399181562e-05,
+      "loss": 0.646,
+      "step": 1309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.265341851793315e-05,
+      "loss": 0.5753,
+      "step": 1310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.258104438540812e-05,
+      "loss": 0.8604,
+      "step": 1311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.250871176672669e-05,
+      "loss": 0.6508,
+      "step": 1312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.243642083427601e-05,
+      "loss": 0.6253,
+      "step": 1313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2364171760343955e-05,
+      "loss": 0.5912,
+      "step": 1314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.229196471711863e-05,
+      "loss": 0.5855,
+      "step": 1315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.221979987668794e-05,
+      "loss": 0.7018,
+      "step": 1316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.214767741103923e-05,
+      "loss": 0.6994,
+      "step": 1317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.207559749205888e-05,
+      "loss": 0.6221,
+      "step": 1318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2003560291531844e-05,
+      "loss": 0.8769,
+      "step": 1319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.193156598114121e-05,
+      "loss": 0.735,
+      "step": 1320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1859614732467954e-05,
+      "loss": 0.7797,
+      "step": 1321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.178770671699036e-05,
+      "loss": 0.6759,
+      "step": 1322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.17158421060837e-05,
+      "loss": 0.5912,
+      "step": 1323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1644021071019784e-05,
+      "loss": 0.782,
+      "step": 1324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1572243782966574e-05,
+      "loss": 0.6351,
+      "step": 1325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.15005104129878e-05,
+      "loss": 0.7267,
+      "step": 1326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1428821132042486e-05,
+      "loss": 0.7189,
+      "step": 1327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.135717611098458e-05,
+      "loss": 0.554,
+      "step": 1328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.128557552056258e-05,
+      "loss": 0.6461,
+      "step": 1329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.121401953141907e-05,
+      "loss": 0.6166,
+      "step": 1330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.114250831409034e-05,
+      "loss": 0.6128,
+      "step": 1331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.107104203900599e-05,
+      "loss": 0.7996,
+      "step": 1332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.09996208764885e-05,
+      "loss": 0.5205,
+      "step": 1333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.092824499675286e-05,
+      "loss": 0.7154,
+      "step": 1334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.085691456990608e-05,
+      "loss": 0.9339,
+      "step": 1335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.078562976594689e-05,
+      "loss": 0.664,
+      "step": 1336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.07143907547653e-05,
+      "loss": 0.6776,
+      "step": 1337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.064319770614213e-05,
+      "loss": 0.6527,
+      "step": 1338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.057205078974873e-05,
+      "loss": 0.6532,
+      "step": 1339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.050095017514646e-05,
+      "loss": 0.6865,
+      "step": 1340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.042989603178634e-05,
+      "loss": 0.5539,
+      "step": 1341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0358888529008612e-05,
+      "loss": 0.9613,
+      "step": 1342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.028792783604242e-05,
+      "loss": 0.7087,
+      "step": 1343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.021701412200528e-05,
+      "loss": 0.6011,
+      "step": 1344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0146147555902805e-05,
+      "loss": 0.7197,
+      "step": 1345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.007532830662822e-05,
+      "loss": 0.585,
+      "step": 1346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.000455654296196e-05,
+      "loss": 0.6527,
+      "step": 1347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.993383243357134e-05,
+      "loss": 0.5603,
+      "step": 1348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9863156147010048e-05,
+      "loss": 0.7321,
+      "step": 1349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9792527851717805e-05,
+      "loss": 0.5674,
+      "step": 1350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9721947716020006e-05,
+      "loss": 0.6385,
+      "step": 1351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9651415908127207e-05,
+      "loss": 0.6112,
+      "step": 1352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.958093259613483e-05,
+      "loss": 0.7384,
+      "step": 1353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9510497948022714e-05,
+      "loss": 0.7583,
+      "step": 1354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.944011213165469e-05,
+      "loss": 0.7434,
+      "step": 1355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.936977531477826e-05,
+      "loss": 0.6374,
+      "step": 1356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9299487665024084e-05,
+      "loss": 0.6243,
+      "step": 1357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9229249349905684e-05,
+      "loss": 0.8422,
+      "step": 1358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9159060536819017e-05,
+      "loss": 0.7176,
+      "step": 1359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9088921393042023e-05,
+      "loss": 0.5448,
+      "step": 1360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9018832085734298e-05,
+      "loss": 0.5818,
+      "step": 1361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.894879278193666e-05,
+      "loss": 0.6566,
+      "step": 1362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8878803648570773e-05,
+      "loss": 0.7781,
+      "step": 1363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.88088648524387e-05,
+      "loss": 0.698,
+      "step": 1364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8738976560222507e-05,
+      "loss": 0.7794,
+      "step": 1365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8669138938483985e-05,
+      "loss": 0.578,
+      "step": 1366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8599352153664094e-05,
+      "loss": 0.6616,
+      "step": 1367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.852961637208268e-05,
+      "loss": 0.7231,
+      "step": 1368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8459931759938018e-05,
+      "loss": 0.7106,
+      "step": 1369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.839029848330642e-05,
+      "loss": 0.6069,
+      "step": 1370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8320716708141836e-05,
+      "loss": 0.6535,
+      "step": 1371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8251186600275533e-05,
+      "loss": 0.6005,
+      "step": 1372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.81817083254156e-05,
+      "loss": 0.6321,
+      "step": 1373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.811228204914661e-05,
+      "loss": 0.7101,
+      "step": 1374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.804290793692922e-05,
+      "loss": 0.7236,
+      "step": 1375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7973586154099712e-05,
+      "loss": 1.0866,
+      "step": 1376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.790431686586974e-05,
+      "loss": 0.641,
+      "step": 1377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.783510023732575e-05,
+      "loss": 0.63,
+      "step": 1378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7765936433428775e-05,
+      "loss": 0.6777,
+      "step": 1379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7696825619013904e-05,
+      "loss": 0.5729,
+      "step": 1380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.762776795878998e-05,
+      "loss": 0.5604,
+      "step": 1381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7558763617339096e-05,
+      "loss": 0.8026,
+      "step": 1382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.748981275911633e-05,
+      "loss": 0.5487,
+      "step": 1383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7420915548449288e-05,
+      "loss": 0.618,
+      "step": 1384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7352072149537733e-05,
+      "loss": 0.7623,
+      "step": 1385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.728328272645313e-05,
+      "loss": 0.6752,
+      "step": 1386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7214547443138328e-05,
+      "loss": 0.7725,
+      "step": 1387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.714586646340716e-05,
+      "loss": 0.5293,
+      "step": 1388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.707723995094405e-05,
+      "loss": 0.4898,
+      "step": 1389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.700866806930359e-05,
+      "loss": 0.7433,
+      "step": 1390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.69401509819102e-05,
+      "loss": 0.577,
+      "step": 1391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6871688852057675e-05,
+      "loss": 0.6277,
+      "step": 1392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6803281842908825e-05,
+      "loss": 0.756,
+      "step": 1393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6734930117495134e-05,
+      "loss": 0.5483,
+      "step": 1394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6666633838716314e-05,
+      "loss": 0.4967,
+      "step": 1395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.659839316933994e-05,
+      "loss": 0.6004,
+      "step": 1396
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6530208272001057e-05,
+      "loss": 0.5558,
+      "step": 1397
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.646207930920175e-05,
+      "loss": 0.5571,
+      "step": 1398
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.639400644331087e-05,
+      "loss": 0.8047,
+      "step": 1399
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6325989836563486e-05,
+      "loss": 0.743,
+      "step": 1400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.625802965106066e-05,
+      "loss": 0.555,
+      "step": 1401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6190126048768972e-05,
+      "loss": 0.6282,
+      "step": 1402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6122279191520138e-05,
+      "loss": 0.6031,
+      "step": 1403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.605448924101062e-05,
+      "loss": 0.6475,
+      "step": 1404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.598675635880129e-05,
+      "loss": 0.6843,
+      "step": 1405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.591908070631699e-05,
+      "loss": 0.5643,
+      "step": 1406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5851462444846204e-05,
+      "loss": 0.6771,
+      "step": 1407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5783901735540584e-05,
