diff --git "a/checkpoints/llava-v1.5-13b-task-lora-diff-caption-2/trainer_state.json" "b/checkpoints/llava-v1.5-13b-task-lora-diff-caption-2/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoints/llava-v1.5-13b-task-lora-diff-caption-2/trainer_state.json"
@@ -0,0 +1,23778 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 3958,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.680672268907563e-07,
+      "loss": 1.609,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.361344537815126e-07,
+      "loss": 1.476,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.042016806722689e-07,
+      "loss": 1.5472,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.722689075630252e-07,
+      "loss": 1.6925,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.403361344537816e-07,
+      "loss": 1.639,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0084033613445378e-06,
+      "loss": 1.6607,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 1.6896,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3445378151260504e-06,
+      "loss": 1.622,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5126050420168068e-06,
+      "loss": 1.5708,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6806722689075632e-06,
+      "loss": 1.8501,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8487394957983196e-06,
+      "loss": 1.772,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0168067226890756e-06,
+      "loss": 1.6672,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.184873949579832e-06,
+      "loss": 1.693,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 1.6933,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.521008403361345e-06,
+      "loss": 1.665,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.689075630252101e-06,
+      "loss": 1.7308,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.6451,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0252100840336137e-06,
+      "loss": 1.5366,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1932773109243696e-06,
+      "loss": 1.6051,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3613445378151265e-06,
+      "loss": 1.6633,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 1.6364,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6974789915966393e-06,
+      "loss": 1.5778,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.865546218487396e-06,
+      "loss": 1.5677,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.033613445378151e-06,
+      "loss": 1.6372,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.201680672268908e-06,
+      "loss": 1.5491,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.369747899159664e-06,
+      "loss": 1.7031,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5378151260504205e-06,
+      "loss": 1.4363,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 1.5256,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.873949579831933e-06,
+      "loss": 1.5449,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.04201680672269e-06,
+      "loss": 1.6577,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.210084033613446e-06,
+      "loss": 1.4077,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.378151260504202e-06,
+      "loss": 1.6275,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.546218487394959e-06,
+      "loss": 1.5949,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.4772,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 1.5442,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.050420168067227e-06,
+      "loss": 1.483,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.218487394957984e-06,
+      "loss": 1.4361,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.386554621848739e-06,
+      "loss": 1.4979,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5546218487394966e-06,
+      "loss": 1.5242,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.722689075630253e-06,
+      "loss": 1.6659,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.8907563025210085e-06,
+      "loss": 1.5235,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 1.5889,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.226890756302521e-06,
+      "loss": 1.2259,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.394957983193279e-06,
+      "loss": 1.3594,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.563025210084034e-06,
+      "loss": 1.3563,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.731092436974791e-06,
+      "loss": 1.3298,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.899159663865547e-06,
+      "loss": 1.3096,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.067226890756303e-06,
+      "loss": 1.2215,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 1.3003,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.403361344537815e-06,
+      "loss": 1.3918,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.2997,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.739495798319328e-06,
+      "loss": 1.3155,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.907563025210085e-06,
+      "loss": 1.2148,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.075630252100841e-06,
+      "loss": 1.264,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.243697478991598e-06,
+      "loss": 1.2412,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 1.3218,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.579831932773111e-06,
+      "loss": 1.1579,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.747899159663867e-06,
+      "loss": 1.2279,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.915966386554622e-06,
+      "loss": 1.3128,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.008403361344538e-05,
+      "loss": 1.2283,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0252100840336135e-05,
+      "loss": 1.2546,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0420168067226892e-05,
+      "loss": 1.3098,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 1.2626,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0756302521008403e-05,
+      "loss": 1.1757,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0924369747899159e-05,
+      "loss": 1.1259,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1092436974789918e-05,
+      "loss": 1.1743,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1260504201680673e-05,
+      "loss": 1.3444,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.1473,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1596638655462186e-05,
+      "loss": 1.2014,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 1.2612,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1932773109243699e-05,
+      "loss": 1.1639,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2100840336134455e-05,
+      "loss": 1.1839,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2268907563025212e-05,
+      "loss": 1.2222,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2436974789915967e-05,
+      "loss": 1.2092,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2605042016806723e-05,
+      "loss": 1.2326,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2773109243697479e-05,
+      "loss": 1.2045,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 1.1483,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3109243697478993e-05,
+      "loss": 1.1844,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3277310924369749e-05,
+      "loss": 1.2756,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3445378151260506e-05,
+      "loss": 1.1276,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3613445378151261e-05,
+      "loss": 1.0973,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3781512605042017e-05,
+      "loss": 1.2206,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3949579831932774e-05,
+      "loss": 1.1175,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 1.22,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 1.1315,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4453781512605043e-05,
+      "loss": 1.0834,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4621848739495798e-05,
+      "loss": 1.0925,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4789915966386557e-05,
+      "loss": 1.0307,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4957983193277313e-05,
+      "loss": 1.1897,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5126050420168068e-05,
+      "loss": 1.128,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5294117647058822e-05,
+      "loss": 1.1891,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5462184873949583e-05,
+      "loss": 1.1092,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5630252100840337e-05,
+      "loss": 1.0794,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5798319327731094e-05,
+      "loss": 1.0865,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.596638655462185e-05,
+      "loss": 1.0572,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6134453781512605e-05,
+      "loss": 1.0677,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6302521008403362e-05,
+      "loss": 1.1558,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 1.149,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6638655462184877e-05,
+      "loss": 1.1098,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.680672268907563e-05,
+      "loss": 1.0762,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6974789915966388e-05,
+      "loss": 1.0914,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 1.1198,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7310924369747902e-05,
+      "loss": 1.0475,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7478991596638656e-05,
+      "loss": 1.0612,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 1.1266,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.781512605042017e-05,
+      "loss": 1.1456,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7983193277310925e-05,
+      "loss": 1.0426,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8151260504201682e-05,
+      "loss": 1.035,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.831932773109244e-05,
+      "loss": 1.1023,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8487394957983196e-05,
+      "loss": 1.1103,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.865546218487395e-05,
+      "loss": 1.0977,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 1.1595,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.899159663865546e-05,
+      "loss": 1.062,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9159663865546222e-05,
+      "loss": 1.1355,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9327731092436976e-05,
+      "loss": 0.99,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9495798319327733e-05,
+      "loss": 1.0512,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9663865546218487e-05,
+      "loss": 1.0298,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9831932773109244e-05,
+      "loss": 0.986,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9545,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996651629545e-05,
+      "loss": 1.0538,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999986606520422e-05,
+      "loss": 1.2241,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999969864679358e-05,
+      "loss": 1.1928,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999994642611756e-05,
+      "loss": 1.0822,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999916290850733e-05,
+      "loss": 1.1318,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999879458899046e-05,
+      "loss": 1.0122,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999835930287176e-05,
+      "loss": 1.0833,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999978570504427e-05,
+      "loss": 0.99,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999728783203957e-05,
+      "loss": 1.1349,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999966516480436e-05,
+      "loss": 1.0816,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999594849888083e-05,
+      "loss": 1.1803,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999517838502215e-05,
+      "loss": 1.1441,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999943413069833e-05,
+      "loss": 1.0832,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999343726532476e-05,
+      "loss": 1.0936,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999246626065205e-05,
+      "loss": 1.0169,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999142829361534e-05,
+      "loss": 1.1245,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999903233649098e-05,
+      "loss": 1.0781,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998915147527532e-05,
+      "loss": 1.0578,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998791262549674e-05,
+      "loss": 1.1214,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999866068164036e-05,
+      "loss": 1.095,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998523404887046e-05,
+      "loss": 0.9331,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999837943238166e-05,
+      "loss": 1.0818,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998228764220614e-05,
+      "loss": 0.997,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998071400504806e-05,
+      "loss": 1.0978,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999790734133962e-05,
+      "loss": 1.0154,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997736586834924e-05,
+      "loss": 1.0825,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997559137105064e-05,
+      "loss": 1.0854,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997374992268877e-05,
+      "loss": 0.9502,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999718415244968e-05,
+      "loss": 1.0463,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999698661777527e-05,
+      "loss": 1.1621,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996782388377934e-05,
+      "loss": 1.1682,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996571464394438e-05,
+      "loss": 1.0526,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996353845966033e-05,
+      "loss": 0.9902,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996129533238454e-05,
+      "loss": 1.07,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999589852636191e-05,
+      "loss": 1.0413,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999566082549111e-05,
+      "loss": 0.9605,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995416430785228e-05,
+      "loss": 0.9861,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995165342407935e-05,
+      "loss": 1.1316,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994907560527376e-05,
+      "loss": 1.0079,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999464308531618e-05,
+      "loss": 1.0839,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999437191695146e-05,
+      "loss": 1.0555,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994094055614814e-05,
+      "loss": 1.0061,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993809501492314e-05,
+      "loss": 1.1547,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993518254774517e-05,
+      "loss": 1.0973,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993220315656465e-05,
+      "loss": 0.944,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992915684337685e-05,
+      "loss": 1.0282,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992604361022173e-05,
+      "loss": 1.0189,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999228634591842e-05,
+      "loss": 1.0562,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999196163923939e-05,
+      "loss": 1.0178,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991630241202527e-05,
+      "loss": 1.0988,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991292152029767e-05,
+      "loss": 1.0395,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999094737194752e-05,
+      "loss": 1.1189,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990595901186668e-05,
+      "loss": 1.1058,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990237739982587e-05,
+      "loss": 1.0069,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998987288857513e-05,
+      "loss": 1.0596,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989501347208626e-05,
+      "loss": 1.0324,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998912311613188e-05,
+      "loss": 1.0903,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988738195598198e-05,
+      "loss": 1.002,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998834658586534e-05,
+      "loss": 0.9157,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998794828719556e-05,
+      "loss": 1.0499,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987543299855595e-05,
+      "loss": 1.0186,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987131624116642e-05,
+      "loss": 1.0439,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99867132602544e-05,
+      "loss": 1.088,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986288208549028e-05,