+      "loss": 0.8221,
+      "step": 1408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5716398739414627e-05,
+      "loss": 0.6555,
+      "step": 1409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.564895361734533e-05,
+      "loss": 0.7617,
+      "step": 1410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5581566530071734e-05,
+      "loss": 0.7073,
+      "step": 1411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.551423763819456e-05,
+      "loss": 0.7483,
+      "step": 1412
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.544696710217588e-05,
+      "loss": 0.6037,
+      "step": 1413
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.537975508233862e-05,
+      "loss": 0.6486,
+      "step": 1414
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.531260173886626e-05,
+      "loss": 0.6772,
+      "step": 1415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.524550723180249e-05,
+      "loss": 0.796,
+      "step": 1416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.517847172105074e-05,
+      "loss": 0.6188,
+      "step": 1417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5111495366373843e-05,
+      "loss": 0.7439,
+      "step": 1418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5044578327393674e-05,
+      "loss": 0.584,
+      "step": 1419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4977720763590672e-05,
+      "loss": 0.783,
+      "step": 1420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4910922834303617e-05,
+      "loss": 0.6888,
+      "step": 1421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.484418469872909e-05,
+      "loss": 0.597,
+      "step": 1422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4777506515921223e-05,
+      "loss": 0.6398,
+      "step": 1423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4710888444791236e-05,
+      "loss": 0.7332,
+      "step": 1424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.464433064410713e-05,
+      "loss": 0.7637,
+      "step": 1425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4577833272493183e-05,
+      "loss": 0.5816,
+      "step": 1426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4511396488429727e-05,
+      "loss": 0.5852,
+      "step": 1427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.444502045025268e-05,
+      "loss": 0.678,
+      "step": 1428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.43787053161532e-05,
+      "loss": 0.6813,
+      "step": 1429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.431245124417724e-05,
+      "loss": 0.6723,
+      "step": 1430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.42462583922253e-05,
+      "loss": 0.5834,
+      "step": 1431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.418012691805191e-05,
+      "loss": 0.5244,
+      "step": 1432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4114056979265364e-05,
+      "loss": 0.6258,
+      "step": 1433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4048048733327294e-05,
+      "loss": 0.822,
+      "step": 1434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.398210233755231e-05,
+      "loss": 0.7021,
+      "step": 1435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3916217949107593e-05,
+      "loss": 0.7506,
+      "step": 1436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3850395725012543e-05,
+      "loss": 0.6313,
+      "step": 1437
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3784635822138424e-05,
+      "loss": 0.6193,
+      "step": 1438
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.371893839720798e-05,
+      "loss": 0.507,
+      "step": 1439
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3653303606795034e-05,
+      "loss": 0.7268,
+      "step": 1440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.358773160732417e-05,
+      "loss": 0.618,
+      "step": 1441
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3522222555070262e-05,
+      "loss": 0.7022,
+      "step": 1442
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3456776606158227e-05,
+      "loss": 0.5715,
+      "step": 1443
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3391393916562536e-05,
+      "loss": 0.5709,
+      "step": 1444
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.332607464210695e-05,
+      "loss": 0.8621,
+      "step": 1445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3260818938464057e-05,
+      "loss": 0.5629,
+      "step": 1446
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3195626961154977e-05,
+      "loss": 0.5394,
+      "step": 1447
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.31304988655489e-05,
+      "loss": 0.7171,
+      "step": 1448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3065434806862807e-05,
+      "loss": 0.5404,
+      "step": 1449
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3000434940161063e-05,
+      "loss": 0.6561,
+      "step": 1450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2935499420355057e-05,
+      "loss": 0.7763,
+      "step": 1451
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2870628402202764e-05,
+      "loss": 0.533,
+      "step": 1452
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2805822040308517e-05,
+      "loss": 0.5967,
+      "step": 1453
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2741080489122472e-05,
+      "loss": 0.6131,
+      "step": 1454
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2676403902940395e-05,
+      "loss": 0.8155,
+      "step": 1455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2611792435903196e-05,
+      "loss": 0.5122,
+      "step": 1456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2547246241996584e-05,
+      "loss": 0.6178,
+      "step": 1457
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2482765475050733e-05,
+      "loss": 0.5749,
+      "step": 1458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.241835028873985e-05,
+      "loss": 0.5989,
+      "step": 1459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2354000836581834e-05,
+      "loss": 0.5719,
+      "step": 1460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.228971727193798e-05,
+      "loss": 0.5755,
+      "step": 1461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.222549974801253e-05,
+      "loss": 0.7157,
+      "step": 1462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2161348417852346e-05,
+      "loss": 0.5647,
+      "step": 1463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2097263434346476e-05,
+      "loss": 0.6441,
+      "step": 1464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2033244950225933e-05,
+      "loss": 0.5813,
+      "step": 1465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.196929311806316e-05,
+      "loss": 0.644,
+      "step": 1466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.190540809027182e-05,
+      "loss": 0.5488,
+      "step": 1467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.184159001910633e-05,
+      "loss": 0.5993,
+      "step": 1468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1777839056661554e-05,
+      "loss": 0.6164,
+      "step": 1469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1714155354872374e-05,
+      "loss": 0.5862,
+      "step": 1470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1650539065513413e-05,
+      "loss": 0.8451,
+      "step": 1471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.158699034019863e-05,
+      "loss": 0.5582,
+      "step": 1472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.152350933038097e-05,
+      "loss": 0.6017,
+      "step": 1473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1460096187351942e-05,
+      "loss": 0.6321,
+      "step": 1474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.139675106224138e-05,
+      "loss": 0.597,
+      "step": 1475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1333474106016948e-05,
+      "loss": 0.6813,
+      "step": 1476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1270265469483898e-05,
+      "loss": 0.6287,
+      "step": 1477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1207125303284637e-05,
+      "loss": 0.5243,
+      "step": 1478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.114405375789839e-05,
+      "loss": 0.6957,
+      "step": 1479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1081050983640866e-05,
+      "loss": 0.6314,
+      "step": 1480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1018117130663832e-05,
+      "loss": 0.5585,
+      "step": 1481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0955252348954807e-05,
+      "loss": 0.61,
+      "step": 1482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.089245678833671e-05,
+      "loss": 0.63,
+      "step": 1483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.08297305984675e-05,
+      "loss": 0.7016,
+      "step": 1484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0767073928839802e-05,
+      "loss": 0.7004,
+      "step": 1485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0704486928780513e-05,
+      "loss": 0.7308,
+      "step": 1486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0641969747450562e-05,
+      "loss": 0.6342,
+      "step": 1487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0579522533844397e-05,
+      "loss": 0.9144,
+      "step": 1488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0517145436789792e-05,
+      "loss": 0.6429,
+      "step": 1489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.045483860494738e-05,
+      "loss": 0.496,
+      "step": 1490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0392602186810356e-05,
+      "loss": 0.7459,
+      "step": 1491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0330436330704044e-05,
+      "loss": 0.5751,
+      "step": 1492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.026834118478567e-05,
+      "loss": 0.656,
+      "step": 1493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0206316897043897e-05,
+      "loss": 0.5786,
+      "step": 1494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.014436361529855e-05,
+      "loss": 0.6516,
+      "step": 1495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0082481487200172e-05,
+      "loss": 0.6674,
+      "step": 1496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.00206706602298e-05,
+      "loss": 0.6561,
+      "step": 1497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9958931281698463e-05,
+      "loss": 0.5772,
+      "step": 1498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9897263498746978e-05,
+      "loss": 0.7714,
+      "step": 1499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9835667458345493e-05,
+      "loss": 0.6226,
+      "step": 1500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.977414330729319e-05,
+      "loss": 0.7208,
+      "step": 1501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9712691192217946e-05,
+      "loss": 0.5344,
+      "step": 1502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9651311259575906e-05,
+      "loss": 0.731,
+      "step": 1503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.959000365565119e-05,
+      "loss": 0.6204,
+      "step": 1504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.952876852655557e-05,
+      "loss": 0.9347,
+      "step": 1505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.946760601822809e-05,
+      "loss": 0.7141,
+      "step": 1506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.940651627643471e-05,