+      "loss": 0.9776,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985856469285178e-05,
+      "loss": 0.943,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985418042751975e-05,
+      "loss": 0.9531,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984972929243017e-05,
+      "loss": 1.0424,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984521129056392e-05,
+      "loss": 1.0978,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984062642494653e-05,
+      "loss": 0.9737,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983597469864837e-05,
+      "loss": 1.0433,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998312561147846e-05,
+      "loss": 1.0156,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982647067651515e-05,
+      "loss": 0.9891,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998216183870447e-05,
+      "loss": 1.0588,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981669924962263e-05,
+      "loss": 0.9509,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998117132675432e-05,
+      "loss": 0.9677,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980666044414546e-05,
+      "loss": 0.9975,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998015407828131e-05,
+      "loss": 1.0137,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979635428697462e-05,
+      "loss": 0.9912,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979110096010328e-05,
+      "loss": 1.0422,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997857808057171e-05,
+      "loss": 1.0896,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997803938273789e-05,
+      "loss": 0.9549,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997749400286961e-05,
+      "loss": 1.1128,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976941941332103e-05,
+      "loss": 1.0432,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976383198495067e-05,
+      "loss": 1.036,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975817774732684e-05,
+      "loss": 1.0277,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99752456704236e-05,
+      "loss": 0.9838,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974666885950937e-05,
+      "loss": 1.0019,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974081421702296e-05,
+      "loss": 1.0642,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973489278069744e-05,
+      "loss": 1.019,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997289045544982e-05,
+      "loss": 1.0666,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997228495424355e-05,
+      "loss": 1.1142,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971672774856413e-05,
+      "loss": 1.0013,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971053917698376e-05,
+      "loss": 1.0154,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970428383183866e-05,
+      "loss": 1.014,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969796171731795e-05,
+      "loss": 0.966,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969157283765532e-05,
+      "loss": 1.0976,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968511719712925e-05,
+      "loss": 1.0778,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967859480006294e-05,
+      "loss": 1.0439,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967200565082426e-05,
+      "loss": 0.9379,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996653497538258e-05,
+      "loss": 1.0584,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965862711352483e-05,
+      "loss": 0.9976,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996518377344233e-05,
+      "loss": 0.8875,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964498162106795e-05,
+      "loss": 1.1192,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963805877805004e-05,
+      "loss": 0.9411,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963106921000572e-05,
+      "loss": 1.1475,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996240129216157e-05,
+      "loss": 0.9235,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961688991760533e-05,
+      "loss": 0.9343,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960970020274474e-05,
+      "loss": 1.1694,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960244378184876e-05,
+      "loss": 0.9976,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959512065977673e-05,
+      "loss": 1.0745,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958773084143283e-05,
+      "loss": 1.0321,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958027433176574e-05,
+      "loss": 1.1256,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957275113576897e-05,
+      "loss": 1.1029,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995651612584806e-05,
+      "loss": 0.9976,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955750470498335e-05,
+      "loss": 1.1054,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954978148040462e-05,
+      "loss": 1.0558,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995419915899165e-05,
+      "loss": 1.0897,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953413503873554e-05,
+      "loss": 1.0229,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952621183212325e-05,
+      "loss": 0.9921,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951822197538547e-05,
+      "loss": 0.9899,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951016547387286e-05,
+      "loss": 1.0369,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950204233298063e-05,
+      "loss": 1.0466,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949385255814865e-05,
+      "loss": 0.972,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948559615486136e-05,
+      "loss": 0.9056,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994772731286479e-05,
+      "loss": 0.9606,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946888348508197e-05,
+      "loss": 1.007,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946042722978192e-05,
+      "loss": 1.0022,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945190436841067e-05,
+      "loss": 0.9328,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994433149066757e-05,
+      "loss": 1.0339,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943465885032926e-05,
+      "loss": 1.0895,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99425936205168e-05,
+      "loss": 0.9736,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941714697703333e-05,
+      "loss": 1.0656,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940829117181106e-05,
+      "loss": 1.0317,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939936879543177e-05,
+      "loss": 1.0418,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993903798538705e-05,
+      "loss": 0.9862,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99381324353147e-05,
+      "loss": 0.996,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937220229932543e-05,
+      "loss": 1.0849,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936301369851455e-05,
+      "loss": 1.0945,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935375855686782e-05,
+      "loss": 1.0477,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993444368805831e-05,
+      "loss": 1.1008,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933504867590297e-05,
+      "loss": 0.8161,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932559394911437e-05,
+      "loss": 1.0339,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993160727065489e-05,
+      "loss": 1.0144,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930648495458275e-05,
+      "loss": 0.9828,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992968306996365e-05,
+      "loss": 1.0712,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928710994817545e-05,
+      "loss": 0.989,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927732270670927e-05,
+      "loss": 1.0073,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926746898179222e-05,
+      "loss": 1.0167,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925754878002314e-05,
+      "loss": 0.9082,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924756210804528e-05,
+      "loss": 0.9733,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923750897254646e-05,
+      "loss": 1.1154,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922738938025904e-05,
+      "loss": 0.9071,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992172033379598e-05,
+      "loss": 0.968,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920695085247012e-05,