+      "loss": 0.583,
+      "step": 1507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9345499446767982e-05,
+      "loss": 0.6868,
+      "step": 1508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9284555674646664e-05,
+      "loss": 0.678,
+      "step": 1509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9223685105315397e-05,
+      "loss": 0.6165,
+      "step": 1510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.916288788384439e-05,
+      "loss": 0.5915,
+      "step": 1511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9102164155129025e-05,
+      "loss": 0.5566,
+      "step": 1512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.904151406388957e-05,
+      "loss": 1.0993,
+      "step": 1513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8980937754670724e-05,
+      "loss": 0.6403,
+      "step": 1514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8920435371841394e-05,
+      "loss": 0.6678,
+      "step": 1515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8860007059594302e-05,
+      "loss": 0.6589,
+      "step": 1516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8799652961945653e-05,
+      "loss": 0.6553,
+      "step": 1517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8739373222734708e-05,
+      "loss": 0.5812,
+      "step": 1518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8679167985623603e-05,
+      "loss": 0.6353,
+      "step": 1519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.861903739409685e-05,
+      "loss": 0.624,
+      "step": 1520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8558981591461095e-05,
+      "loss": 0.6626,
+      "step": 1521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8499000720844734e-05,
+      "loss": 0.652,
+      "step": 1522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8439094925197593e-05,
+      "loss": 0.5833,
+      "step": 1523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8379264347290566e-05,
+      "loss": 0.6519,
+      "step": 1524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.831950912971528e-05,
+      "loss": 0.6662,
+      "step": 1525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8259829414883728e-05,
+      "loss": 0.559,
+      "step": 1526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8200225345028023e-05,
+      "loss": 0.6278,
+      "step": 1527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8140697062199963e-05,
+      "loss": 0.6195,
+      "step": 1528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.808124470827072e-05,
+      "loss": 0.6936,
+      "step": 1529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8021868424930555e-05,
+      "loss": 0.5996,
+      "step": 1530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7962568353688364e-05,
+      "loss": 0.5652,
+      "step": 1531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7903344635871434e-05,
+      "loss": 0.5533,
+      "step": 1532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7844197412625096e-05,
+      "loss": 0.5615,
+      "step": 1533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7785126824912378e-05,
+      "loss": 0.6175,
+      "step": 1534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772613301351366e-05,
+      "loss": 0.6786,
+      "step": 1535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7667216119026347e-05,
+      "loss": 0.6569,
+      "step": 1536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76083762818645e-05,
+      "loss": 0.7291,
+      "step": 1537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7549613642258573e-05,
+      "loss": 0.5694,
+      "step": 1538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.749092834025504e-05,
+      "loss": 0.6735,
+      "step": 1539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.743232051571601e-05,
+      "loss": 0.5267,
+      "step": 1540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.737379030831901e-05,
+      "loss": 0.7321,
+      "step": 1541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.731533785755652e-05,
+      "loss": 0.626,
+      "step": 1542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.725696330273575e-05,
+      "loss": 0.6458,
+      "step": 1543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7198666782978258e-05,
+      "loss": 0.8357,
+      "step": 1544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7140448437219614e-05,
+      "loss": 0.5754,
+      "step": 1545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7082308404209102e-05,
+      "loss": 0.5675,
+      "step": 1546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7024246822509325e-05,
+      "loss": 0.5379,
+      "step": 1547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6966263830495936e-05,
+      "loss": 0.5182,
+      "step": 1548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.69083595663573e-05,
+      "loss": 0.6632,
+      "step": 1549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.685053416809415e-05,
+      "loss": 0.71,
+      "step": 1550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6792787773519253e-05,
+      "loss": 0.6975,
+      "step": 1551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.673512052025711e-05,
+      "loss": 0.5441,
+      "step": 1552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.667753254574358e-05,
+      "loss": 0.7881,
+      "step": 1553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6620023987225563e-05,
+      "loss": 0.7055,
+      "step": 1554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6562594981760754e-05,
+      "loss": 0.547,
+      "step": 1555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6505245666217206e-05,
+      "loss": 0.5188,
+      "step": 1556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6447976177273066e-05,
+      "loss": 0.6209,
+      "step": 1557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6390786651416245e-05,
+      "loss": 0.5846,
+      "step": 1558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.633367722494404e-05,
+      "loss": 1.0263,
+      "step": 1559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6276648033962898e-05,
+      "loss": 0.7784,
+      "step": 1560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6219699214388e-05,
+      "loss": 0.799,
+      "step": 1561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6162830901943023e-05,
+      "loss": 0.5416,
+      "step": 1562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6106043232159746e-05,
+      "loss": 0.6199,
+      "step": 1563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6049336340377792e-05,
+      "loss": 0.7167,
+      "step": 1564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5992710361744205e-05,
+      "loss": 0.6931,
+      "step": 1565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.593616543121325e-05,
+      "loss": 0.6406,
+      "step": 1566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5879701683546023e-05,
+      "loss": 0.7903,
+      "step": 1567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5823319253310133e-05,
+      "loss": 0.7682,
+      "step": 1568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5767018274879382e-05,
+      "loss": 0.5636,
+      "step": 1569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5710798882433432e-05,
+      "loss": 0.7264,
+      "step": 1570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5654661209957554e-05,
+      "loss": 0.6036,
+      "step": 1571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5598605391242232e-05,
+      "loss": 0.7081,
+      "step": 1572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.554263155988287e-05,
+      "loss": 0.6884,
+      "step": 1573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.54867398492795e-05,
+      "loss": 0.6141,
+      "step": 1574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.543093039263641e-05,
+      "loss": 0.6111,
+      "step": 1575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5375203322961834e-05,
+      "loss": 0.7722,
+      "step": 1576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.531955877306771e-05,
+      "loss": 0.519,
+      "step": 1577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5263996875569288e-05,
+      "loss": 0.6501,
+      "step": 1578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5208517762884833e-05,
+      "loss": 0.5788,
+      "step": 1579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5153121567235335e-05,
+      "loss": 0.6994,
+      "step": 1580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5097808420644115e-05,
+      "loss": 0.5023,
+      "step": 1581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5042578454936634e-05,
+      "loss": 0.7767,
+      "step": 1582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4987431801740054e-05,
+      "loss": 0.5704,
+      "step": 1583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.493236859248302e-05,
+      "loss": 0.8919,
+      "step": 1584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.48773889583953e-05,
+      "loss": 0.752,
+      "step": 1585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4822493030507489e-05,
+      "loss": 0.5987,
+      "step": 1586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4767680939650658e-05,
+      "loss": 0.6097,
+      "step": 1587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4712952816456093e-05,
+      "loss": 0.7816,
+      "step": 1588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4658308791354975e-05,
+      "loss": 0.6896,
+      "step": 1589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4603748994578054e-05,
+      "loss": 0.5966,
+      "step": 1590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4549273556155319e-05,
+      "loss": 0.7563,
+      "step": 1591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4494882605915717e-05,
+      "loss": 0.5918,
+      "step": 1592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4440576273486855e-05,
+      "loss": 0.6783,
+      "step": 1593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.438635468829468e-05,
+      "loss": 0.7251,
+      "step": 1594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4332217979563134e-05,
+      "loss": 0.7353,
+      "step": 1595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4278166276313915e-05,
+      "loss": 0.643,
+      "step": 1596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4224199707366099e-05,
+      "loss": 0.707,
+      "step": 1597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4170318401335859e-05,
+      "loss": 0.8204,
+      "step": 1598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4116522486636191e-05,
+      "loss": 0.6249,
+      "step": 1599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4062812091476573e-05,
+      "loss": 0.6155,
+      "step": 1600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4009187343862673e-05,
+      "loss": 0.671,
+      "step": 1601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3955648371596048e-05,
+      "loss": 0.5932,
+      "step": 1602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3902195302273779e-05,
+      "loss": 0.7469,
+      "step": 1603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3848828263288288e-05,
+      "loss": 0.6382,
+      "step": 1604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3795547381826902e-05,
+      "loss": 0.5975,
+      "step": 1605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3742352784871648e-05,
+      "loss": 0.6445,