+      "loss": 0.9833,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919663193065575e-05,
+      "loss": 1.0728,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991862465794271e-05,
+      "loss": 1.0463,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991757948057389e-05,
+      "loss": 0.9573,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916527661659047e-05,
+      "loss": 0.8704,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915469201902552e-05,
+      "loss": 0.955,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914404102013238e-05,
+      "loss": 1.0961,
+      "step": 279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913332362704362e-05,
+      "loss": 1.1124,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912253984693647e-05,
+      "loss": 1.1377,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911168968703256e-05,
+      "loss": 0.9062,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910077315459794e-05,
+      "loss": 0.9528,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908979025694312e-05,
+      "loss": 1.0268,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990787410014231e-05,
+      "loss": 0.9945,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990676253954372e-05,
+      "loss": 1.1337,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905644344642936e-05,
+      "loss": 1.041,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904519516188777e-05,
+      "loss": 0.9634,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903388054934515e-05,
+      "loss": 0.9562,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902249961637855e-05,
+      "loss": 0.9156,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901105237060956e-05,
+      "loss": 0.9831,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899953881970404e-05,
+      "loss": 0.9939,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898795897137236e-05,
+      "loss": 0.8948,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897631283336917e-05,
+      "loss": 1.0365,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989646004134937e-05,
+      "loss": 0.9623,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895282171958937e-05,
+      "loss": 1.1275,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894097675954415e-05,
+      "loss": 1.144,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989290655412902e-05,
+      "loss": 0.9686,
+      "step": 298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989170880728042e-05,
+      "loss": 1.0106,
+      "step": 299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890504436210712e-05,
+      "loss": 1.0606,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988929344172644e-05,
+      "loss": 1.0831,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888075824638567e-05,
+      "loss": 0.9368,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886851585762508e-05,
+      "loss": 0.91,
+      "step": 303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98856207259181e-05,
+      "loss": 1.0081,
+      "step": 304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884383245929615e-05,
+      "loss": 1.0329,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883139146625763e-05,
+      "loss": 0.964,
+      "step": 306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881888428839687e-05,
+      "loss": 0.9781,
+      "step": 307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880631093408956e-05,
+      "loss": 0.9632,
+      "step": 308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879367141175582e-05,
+      "loss": 0.9565,
+      "step": 309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878096572985996e-05,
+      "loss": 1.0271,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876819389691068e-05,
+      "loss": 0.9512,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875535592146085e-05,
+      "loss": 0.9805,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987424518121079e-05,
+      "loss": 1.01,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987294815774932e-05,
+      "loss": 1.0319,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987164452263027e-05,
+      "loss": 0.9367,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870334276726648e-05,
+      "loss": 0.9867,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869017420915888e-05,
+      "loss": 0.9575,
+      "step": 317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867693956079857e-05,
+      "loss": 0.9824,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866363883104844e-05,
+      "loss": 1.1046,
+      "step": 319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865027202881566e-05,
+      "loss": 0.9397,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863683916305163e-05,
+      "loss": 1.03,
+      "step": 321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98623340242752e-05,
+      "loss": 0.9862,
+      "step": 322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860977527695657e-05,
+      "loss": 0.9594,
+      "step": 323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859614427474957e-05,
+      "loss": 1.1012,
+      "step": 324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858244724525924e-05,
+      "loss": 0.9273,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856868419765814e-05,
+      "loss": 0.9672,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985548551411631e-05,
+      "loss": 1.0672,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854096008503495e-05,
+      "loss": 0.9901,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852699903857895e-05,
+      "loss": 0.9892,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985129720111444e-05,
+      "loss": 1.0678,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984988790121249e-05,
+      "loss": 0.9953,
+      "step": 331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984847200509581e-05,
+      "loss": 0.8892,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984704951371259e-05,
+      "loss": 1.0338,
+      "step": 333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845620428015438e-05,
+      "loss": 0.9652,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844184748961374e-05,
+      "loss": 0.8412,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984274247751184e-05,
+      "loss": 0.9756,
+      "step": 336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841293614632677e-05,
+      "loss": 0.9994,
+      "step": 337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839838161294162e-05,
+      "loss": 0.9043,
+      "step": 338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838376118470965e-05,
+      "loss": 0.9883,
+      "step": 339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983690748714219e-05,
+      "loss": 0.9887,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835432268291327e-05,
+      "loss": 1.0107,
+      "step": 341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833950462906304e-05,
+      "loss": 1.1004,
+      "step": 342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983246207197944e-05,
+      "loss": 0.9886,
+      "step": 343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830967096507478e-05,
+      "loss": 1.007,
+      "step": 344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982946553749156e-05,
+      "loss": 1.0945,
+      "step": 345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982795739593724e-05,
+      "loss": 0.9527,
+      "step": 346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826442672854488e-05,
+      "loss": 0.9726,
+      "step": 347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824921369257665e-05,
+      "loss": 0.9984,
+      "step": 348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982339348616556e-05,
+      "loss": 0.9588,
+      "step": 349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821859024601345e-05,
+      "loss": 0.9292,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820317985592617e-05,
+      "loss": 1.0254,
+      "step": 351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818770370171367e-05,
+      "loss": 0.9493,
+      "step": 352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817216179373993e-05,
+      "loss": 0.9647,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815655414241298e-05,
+      "loss": 1.0137,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814088075818486e-05,
+      "loss": 1.0547,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812514165155164e-05,
+      "loss": 0.991,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810933683305335e-05,
+      "loss": 1.044,
+      "step": 357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809346631327408e-05,
+      "loss": 0.9047,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980775301028419e-05,
+      "loss": 0.9063,
+      "step": 359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806152821242894e-05,
+      "loss": 1.0476,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804546065275116e-05,
+      "loss": 0.9975,
+      "step": 361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802932743456865e-05,
+      "loss": 0.9211,
+      "step": 362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801312856868534e-05,
+      "loss": 0.9487,
+      "step": 363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799686406594932e-05,
+      "loss": 0.9084,
+      "step": 364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798053393725238e-05,
+      "loss": 0.9369,
+      "step": 365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796413819353043e-05,
+      "loss": 0.9556,
+      "step": 366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979476768457633e-05,
+      "loss": 0.9755,
+      "step": 367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979311499049747e-05,
+      "loss": 1.0035,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791455738223226e-05,
+      "loss": 0.9823,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789789928864764e-05,
+      "loss": 0.9931,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788117563537628e-05,
+      "loss": 1.0504,
+      "step": 371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978643864336176e-05,
+      "loss": 1.112,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784753169461486e-05,
+      "loss": 0.8965,
+      "step": 373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783061142965528e-05,
+      "loss": 0.9659,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978136256500699e-05,
+      "loss": 1.0607,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.977965743672337e-05,
+      "loss": 1.1661,
+      "step": 376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977794575925654e-05,
+      "loss": 1.0526,
+      "step": 377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776227533752776e-05,
+      "loss": 1.0379,
+      "step": 378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977450276136272e-05,
+      "loss": 0.9568,
+      "step": 379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772771443241413e-05,
+      "loss": 0.9177,
+      "step": 380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771033580548272e-05,
+      "loss": 0.933,
+      "step": 381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97692891744471e-05,
+      "loss": 1.0342,
+      "step": 382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767538226106078e-05,
+      "loss": 0.8861,
+      "step": 383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765780736697773e-05,
+      "loss": 1.0269,
+      "step": 384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976401670739913e-05,
+      "loss": 0.9866,
+      "step": 385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976224613939147e-05,
+      "loss": 0.8786,
+      "step": 386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760469033860505e-05,
+      "loss": 0.9947,
+      "step": 387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758685391996308e-05,
+      "loss": 0.9281,
+      "step": 388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756895214993337e-05,
+      "loss": 0.9933,
+      "step": 389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755098504050436e-05,
+      "loss": 0.8775,
+      "step": 390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975329526037081e-05,
+      "loss": 0.9529,
+      "step": 391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751485485162042e-05,
+      "loss": 1.1179,
+      "step": 392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749669179636098e-05,
+      "loss": 0.9486,
+      "step": 393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747846345009306e-05,
+      "loss": 0.9875,
+      "step": 394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746016982502376e-05,
+      "loss": 0.915,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974418109334038e-05,
+      "loss": 1.049,
+      "step": 396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742338678752766e-05,
+      "loss": 0.9871,
+      "step": 397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974048973997335e-05,
+      "loss": 0.9664,
+      "step": 398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738634278240326e-05,
+      "loss": 0.9914,
+      "step": 399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736772294796238e-05,
+      "loss": 0.9428,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734903790888017e-05,
+      "loss": 0.9377,
+      "step": 401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733028767766946e-05,
+      "loss": 1.0193,
+      "step": 402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731147226688683e-05,
+      "loss": 1.0571,
+      "step": 403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729259168913242e-05,
+      "loss": 0.9714,
+      "step": 404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727364595705012e-05,
+      "loss": 0.8692,
+      "step": 405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972546350833274e-05,
+      "loss": 0.9211,
+      "step": 406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972355590806953e-05,
+      "loss": 1.1492,
+      "step": 407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721641796192857e-05,
+      "loss": 0.9891,
+      "step": 408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719721173984546e-05,
+      "loss": 0.9182,
+      "step": 409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97177940427308e-05,
+      "loss": 0.8669,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715860403722153e-05,
+      "loss": 0.9172,
+      "step": 411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713920258253524e-05,
+      "loss": 0.9417,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711973607624177e-05,
+      "loss": 0.8922,
+      "step": 413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710020453137728e-05,
+      "loss": 0.9372,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970806079610216e-05,
+      "loss": 1.0511,
+      "step": 415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706094637829797e-05,
+      "loss": 0.9474,
+      "step": 416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704121979637335e-05,
+      "loss": 1.0655,
+      "step": 417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702142822845807e-05,
+      "loss": 1.0072,
+      "step": 418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97001571687806e-05,
+      "loss": 1.0731,
+      "step": 419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698165018771454e-05,
+      "loss": 0.9505,
+      "step": 420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969616637415247e-05,
+      "loss": 1.0984,
+      "step": 421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969416123626208e-05,
+      "loss": 1.0312,
+      "step": 422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692149606443076e-05,
+      "loss": 1.1069,
+      "step": 423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969013148604259e-05,
+      "loss": 1.0114,
+      "step": 424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968810687641211e-05,
+      "loss": 1.0924,
+      "step": 425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968607577890747e-05,
+      "loss": 1.0781,
+      "step": 426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684038194888827e-05,
+      "loss": 0.9142,
+      "step": 427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681994125720712e-05,
+      "loss": 0.9121,
+      "step": 428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679943572771982e-05,
+      "loss": 1.0341,
+      "step": 429
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677886537415837e-05,
+      "loss": 1.0266,
+      "step": 430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967582302102982e-05,
+      "loss": 1.0508,
+      "step": 431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673753024995816e-05,
+      "loss": 0.9132,
+      "step": 432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967167655070005e-05,