+      "step": 1606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.368924459919892e-05,
+      "loss": 0.6504,
+      "step": 1607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3636222951379168e-05,
+      "loss": 0.6699,
+      "step": 1608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3583287967776564e-05,
+      "loss": 0.5322,
+      "step": 1609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.353043977454878e-05,
+      "loss": 0.5804,
+      "step": 1610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3477678497646628e-05,
+      "loss": 0.5843,
+      "step": 1611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3425004262813795e-05,
+      "loss": 0.5745,
+      "step": 1612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3372417195586478e-05,
+      "loss": 0.5888,
+      "step": 1613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3319917421293182e-05,
+      "loss": 0.64,
+      "step": 1614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3267505065054331e-05,
+      "loss": 0.6937,
+      "step": 1615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3215180251782028e-05,
+      "loss": 0.5955,
+      "step": 1616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3162943106179749e-05,
+      "loss": 0.5401,
+      "step": 1617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3110793752742035e-05,
+      "loss": 0.5894,
+      "step": 1618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3058732315754169e-05,
+      "loss": 0.5576,
+      "step": 1619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3006758919291906e-05,
+      "loss": 0.5452,
+      "step": 1620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2954873687221208e-05,
+      "loss": 0.5902,
+      "step": 1621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2903076743197912e-05,
+      "loss": 0.9364,
+      "step": 1622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2851368210667426e-05,
+      "loss": 0.5031,
+      "step": 1623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2799748212864477e-05,
+      "loss": 0.4507,
+      "step": 1624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2748216872812745e-05,
+      "loss": 0.6638,
+      "step": 1625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2696774313324672e-05,
+      "loss": 0.5732,
+      "step": 1626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2645420657001062e-05,
+      "loss": 0.7568,
+      "step": 1627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2594156026230869e-05,
+      "loss": 0.5935,
+      "step": 1628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2542980543190874e-05,
+      "loss": 0.5709,
+      "step": 1629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2491894329845406e-05,
+      "loss": 0.5772,
+      "step": 1630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2440897507945997e-05,
+      "loss": 0.8734,
+      "step": 1631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2389990199031193e-05,
+      "loss": 0.4963,
+      "step": 1632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2339172524426162e-05,
+      "loss": 0.5985,
+      "step": 1633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2288444605242505e-05,
+      "loss": 0.5301,
+      "step": 1634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2237806562377835e-05,
+      "loss": 0.6055,
+      "step": 1635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2187258516515642e-05,
+      "loss": 0.5772,
+      "step": 1636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2136800588124874e-05,
+      "loss": 0.6151,
+      "step": 1637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2086432897459737e-05,
+      "loss": 0.9748,
+      "step": 1638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.203615556455937e-05,
+      "loss": 0.5705,
+      "step": 1639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1985968709247574e-05,
+      "loss": 0.6689,
+      "step": 1640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1935872451132518e-05,
+      "loss": 0.676,
+      "step": 1641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1885866909606435e-05,
+      "loss": 0.588,
+      "step": 1642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1835952203845352e-05,
+      "loss": 0.6122,
+      "step": 1643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1786128452808847e-05,
+      "loss": 0.6295,
+      "step": 1644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1736395775239712e-05,
+      "loss": 0.5283,
+      "step": 1645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1686754289663693e-05,
+      "loss": 0.6347,
+      "step": 1646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1637204114389177e-05,
+      "loss": 0.7181,
+      "step": 1647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.158774536750698e-05,
+      "loss": 0.5231,
+      "step": 1648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1538378166889962e-05,
+      "loss": 0.6227,
+      "step": 1649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.148910263019285e-05,
+      "loss": 0.6199,
+      "step": 1650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1439918874851908e-05,
+      "loss": 0.642,
+      "step": 1651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.139082701808466e-05,
+      "loss": 0.6337,
+      "step": 1652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1341827176889574e-05,
+      "loss": 0.7974,
+      "step": 1653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1292919468045877e-05,
+      "loss": 0.6575,
+      "step": 1654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1244104008113187e-05,
+      "loss": 0.6272,
+      "step": 1655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1195380913431303e-05,
+      "loss": 0.6634,
+      "step": 1656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.114675030011983e-05,
+      "loss": 0.5897,
+      "step": 1657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1098212284078036e-05,
+      "loss": 0.6015,
+      "step": 1658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.104976698098445e-05,
+      "loss": 0.6572,
+      "step": 1659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1001414506296681e-05,
+      "loss": 1.031,
+      "step": 1660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.095315497525109e-05,
+      "loss": 0.6575,
+      "step": 1661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0904988502862534e-05,
+      "loss": 0.5993,
+      "step": 1662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0856915203924096e-05,
+      "loss": 0.7561,
+      "step": 1663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0808935193006769e-05,
+      "loss": 0.5775,
+      "step": 1664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0761048584459232e-05,
+      "loss": 0.5819,
+      "step": 1665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.071325549240758e-05,
+      "loss": 1.0201,
+      "step": 1666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0665556030755026e-05,
+      "loss": 0.6283,
+      "step": 1667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.061795031318164e-05,
+      "loss": 0.8327,
+      "step": 1668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0570438453144043e-05,
+      "loss": 0.5755,
+      "step": 1669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0523020563875236e-05,
+      "loss": 0.5852,
+      "step": 1670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.047569675838418e-05,
+      "loss": 0.7616,
+      "step": 1671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0428467149455679e-05,
+      "loss": 0.5072,
+      "step": 1672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0381331849650011e-05,
+      "loss": 0.6235,
+      "step": 1673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0334290971302712e-05,
+      "loss": 0.7575,
+      "step": 1674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0287344626524248e-05,
+      "loss": 0.6692,
+      "step": 1675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0240492927199824e-05,
+      "loss": 0.7645,
+      "step": 1676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0193735984989072e-05,
+      "loss": 0.9141,
+      "step": 1677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.01470739113258e-05,
+      "loss": 0.6553,
+      "step": 1678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0100506817417688e-05,
+      "loss": 0.6692,
+      "step": 1679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0054034814246094e-05,
+      "loss": 0.6348,
+      "step": 1680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0007658012565718e-05,
+      "loss": 0.5832,
+      "step": 1681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.961376522904393e-06,
+      "loss": 0.6773,
+      "step": 1682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.91519045556279e-06,
+      "loss": 0.7213,
+      "step": 1683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.869099920614166e-06,
+      "loss": 0.5994,
+      "step": 1684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.823105027904123e-06,
+      "loss": 0.646,
+      "step": 1685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.777205887050272e-06,
+      "loss": 1.0717,
+      "step": 1686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.731402607442047e-06,
+      "loss": 0.6002,
+      "step": 1687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.685695298240432e-06,
+      "loss": 0.6578,
+      "step": 1688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.64008406837768e-06,
+      "loss": 0.6105,
+      "step": 1689
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.594569026557066e-06,
+      "loss": 0.5751,
+      "step": 1690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549150281252633e-06,
+      "loss": 0.5657,
+      "step": 1691
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.503827940708898e-06,
+      "loss": 0.8128,
+      "step": 1692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.458602112940617e-06,
+      "loss": 1.1223,
+      "step": 1693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.413472905732573e-06,
+      "loss": 0.7516,
+      "step": 1694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.368440426639246e-06,
+      "loss": 0.8094,
+      "step": 1695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.323504782984616e-06,
+      "loss": 0.5854,
+      "step": 1696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.278666081861825e-06,
+      "loss": 0.6684,
+      "step": 1697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.233924430133023e-06,
+      "loss": 0.6331,
+      "step": 1698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.189279934429063e-06,
+      "loss": 0.7127,
+      "step": 1699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.144732701149245e-06,
+      "loss": 0.7562,
+      "step": 1700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.100282836461038e-06,
+      "loss": 0.6869,
+      "step": 1701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.055930446299915e-06,
+      "loss": 1.2033,
+      "step": 1702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.011675636368977e-06,
+      "loss": 0.64,
+      "step": 1703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.967518512138811e-06,
+      "loss": 0.5092,
+      "step": 1704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.923459178847188e-06,
+      "loss": 0.5598,