+      "loss": 0.9269,
+      "step": 433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669593599533078e-05,
+      "loss": 0.8069,
+      "step": 434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96675041728898e-05,
+      "loss": 0.9501,
+      "step": 435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966540827216945e-05,
+      "loss": 0.9317,
+      "step": 436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663305898775607e-05,
+      "loss": 0.9943,
+      "step": 437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661197054116165e-05,
+      "loss": 0.8313,
+      "step": 438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659081739603366e-05,
+      "loss": 0.9979,
+      "step": 439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656959956653786e-05,
+      "loss": 0.9456,
+      "step": 440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654831706688324e-05,
+      "loss": 1.0392,
+      "step": 441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652696991132213e-05,
+      "loss": 1.0748,
+      "step": 442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650555811415017e-05,
+      "loss": 1.0089,
+      "step": 443
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648408168970627e-05,
+      "loss": 0.9596,
+      "step": 444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964625406523727e-05,
+      "loss": 0.9734,
+      "step": 445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644093501657487e-05,
+      "loss": 1.0,
+      "step": 446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641926479678157e-05,
+      "loss": 1.0192,
+      "step": 447
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963975300075047e-05,
+      "loss": 1.0375,
+      "step": 448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963757306632996e-05,
+      "loss": 0.9553,
+      "step": 449
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635386677876463e-05,
+      "loss": 1.027,
+      "step": 450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963319383685415e-05,
+      "loss": 0.9247,
+      "step": 451
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963099454473151e-05,
+      "loss": 0.9041,
+      "step": 452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628788802981354e-05,
+      "loss": 0.879,
+      "step": 453
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626576613080804e-05,
+      "loss": 1.0201,
+      "step": 454
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624357976511315e-05,
+      "loss": 0.9513,
+      "step": 455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622132894758647e-05,
+      "loss": 0.9945,
+      "step": 456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961990136931287e-05,
+      "loss": 0.9853,
+      "step": 457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617663401668393e-05,
+      "loss": 1.012,
+      "step": 458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615418993323917e-05,
+      "loss": 0.9663,
+      "step": 459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613168145782468e-05,
+      "loss": 0.9894,
+      "step": 460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961091086055138e-05,
+      "loss": 1.0136,
+      "step": 461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960864713914229e-05,
+      "loss": 0.9519,
+      "step": 462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960637698307117e-05,
+      "loss": 1.0442,
+      "step": 463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604100393858266e-05,
+      "loss": 0.9714,
+      "step": 464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601817373028166e-05,
+      "loss": 1.0071,
+      "step": 465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959952792210974e-05,
+      "loss": 0.9137,
+      "step": 466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959723204263618e-05,
+      "loss": 0.9464,
+      "step": 467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 1.0031,
+      "step": 468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959262100417792e-05,
+      "loss": 1.0472,
+      "step": 469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590305848281113e-05,
+      "loss": 1.0241,
+      "step": 470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958798427000495e-05,
+      "loss": 1.0538,
+      "step": 471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958565627090414e-05,
+      "loss": 0.9177,
+      "step": 472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958332185253768e-05,
+      "loss": 1.0354,
+      "step": 473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580981016468864e-05,
+      "loss": 0.9507,
+      "step": 474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95786337642653e-05,
+      "loss": 0.9123,
+      "step": 475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957628009749887e-05,
+      "loss": 0.9359,
+      "step": 476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957392001774577e-05,
+      "loss": 1.0027,
+      "step": 477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571553526586487e-05,
+      "loss": 0.9649,
+      "step": 478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569180625605796e-05,
+      "loss": 0.9569,
+      "step": 479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566801316392765e-05,
+      "loss": 1.0343,
+      "step": 480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564415600540755e-05,
+      "loss": 0.9886,
+      "step": 481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956202347964743e-05,
+      "loss": 0.9908,
+      "step": 482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559624955314712e-05,
+      "loss": 0.9168,
+      "step": 483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557220029148848e-05,
+      "loss": 1.0318,
+      "step": 484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554808702760342e-05,
+      "loss": 0.9398,
+      "step": 485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552390977764008e-05,
+      "loss": 1.168,
+      "step": 486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549966855778923e-05,
+      "loss": 0.9975,
+      "step": 487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547536338428464e-05,
+      "loss": 0.7627,
+      "step": 488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954509942734029e-05,
+      "loss": 0.9511,
+      "step": 489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954265612414633e-05,
+      "loss": 0.9696,
+      "step": 490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540206430482804e-05,
+      "loss": 0.9076,
+      "step": 491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537750347990206e-05,
+      "loss": 0.9213,
+      "step": 492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535287878313315e-05,
+      "loss": 0.9386,
+      "step": 493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532819023101183e-05,
+      "loss": 0.9665,
+      "step": 494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530343784007132e-05,
+      "loss": 0.9357,
+      "step": 495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952786216268877e-05,
+      "loss": 0.875,
+      "step": 496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525374160807975e-05,
+      "loss": 0.9121,
+      "step": 497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522879780030894e-05,
+      "loss": 0.9435,
+      "step": 498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520379022027955e-05,
+      "loss": 0.9141,
+      "step": 499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517871888473847e-05,
+      "loss": 0.8633,
+      "step": 500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515358381047534e-05,
+      "loss": 0.9859,
+      "step": 501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512838501432244e-05,
+      "loss": 1.0109,
+      "step": 502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510312251315476e-05,
+      "loss": 0.9877,
+      "step": 503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507779632388997e-05,
+      "loss": 1.0716,
+      "step": 504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505240646348833e-05,
+      "loss": 0.9126,
+      "step": 505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950269529489528e-05,
+      "loss": 0.8982,
+      "step": 506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500143579732887e-05,
+      "loss": 1.0378,
+      "step": 507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497585502570486e-05,