+      "step": 1705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.879497741498804e-06,
+      "loss": 0.6555,
+      "step": 1706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.835634304865071e-06,
+      "loss": 0.5434,
+      "step": 1707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.791868973483803e-06,
+      "loss": 0.6644,
+      "step": 1708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.74820185165901e-06,
+      "loss": 0.556,
+      "step": 1709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.704633043460664e-06,
+      "loss": 0.6667,
+      "step": 1710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.66116265272443e-06,
+      "loss": 0.6006,
+      "step": 1711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.617790783051399e-06,
+      "loss": 0.6539,
+      "step": 1712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.574517537807897e-06,
+      "loss": 0.5757,
+      "step": 1713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.531343020125159e-06,
+      "loss": 0.5524,
+      "step": 1714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.48826733289913e-06,
+      "loss": 0.5986,
+      "step": 1715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.445290578790248e-06,
+      "loss": 0.5942,
+      "step": 1716
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.402412860223158e-06,
+      "loss": 0.6468,
+      "step": 1717
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.359634279386464e-06,
+      "loss": 0.8138,
+      "step": 1718
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.316954938232523e-06,
+      "loss": 0.5407,
+      "step": 1719
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.274374938477136e-06,
+      "loss": 0.6141,
+      "step": 1720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.231894381599392e-06,
+      "loss": 0.5732,
+      "step": 1721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.189513368841373e-06,
+      "loss": 0.6154,
+      "step": 1722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.147232001207894e-06,
+      "loss": 0.5743,
+      "step": 1723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.105050379466333e-06,
+      "loss": 0.5409,
+      "step": 1724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.062968604146293e-06,
+      "loss": 0.6069,
+      "step": 1725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.020986775539468e-06,
+      "loss": 0.6036,
+      "step": 1726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.979104993699339e-06,
+      "loss": 0.5906,
+      "step": 1727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.937323358440935e-06,
+      "loss": 0.63,
+      "step": 1728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.895641969340639e-06,
+      "loss": 0.6941,
+      "step": 1729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.85406092573589e-06,
+      "loss": 0.7143,
+      "step": 1730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.812580326724978e-06,
+      "loss": 0.6265,
+      "step": 1731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.771200271166833e-06,
+      "loss": 0.5241,
+      "step": 1732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.72992085768075e-06,
+      "loss": 0.5434,
+      "step": 1733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.688742184646169e-06,
+      "loss": 0.4872,
+      "step": 1734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.64766435020246e-06,
+      "loss": 0.6175,
+      "step": 1735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.606687452248629e-06,
+      "loss": 0.5887,
+      "step": 1736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.565811588443139e-06,
+      "loss": 0.5955,
+      "step": 1737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5250368562036765e-06,
+      "loss": 0.6562,
+      "step": 1738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.484363352706897e-06,
+      "loss": 0.6461,
+      "step": 1739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.443791174888204e-06,
+      "loss": 0.7503,
+      "step": 1740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.403320419441528e-06,
+      "loss": 0.6548,
+      "step": 1741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.362951182819044e-06,
+      "loss": 0.6224,
+      "step": 1742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.322683561231025e-06,
+      "loss": 0.5958,
+      "step": 1743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.282517650645543e-06,
+      "loss": 0.6933,
+      "step": 1744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.242453546788275e-06,
+      "loss": 0.7559,
+      "step": 1745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2024913451422875e-06,
+      "loss": 0.6208,
+      "step": 1746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.162631140947734e-06,
+      "loss": 0.5465,
+      "step": 1747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.122873029201732e-06,
+      "loss": 0.6418,
+      "step": 1748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.083217104658074e-06,
+      "loss": 0.7065,
+      "step": 1749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.043663461827004e-06,
+      "loss": 0.5052,
+      "step": 1750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.004212194975024e-06,
+      "loss": 0.7282,
+      "step": 1751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.964863398124621e-06,
+      "loss": 0.65,
+      "step": 1752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.92561716505406e-06,
+      "loss": 0.6926,
+      "step": 1753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.886473589297215e-06,
+      "loss": 0.8272,
+      "step": 1754
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8474327641432686e-06,
+      "loss": 0.6136,
+      "step": 1755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.808494782636532e-06,
+      "loss": 0.5653,
+      "step": 1756
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.769659737576229e-06,
+      "loss": 0.8037,
+      "step": 1757
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.730927721516228e-06,
+      "loss": 0.6109,
+      "step": 1758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.692298826764859e-06,
+      "loss": 0.6259,
+      "step": 1759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6537731453847e-06,
+      "loss": 0.4991,
+      "step": 1760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.615350769192347e-06,
+      "loss": 0.7911,
+      "step": 1761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.577031789758193e-06,
+      "loss": 0.5382,
+      "step": 1762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.538816298406203e-06,
+      "loss": 0.7098,
+      "step": 1763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.500704386213691e-06,
+      "loss": 0.6766,
+      "step": 1764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.462696144011149e-06,
+      "loss": 0.7184,
+      "step": 1765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.424791662381957e-06,
+      "loss": 0.672,
+      "step": 1766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.386991031662232e-06,
+      "loss": 0.5919,
+      "step": 1767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.349294341940593e-06,
+      "loss": 0.5943,
+      "step": 1768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.31170168305793e-06,
+      "loss": 0.6431,
+      "step": 1769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.274213144607183e-06,
+      "loss": 0.5868,
+      "step": 1770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.236828815933176e-06,
+      "loss": 0.5993,
+      "step": 1771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.199548786132364e-06,
+      "loss": 0.5965,
+      "step": 1772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.162373144052636e-06,
+      "loss": 0.6514,
+      "step": 1773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.125301978293091e-06,
+      "loss": 0.611,
+      "step": 1774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.088335377203814e-06,
+      "loss": 0.7747,
+      "step": 1775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.051473428885718e-06,
+      "loss": 0.5929,
+      "step": 1776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0147162211902975e-06,
+      "loss": 0.7349,
+      "step": 1777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.978063841719411e-06,
+      "loss": 0.7574,
+      "step": 1778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.941516377825102e-06,
+      "loss": 0.7227,
+      "step": 1779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.905073916609344e-06,
+      "loss": 0.6477,
+      "step": 1780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.868736544923875e-06,
+      "loss": 0.6646,
+      "step": 1781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.832504349369988e-06,
+      "loss": 0.6828,
+      "step": 1782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.796377416298315e-06,
+      "loss": 0.7446,
+      "step": 1783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.760355831808606e-06,
+      "loss": 0.5834,
+      "step": 1784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.724439681749555e-06,
+      "loss": 0.669,
+      "step": 1785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.688629051718547e-06,
+      "loss": 0.6908,
+      "step": 1786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.65292402706154e-06,
+      "loss": 0.5937,
+      "step": 1787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.617324692872744e-06,
+      "loss": 0.6144,
+      "step": 1788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.581831133994525e-06,
+      "loss": 0.5987,
+      "step": 1789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.546443435017146e-06,
+      "loss": 0.5733,
+      "step": 1790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.511161680278592e-06,
+      "loss": 0.5862,
+      "step": 1791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.475985953864321e-06,
+      "loss": 0.7222,
+      "step": 1792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.440916339607138e-06,
+      "loss": 0.7207,
+      "step": 1793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.405952921086926e-06,
+      "loss": 0.6279,
+      "step": 1794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.371095781630503e-06,
+      "loss": 0.61,
+      "step": 1795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.336345004311366e-06,
+      "loss": 0.6858,
+      "step": 1796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.30170067194955e-06,
+      "loss": 0.5055,
+      "step": 1797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.267162867111386e-06,
+      "loss": 0.6511,
+      "step": 1798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.232731672109336e-06,
+      "loss": 0.7006,
+      "step": 1799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.198407169001773e-06,
+      "loss": 0.5947,
+      "step": 1800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1641894395928205e-06,
+      "loss": 0.7206,
+      "step": 1801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.13007856543209e-06,
+      "loss": 0.5948,
+      "step": 1802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.096074627814557e-06,
+      "loss": 0.6532,
+      "step": 1803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.06217770778033e-06,
+      "loss": 0.6736,