+      "loss": 1.0823,
+      "step": 508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949502106512114e-05,
+      "loss": 0.9331,
+      "step": 509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492450269102195e-05,
+      "loss": 1.0454,
+      "step": 510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489873116235242e-05,
+      "loss": 0.9492,
+      "step": 511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487289608246135e-05,
+      "loss": 0.8282,
+      "step": 512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948469974686498e-05,
+      "loss": 0.9873,
+      "step": 513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948210353382615e-05,
+      "loss": 0.983,
+      "step": 514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947950097086825e-05,
+      "loss": 0.9466,
+      "step": 515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476892059734154e-05,
+      "loss": 1.0009,
+      "step": 516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947427680217098e-05,
+      "loss": 1.0702,
+      "step": 517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94716551999301e-05,
+      "loss": 1.0741,
+      "step": 518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469027254767134e-05,
+      "loss": 0.9581,
+      "step": 519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466392968441946e-05,
+      "loss": 0.9624,
+      "step": 520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463752342718655e-05,
+      "loss": 0.8537,
+      "step": 521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461105379365612e-05,
+      "loss": 0.8857,
+      "step": 522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458452080155425e-05,
+      "loss": 0.9485,
+      "step": 523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455792446864933e-05,
+      "loss": 0.8767,
+      "step": 524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453126481275237e-05,
+      "loss": 1.0529,
+      "step": 525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945045418517165e-05,
+      "loss": 0.9236,
+      "step": 526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944777556034375e-05,
+      "loss": 0.9368,
+      "step": 527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445090608585336e-05,
+      "loss": 0.9903,
+      "step": 528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442399331694456e-05,
+      "loss": 0.9621,
+      "step": 529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439701731473383e-05,
+      "loss": 0.9888,
+      "step": 530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436997809728636e-05,
+      "loss": 0.9009,
+      "step": 531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434287568270954e-05,
+      "loss": 0.9388,
+      "step": 532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431571008915323e-05,
+      "loss": 0.988,
+      "step": 533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428848133480948e-05,
+      "loss": 0.9851,
+      "step": 534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942611894379127e-05,
+      "loss": 0.8986,
+      "step": 535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423383441673954e-05,
+      "loss": 0.8113,
+      "step": 536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420641628960897e-05,
+      "loss": 1.0037,
+      "step": 537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941789350748822e-05,
+      "loss": 0.942,
+      "step": 538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941513907909627e-05,
+      "loss": 0.9331,
+      "step": 539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412378345629613e-05,
+      "loss": 0.9705,
+      "step": 540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940961130893704e-05,
+      "loss": 0.92,
+      "step": 541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940683797087157e-05,
+      "loss": 0.9074,
+      "step": 542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940405833329043e-05,
+      "loss": 0.8661,
+      "step": 543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940127239805507e-05,
+      "loss": 0.9529,
+      "step": 544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939848016703117e-05,
+      "loss": 0.9029,
+      "step": 545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93956816420886e-05,
+      "loss": 0.9436,
+      "step": 546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392876825101465e-05,
+      "loss": 1.0003,
+      "step": 547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390065717948084e-05,
+      "loss": 1.0145,
+      "step": 548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387248322510977e-05,
+      "loss": 0.8913,
+      "step": 549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938442464067688e-05,
+      "loss": 1.0745,
+      "step": 550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938159467433674e-05,
+      "loss": 0.9766,
+      "step": 551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378758425385716e-05,
+      "loss": 0.8942,
+      "step": 552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937591589572317e-05,
+      "loss": 0.9739,
+      "step": 553
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373067087252663e-05,
+      "loss": 1.0571,
+      "step": 554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370212001881978e-05,
+      "loss": 0.9844,
+      "step": 555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367350641523083e-05,
+      "loss": 0.905,
+      "step": 556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364483008092164e-05,
+      "loss": 0.9976,
+      "step": 557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361609103509597e-05,
+      "loss": 0.91,
+      "step": 558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358728929699966e-05,
+      "loss": 1.0172,
+      "step": 559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935584248859204e-05,
+      "loss": 0.926,
+      "step": 560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352949782118805e-05,
+      "loss": 0.9582,
+      "step": 561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935005081221742e-05,
+      "loss": 0.9993,
+      "step": 562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934714558082926e-05,
+      "loss": 0.9855,
+      "step": 563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344234089899872e-05,
+      "loss": 1.0098,
+      "step": 564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341316341379016e-05,
+      "loss": 0.9316,
+      "step": 565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338392337220634e-05,
+      "loss": 1.0519,
+      "step": 566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335462079382848e-05,
+      "loss": 0.8369,
+      "step": 567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332525569827978e-05,
+      "loss": 0.9161,
+      "step": 568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932958281052253e-05,
+      "loss": 1.0608,
+      "step": 569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326633803437197e-05,
+      "loss": 0.9283,
+      "step": 570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323678550546844e-05,
+      "loss": 0.9542,
+      "step": 571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320717053830532e-05,
+      "loss": 0.9905,
+      "step": 572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93177493152715e-05,
+      "loss": 0.8875,
+      "step": 573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314775336857165e-05,
+      "loss": 0.9548,
+      "step": 574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931179512057912e-05,
+      "loss": 1.0203,
+      "step": 575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308808668433142e-05,
+      "loss": 1.0569,
+      "step": 576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305815982419182e-05,
+      "loss": 0.9739,
+      "step": 577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930281706454136e-05,
+      "loss": 0.9398,
+      "step": 578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929981191680798e-05,
+      "loss": 0.9784,
+      "step": 579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92968005412315e-05,
+      "loss": 0.9616,
+      "step": 580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929378293982857e-05,
+      "loss": 0.8958,
+      "step": 581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929075911462e-05,
+      "loss": 0.9131,