+      "step": 1804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.028387886114489e-06,
+      "loss": 0.5173,
+      "step": 1805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9947052433468425e-06,
+      "loss": 0.54,
+      "step": 1806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.96112985975179e-06,
+      "loss": 0.7262,
+      "step": 1807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9276618153480724e-06,
+      "loss": 0.7727,
+      "step": 1808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8943011898986535e-06,
+      "loss": 0.6338,
+      "step": 1809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.861048062910445e-06,
+      "loss": 0.5467,
+      "step": 1810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.827902513634197e-06,
+      "loss": 0.523,
+      "step": 1811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.794864621064266e-06,
+      "loss": 0.6115,
+      "step": 1812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.761934463938444e-06,
+      "loss": 0.5681,
+      "step": 1813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.729112120737728e-06,
+      "loss": 0.5939,
+      "step": 1814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.696397669686192e-06,
+      "loss": 0.6074,
+      "step": 1815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.663791188750788e-06,
+      "loss": 0.9338,
+      "step": 1816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.631292755641125e-06,
+      "loss": 0.6586,
+      "step": 1817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.598902447809311e-06,
+      "loss": 0.6175,
+      "step": 1818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.566620342449768e-06,
+      "loss": 0.7125,
+      "step": 1819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.53444651649903e-06,
+      "loss": 0.5991,
+      "step": 1820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.502381046635601e-06,
+      "loss": 0.643,
+      "step": 1821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.470424009279717e-06,
+      "loss": 0.691,
+      "step": 1822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.43857548059321e-06,
+      "loss": 0.5785,
+      "step": 1823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.406835536479309e-06,
+      "loss": 0.5902,
+      "step": 1824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3752042525824375e-06,
+      "loss": 0.8398,
+      "step": 1825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.343681704288061e-06,
+      "loss": 0.7819,
+      "step": 1826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.312267966722516e-06,
+      "loss": 0.5817,
+      "step": 1827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2809631147528005e-06,
+      "loss": 0.7139,
+      "step": 1828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.249767222986423e-06,
+      "loss": 0.7232,
+      "step": 1829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2186803657711885e-06,
+      "loss": 0.7256,
+      "step": 1830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.187702617195077e-06,
+      "loss": 0.5834,
+      "step": 1831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.156834051086001e-06,
+      "loss": 0.5937,
+      "step": 1832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.126074741011681e-06,
+      "loss": 0.5237,
+      "step": 1833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.095424760279454e-06,
+      "loss": 0.6729,
+      "step": 1834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.064884181936107e-06,
+      "loss": 0.8045,
+      "step": 1835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.034453078767653e-06,
+      "loss": 0.7014,
+      "step": 1836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.004131523299243e-06,
+      "loss": 0.8916,
+      "step": 1837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.973919587794922e-06,
+      "loss": 0.5882,
+      "step": 1838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9438173442575e-06,
+      "loss": 0.6708,
+      "step": 1839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9138248644283286e-06,
+      "loss": 0.6022,
+      "step": 1840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.883942219787217e-06,
+      "loss": 0.6429,
+      "step": 1841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.854169481552155e-06,
+      "loss": 0.7517,
+      "step": 1842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8245067206792325e-06,
+      "loss": 0.9445,
+      "step": 1843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7949540078624225e-06,
+      "loss": 0.6154,
+      "step": 1844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.765511413533429e-06,
+      "loss": 0.8085,
+      "step": 1845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7361790078615232e-06,
+      "loss": 0.7859,
+      "step": 1846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7069568607533465e-06,
+      "loss": 0.7018,
+      "step": 1847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6778450418527733e-06,
+      "loss": 0.6016,
+      "step": 1848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.648843620540754e-06,
+      "loss": 0.9447,
+      "step": 1849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.619952665935117e-06,
+      "loss": 0.53,
+      "step": 1850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.591172246890434e-06,
+      "loss": 0.6914,
+      "step": 1851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.562502431997816e-06,
+      "loss": 0.608,
+      "step": 1852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5339432895848123e-06,
+      "loss": 0.5682,
+      "step": 1853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.505494887715172e-06,
+      "loss": 0.6463,
+      "step": 1854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.477157294188749e-06,
+      "loss": 0.6054,
+      "step": 1855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.448930576541309e-06,
+      "loss": 0.5652,
+      "step": 1856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4208148020443697e-06,
+      "loss": 0.8546,
+      "step": 1857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.39281003770503e-06,
+      "loss": 0.6054,
+      "step": 1858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3649163502658345e-06,
+      "loss": 0.8424,
+      "step": 1859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.337133806204612e-06,
+      "loss": 0.6794,
+      "step": 1860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.309462471734298e-06,
+      "loss": 0.8606,
+      "step": 1861
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.281902412802773e-06,
+      "loss": 0.5956,
+      "step": 1862
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2544536950927516e-06,
+      "loss": 0.6659,
+      "step": 1863
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.227116384021556e-06,
+      "loss": 0.5886,
+      "step": 1864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.19989054474103e-06,
+      "loss": 0.6493,
+      "step": 1865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.172776242137332e-06,
+      "loss": 0.5378,
+      "step": 1866
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1457735408308154e-06,
+      "loss": 0.6429,
+      "step": 1867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1188825051758575e-06,
+      "loss": 0.8006,
+      "step": 1868
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.092103199260682e-06,
+      "loss": 0.6837,
+      "step": 1869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.065435686907253e-06,
+      "loss": 0.6632,
+      "step": 1870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.038880031671104e-06,
+      "loss": 0.6731,
+      "step": 1871
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0124362968411813e-06,
+      "loss": 0.5387,
+      "step": 1872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.986104545439694e-06,
+      "loss": 0.6263,
+      "step": 1873
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9598848402219716e-06,
+      "loss": 0.6527,
+      "step": 1874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9337772436763002e-06,
+      "loss": 0.8104,
+      "step": 1875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9077818180237693e-06,
+      "loss": 0.6499,
+      "step": 1876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8818986252181714e-06,
+      "loss": 0.618,
+      "step": 1877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.85612772694579e-06,
+      "loss": 0.5409,
+      "step": 1878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8304691846253118e-06,
+      "loss": 0.7014,
+      "step": 1879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.804923059407605e-06,
+      "loss": 0.6324,
+      "step": 1880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7794894121756564e-06,
+      "loss": 0.5237,
+      "step": 1881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7541683035443797e-06,
+      "loss": 0.6532,
+      "step": 1882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7289597938604806e-06,
+      "loss": 0.6742,
+      "step": 1883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7038639432023005e-06,
+      "loss": 0.5938,
+      "step": 1884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.678880811379708e-06,
+      "loss": 0.6634,
+      "step": 1885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.654010457933903e-06,
+      "loss": 0.5933,
+      "step": 1886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6292529421373334e-06,
+      "loss": 0.5623,
+      "step": 1887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.604608322993518e-06,
+      "loss": 0.6954,
+      "step": 1888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5800766592369073e-06,
+      "loss": 0.588,
+      "step": 1889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5556580093327675e-06,
+      "loss": 0.5418,
+      "step": 1890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5313524314769964e-06,
+      "loss": 0.5456,
+      "step": 1891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.507159983596019e-06,
+      "loss": 0.7716,
+      "step": 1892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.483080723346665e-06,
+      "loss": 0.6641,
+      "step": 1893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.459114708115989e-06,
+      "loss": 0.6558,
+      "step": 1894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4352619950211653e-06,
+      "loss": 0.4839,
+      "step": 1895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.411522640909325e-06,
+      "loss": 0.6257,
+      "step": 1896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.387896702357456e-06,
+      "loss": 0.6226,
+      "step": 1897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3643842356722124e-06,
+      "loss": 0.6491,
+      "step": 1898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3409852968898582e-06,
+      "loss": 0.5273,
+      "step": 1899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.317699941776064e-06,
+      "loss": 0.6524,
+      "step": 1900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2945282258258148e-06,
+      "loss": 0.669,
+      "step": 1901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.271470204263265e-06,
+      "loss": 0.706,
+      "step": 1902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.248525932041584e-06,
+      "loss": 0.5368,
+      "step": 1903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2256954638428697e-06,
+      "loss": 0.6712,
+      "step": 1904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.202978854078003e-06,
+      "loss": 0.5477,
+      "step": 1905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1803761568864765e-06,
+      "loss": 0.6462,
+      "step": 1906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.15788742613634e-06,
+      "loss": 0.6153,
+      "step": 1907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1355127154239927e-06,
+      "loss": 0.5491,
+      "step": 1908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1132520780741295e-06,
+      "loss": 0.5372,
+      "step": 1909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.091105567139562e-06,
+      "loss": 0.4883,
+      "step": 1910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.069073235401109e-06,
+      "loss": 0.5765,
+      "step": 1911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0471551353674845e-06,
+      "loss": 0.5556,
+      "step": 1912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0253513192751373e-06,
+      "loss": 0.8237,
+      "step": 1913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0036618390881555e-06,
+      "loss": 0.6473,
+      "step": 1914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9820867464981465e-06,
+      "loss": 0.5466,
+      "step": 1915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9606260929240893e-06,
+      "loss": 0.6039,
+      "step": 1916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.939279929512244e-06,
+      "loss": 0.7144,
+      "step": 1917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9180483071359946e-06,
+      "loss": 0.6383,
+      "step": 1918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8969312763957414e-06,
+      "loss": 0.6092,
+      "step": 1919
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8759288876187887e-06,
+      "loss": 0.6207,
+      "step": 1920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8550411908592302e-06,
+      "loss": 0.5241,
+      "step": 1921
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8342682358978069e-06,
+      "loss": 0.572,
+      "step": 1922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8136100722418004e-06,
+      "loss": 0.7185,
+      "step": 1923
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7930667491249276e-06,
+      "loss": 0.5088,
+      "step": 1924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.772638315507197e-06,
+      "loss": 0.6211,
+      "step": 1925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7523248200748076e-06,
+      "loss": 0.536,
+      "step": 1926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.73212631124004e-06,
+      "loss": 0.7793,
+      "step": 1927
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7120428371411146e-06,
+      "loss": 0.7121,
+      "step": 1928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6920744456421279e-06,
+      "loss": 0.7231,
+      "step": 1929
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6722211843328673e-06,
+      "loss": 1.0526,
+      "step": 1930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6524831005287567e-06,
+      "loss": 0.6092,
+      "step": 1931
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6328602412707117e-06,
+      "loss": 0.741,
+      "step": 1932
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6133526533250565e-06,
+      "loss": 0.6973,
+      "step": 1933
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5939603831833793e-06,
+      "loss": 0.7007,
+      "step": 1934
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.574683477062433e-06,
+      "loss": 0.5549,
+      "step": 1935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5555219809040288e-06,
+      "loss": 0.6122,
+      "step": 1936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.536475940374943e-06,
+      "loss": 0.6838,
+      "step": 1937
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5175454008667712e-06,
+      "loss": 0.5035,
+      "step": 1938
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4987304074958576e-06,
+      "loss": 0.5019,
+      "step": 1939
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4800310051031662e-06,
+      "loss": 0.5799,
+      "step": 1940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4614472382541655e-06,
+      "loss": 0.5657,
+      "step": 1941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.442979151238738e-06,
+      "loss": 0.5743,
+      "step": 1942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4246267880710818e-06,
+      "loss": 0.7204,
+      "step": 1943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4063901924895984e-06,
+      "loss": 0.6046,
+      "step": 1944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.388269407956766e-06,
+      "loss": 0.8218,
+      "step": 1945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3702644776590778e-06,
+      "loss": 0.7591,
+      "step": 1946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3523754445068925e-06,
+      "loss": 0.7839,
+      "step": 1947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3346023511343842e-06,
+      "loss": 0.5803,
+      "step": 1948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.316945239899381e-06,
+      "loss": 0.7323,
+      "step": 1949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2994041528833266e-06,
+      "loss": 0.7468,
+      "step": 1950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2819791318911256e-06,
+      "loss": 0.5774,
+      "step": 1951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2646702184510862e-06,
+      "loss": 0.5969,
+      "step": 1952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2474774538147826e-06,
+      "loss": 0.5703,
+      "step": 1953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2304008789569887e-06,
+      "loss": 1.2667,
+      "step": 1954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2134405345755773e-06,
+      "loss": 0.5958,
+      "step": 1955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1965964610913981e-06,
+      "loss": 0.6654,
+      "step": 1956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.179868698648201e-06,
+      "loss": 0.6818,
+      "step": 1957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1632572871125346e-06,
+      "loss": 0.5629,
+      "step": 1958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1467622660736642e-06,
+      "loss": 0.6603,
+      "step": 1959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.130383674843455e-06,
+      "loss": 0.6659,
+      "step": 1960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1141215524563042e-06,
+      "loss": 0.7431,
+      "step": 1961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0979759376690158e-06,
+      "loss": 0.5453,
+      "step": 1962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0819468689607426e-06,
+      "loss": 0.5986,
+      "step": 1963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0660343845328657e-06,
+      "loss": 0.6855,
+      "step": 1964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0502385223089162e-06,
+      "loss": 0.7125,
+      "step": 1965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.034559319934497e-06,
+      "loss": 0.6078,
+      "step": 1966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0189968147771733e-06,
+      "loss": 0.9459,
+      "step": 1967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.003551043926393e-06,
+      "loss": 0.4722,
+      "step": 1968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.882220441933821e-07,
+      "loss": 0.7407,
+      "step": 1969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.730098521110954e-07,
+      "loss": 0.7013,
+      "step": 1970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.57914503934082e-07,
+      "loss": 1.0182,
+      "step": 1971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.429360356384364e-07,
+      "loss": 0.628,
+      "step": 1972
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.280744829216981e-07,
+      "loss": 0.675,
+      "step": 1973
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.133298812027624e-07,
+      "loss": 0.6148,
+      "step": 1974
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.987022656217925e-07,
+      "loss": 0.7342,
+      "step": 1975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.841916710401576e-07,
+      "loss": 0.4878,
+      "step": 1976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.697981320403337e-07,
+      "loss": 0.6495,
+      "step": 1977
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.555216829258195e-07,
+      "loss": 0.715,
+      "step": 1978
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.413623577210539e-07,
+      "loss": 0.4834,
+      "step": 1979
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.27320190171349e-07,
+      "loss": 0.7212,
+      "step": 1980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.133952137427847e-07,
+      "loss": 0.6027,
+      "step": 1981
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.995874616221699e-07,
+      "loss": 0.5326,
+      "step": 1982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.858969667169092e-07,
+      "loss": 0.7999,
+      "step": 1983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.723237616549805e-07,
+      "loss": 0.562,
+      "step": 1984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.588678787848136e-07,
+      "loss": 0.6799,
+      "step": 1985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.455293501752225e-07,
+      "loss": 0.6171,
+      "step": 1986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.323082076153509e-07,
+      "loss": 0.538,
+      "step": 1987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.192044826145771e-07,
+      "loss": 0.8723,
+      "step": 1988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.062182064024425e-07,
+      "loss": 0.7749,
+      "step": 1989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.933494099285787e-07,
+      "loss": 0.6374,
+      "step": 1990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.805981238626302e-07,
+      "loss": 0.7822,
+      "step": 1991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.679643785941881e-07,
+      "loss": 0.5958,
+      "step": 1992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.554482042327003e-07,
+      "loss": 0.7872,
+      "step": 1993
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.430496306074341e-07,
+      "loss": 1.1722,
+      "step": 1994
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.307686872673746e-07,
+      "loss": 0.6765,
+      "step": 1995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.186054034811595e-07,
+      "loss": 0.5929,
+      "step": 1996
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.065598082370172e-07,
+      "loss": 0.5837,
+      "step": 1997
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.94631930242695e-07,
+      "loss": 0.5938,
+      "step": 1998
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.828217979253869e-07,
+      "loss": 0.564,
+      "step": 1999
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.711294394316724e-07,
+      "loss": 0.5315,
+      "step": 2000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.595548826274388e-07,
+      "loss": 0.7075,
+      "step": 2001
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.48098155097837e-07,
+      "loss": 0.9447,
+      "step": 2002
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.367592841471814e-07,
+      "loss": 0.6183,
+      "step": 2003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.255382967989164e-07,
+      "loss": 0.721,
+      "step": 2004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.144352197955282e-07,
+      "loss": 0.616,
+      "step": 2005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.034500795985164e-07,
+      "loss": 0.7744,
+      "step": 2006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.925829023882778e-07,
+      "loss": 0.576,
+      "step": 2007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.818337140640894e-07,
+      "loss": 0.5553,
+      "step": 2008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.712025402440201e-07,
+      "loss": 0.6924,
+      "step": 2009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.60689406264897e-07,
+      "loss": 0.6522,
+      "step": 2010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.502943371822166e-07,
+      "loss": 0.7412,
+      "step": 2011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.40017357770095e-07,
+      "loss": 0.5314,
+      "step": 2012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.298584925212068e-07,
+      "loss": 0.8272,
+      "step": 2013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1981776564674613e-07,
+      "loss": 0.8516,
+      "step": 2014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.098952010763324e-07,
+      "loss": 0.6146,
+      "step": 2015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0009082245799357e-07,
+      "loss": 0.5939,
+      "step": 2016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9040465315808297e-07,
+      "loss": 0.5459,
+      "step": 2017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8083671626122365e-07,
+      "loss": 0.6326,
+      "step": 2018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7138703457025856e-07,
+      "loss": 0.6272,
+      "step": 2019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.620556306062173e-07,
+      "loss": 0.6179,
+      "step": 2020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.528425266082325e-07,
+      "loss": 0.601,
+      "step": 2021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.437477445334902e-07,
+      "loss": 0.6157,
+      "step": 2022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.347713060571855e-07,
+      "loss": 0.6446,
+      "step": 2023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2591323257248893e-07,
+      "loss": 0.6826,
+      "step": 2024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.171735451904578e-07,
+      "loss": 0.6842,
+      "step": 2025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0855226474000853e-07,
+      "loss": 0.7155,
+      "step": 2026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.00049411767872e-07,
+      "loss": 0.7242,
+      "step": 2027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.916650065385218e-07,
+      "loss": 0.5653,
+      "step": 2028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8339906903415705e-07,
+      "loss": 0.8818,
+      "step": 2029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7525161895462506e-07,
+      "loss": 0.6367,
+      "step": 2030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6722267571739344e-07,
+      "loss": 0.6006,
+      "step": 2031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.593122584574892e-07,
+      "loss": 0.619,
+      "step": 2032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5152038602747064e-07,
+      "loss": 0.5454,
+      "step": 2033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4384707699737773e-07,
+      "loss": 0.5362,
+      "step": 2034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3629234965465986e-07,
+      "loss": 0.659,
+      "step": 2035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2885622200419232e-07,
+      "loss": 0.6758,
+      "step": 2036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2153871176815999e-07,
+      "loss": 0.6973,
+      "step": 2037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.143398363860738e-07,
+      "loss": 0.62,
+      "step": 2038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.072596130147042e-07,
+      "loss": 0.5518,
+      "step": 2039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0029805852804228e-07,
+      "loss": 0.5326,
+      "step": 2040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.934551895172554e-07,
+      "loss": 0.7316,
+      "step": 2041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.867310222906593e-07,
+      "loss": 0.6242,
+      "step": 2042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8012557287367392e-07,
+      "loss": 0.679,
+      "step": 2043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.736388570087788e-07,
+      "loss": 0.6048,
+      "step": 2044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6727089015547982e-07,
+      "loss": 0.6095,
+      "step": 2045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6102168749028146e-07,
+      "loss": 0.725,
+      "step": 2046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.548912639066369e-07,
+      "loss": 1.0971,
+      "step": 2047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.488796340149201e-07,
+      "loss": 0.751,
+      "step": 2048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4298681214238162e-07,
+      "loss": 0.6571,
+      "step": 2049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3721281233314288e-07,
+      "loss": 0.6784,
+      "step": 2050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3155764834811846e-07,
+      "loss": 0.5919,
+      "step": 2051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2602133366501624e-07,
+      "loss": 0.6423,
+      "step": 2052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2060388147829839e-07,
+      "loss": 0.5467,
+      "step": 2053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1530530469914259e-07,
+      "loss": 0.5311,
+      "step": 2054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1012561595541981e-07,
+      "loss": 0.5221,
+      "step": 2055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0506482759165548e-07,
+      "loss": 0.6019,
+      "step": 2056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.001229516690072e-07,
+      "loss": 0.5634,
+      "step": 2057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.529999996524264e-08,
+      "loss": 0.6585,
+      "step": 2058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.05959839746784e-08,
+      "loss": 0.5587,
+      "step": 2059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.601091490820778e-08,
+      "loss": 0.8433,
+      "step": 2060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.15448036932176e-08,
+      "loss": 0.5822,
+      "step": 2061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.719766097359915e-08,
+      "loss": 0.5881,
+      "step": 2062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.296949710971501e-08,
+      "loss": 0.539,
+      "step": 2063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.886032217836014e-08,
+      "loss": 0.5495,
+      "step": 2064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.487014597275631e-08,
+      "loss": 0.7746,
+      "step": 2065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.099897800251885e-08,
+      "loss": 0.5878,
+      "step": 2066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.724682749362886e-08,
+      "loss": 0.6146,
+      "step": 2067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.361370338842764e-08,
+      "loss": 0.6863,
+      "step": 2068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.009961434558896e-08,
+      "loss": 0.6035,
+      "step": 2069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.670456874008022e-08,
+      "loss": 0.6753,
+      "step": 2070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.342857466316796e-08,
+      "loss": 0.5338,
+      "step": 2071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.027163992239014e-08,
+      "loss": 0.6081,
+      "step": 2072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7233772041533885e-08,
+      "loss": 0.6357,
+      "step": 2073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4314978260618915e-08,
+      "loss": 0.7017,
+      "step": 2074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.15152655358697e-08,
+      "loss": 0.9064,
+      "step": 2075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8834640539737723e-08,
+      "loss": 0.9467,
+      "step": 2076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6273109660823747e-08,
+      "loss": 0.6147,
+      "step": 2077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3830679003927768e-08,
+      "loss": 0.6558,
+      "step": 2078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1507354389982415e-08,
+      "loss": 0.5162,
+      "step": 2079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9303141356064037e-08,
+      "loss": 0.5784,
+      "step": 2080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7218045155381613e-08,
+      "loss": 0.641,
+      "step": 2081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5252070757254544e-08,
+      "loss": 0.7774,
+      "step": 2082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.340522284710155e-08,
+      "loss": 0.5309,
+      "step": 2083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1677505826440671e-08,
+      "loss": 0.5839,
+      "step": 2084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0068923812867059e-08,
+      "loss": 0.8782,
+      "step": 2085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.57948064003633e-09,
+      "loss": 0.5847,
+      "step": 2086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.209179857675663e-09,
+      "loss": 0.717,
+      "step": 2087
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.958024731567147e-09,
+      "loss": 0.6955,
+      "step": 2088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.826018243536678e-09,
+      "loss": 0.5498,
+      "step": 2089
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.813163091442862e-09,
+      "loss": 0.6121,
+      "step": 2090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9194616891770054e-09,
+      "loss": 0.545,
+      "step": 2091
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1449161666631246e-09,
+      "loss": 0.5141,
+      "step": 2092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4895283698412865e-09,
+      "loss": 0.6721,
+      "step": 2093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.532998606676113e-10,
+      "loss": 0.5613,
+      "step": 2094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.362319171142715e-10,
+      "loss": 0.7213,
+      "step": 2095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3832553315839e-10,
+      "loss": 0.741,
+      "step": 2096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.958141878759094e-11,
+      "loss": 0.5696,
+      "step": 2097
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.6439,
+      "step": 2098
+    },
+    {
+      "epoch": 1.0,
+      "step": 2098,
+      "total_flos": 108958163927040.0,
+      "train_loss": 0.7254286452602272,
+      "train_runtime": 23589.0086,
+      "train_samples_per_second": 1.423,
+      "train_steps_per_second": 0.089
+    }
+  ],
+  "max_steps": 2098,
+  "num_train_epochs": 1,
+  "total_flos": 108958163927040.0,
+  "trial_name": null,
+  "trial_params": null
+}