+      "step": 582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287729067630764e-05,
+      "loss": 1.0431,
+      "step": 583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284692800890004e-05,
+      "loss": 0.9576,
+      "step": 584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928165031643103e-05,
+      "loss": 0.9258,
+      "step": 585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278601616291318e-05,
+      "loss": 1.0116,
+      "step": 586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275546702512505e-05,
+      "loss": 1.0022,
+      "step": 587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272485577140384e-05,
+      "loss": 0.9777,
+      "step": 588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269418242224905e-05,
+      "loss": 0.9963,
+      "step": 589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266344699820196e-05,
+      "loss": 0.8869,
+      "step": 590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926326495198452e-05,
+      "loss": 0.9844,
+      "step": 591
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926017900078031e-05,
+      "loss": 0.9453,
+      "step": 592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257086848274143e-05,
+      "loss": 0.8989,
+      "step": 593
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253988496536754e-05,
+      "loss": 0.8952,
+      "step": 594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925088394764303e-05,
+      "loss": 0.8558,
+      "step": 595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247773203672005e-05,
+      "loss": 0.8289,
+      "step": 596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924465626670687e-05,
+      "loss": 0.9867,
+      "step": 597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241533138834943e-05,
+      "loss": 0.9394,
+      "step": 598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923840382214772e-05,
+      "loss": 0.9681,
+      "step": 599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235268318740806e-05,
+      "loss": 0.9679,
+      "step": 600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923212663071398e-05,
+      "loss": 1.0052,
+      "step": 601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228978760171143e-05,
+      "loss": 0.9979,
+      "step": 602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922582470922034e-05,
+      "loss": 1.0549,
+      "step": 603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222664479973767e-05,
+      "loss": 0.9121,
+      "step": 604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219498074547735e-05,
+      "loss": 0.9116,
+      "step": 605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921632549506271e-05,
+      "loss": 1.0296,
+      "step": 606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921314674364329e-05,
+      "loss": 1.0295,
+      "step": 607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209961822418194e-05,
+      "loss": 1.0178,
+      "step": 608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920677073352029e-05,
+      "loss": 1.0158,
+      "step": 609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203573479086558e-05,
+      "loss": 1.0702,
+      "step": 610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200370061258125e-05,
+      "loss": 1.0122,
+      "step": 611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919716048218023e-05,
+      "loss": 0.8921,
+      "step": 612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919394474400225e-05,
+      "loss": 0.8492,
+      "step": 613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190722848877683e-05,
+      "loss": 0.9607,
+      "step": 614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187494798964145e-05,
+      "loss": 1.0887,
+      "step": 615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184260596423375e-05,
+      "loss": 0.9555,
+      "step": 616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918102024342124e-05,
+      "loss": 0.9191,
+      "step": 617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177773742127718e-05,
+      "loss": 0.9897,
+      "step": 618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174521094716906e-05,
+      "loss": 0.9918,
+      "step": 619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917126230336702e-05,
+      "loss": 0.9603,
+      "step": 620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167997370260384e-05,
+      "loss": 0.9976,
+      "step": 621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164726297583444e-05,
+      "loss": 0.9352,
+      "step": 622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916144908752675e-05,
+      "loss": 0.9702,
+      "step": 623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158165742284963e-05,
+      "loss": 0.9429,
+      "step": 624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154876264056863e-05,
+      "loss": 0.9635,
+      "step": 625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151580655045317e-05,
+      "loss": 0.9817,
+      "step": 626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148278917457314e-05,
+      "loss": 1.0401,
+      "step": 627
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144971053503942e-05,
+      "loss": 0.9279,
+      "step": 628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141657065400392e-05,
+      "loss": 0.9346,
+      "step": 629
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138336955365955e-05,
+      "loss": 0.922,
+      "step": 630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135010725624025e-05,
+      "loss": 0.9846,
+      "step": 631
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913167837840209e-05,
+      "loss": 0.9515,
+      "step": 632
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128339915931737e-05,
+      "loss": 0.9625,
+      "step": 633
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124995340448644e-05,
+      "loss": 1.0257,
+      "step": 634
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121644654192592e-05,
+      "loss": 0.949,
+      "step": 635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911828785940745e-05,
+      "loss": 0.9039,
+      "step": 636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911492495834117e-05,
+      "loss": 0.9121,
+      "step": 637
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111555953245805e-05,
+      "loss": 0.9898,
+      "step": 638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108180846377486e-05,
+      "loss": 0.9558,
+      "step": 639
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910479963999644e-05,
+      "loss": 0.9491,
+      "step": 640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101412336366967e-05,
+      "loss": 0.9758,
+      "step": 641
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098018937757462e-05,
+      "loss": 0.8521,
+      "step": 642
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094619446440394e-05,
+      "loss": 0.9822,
+      "step": 643
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091213864692317e-05,
+      "loss": 0.9907,
+      "step": 644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087802194793853e-05,
+      "loss": 0.985,
+      "step": 645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908438443902972e-05,
+      "loss": 0.8395,
+      "step": 646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908096059968869e-05,
+      "loss": 0.916,
+      "step": 647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077530679063624e-05,
+      "loss": 0.8961,
+      "step": 648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074094679451454e-05,
+      "loss": 0.9316,
+      "step": 649
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070652603153177e-05,
+      "loss": 1.0228,
+      "step": 650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906720445247386e-05,
+      "loss": 0.884,
+      "step": 651
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063750229722645e-05,
+      "loss": 0.9177,
+      "step": 652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906028993721273e-05,
+      "loss": 1.0095,
+      "step": 653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905682357726139e-05,
+      "loss": 0.9163,
+      "step": 654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905335115218995e-05,
+      "loss": 0.9513,
+      "step": 655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049872664323806e-05,
+      "loss": 0.9976,
+      "step": 656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046388115992412e-05,
+      "loss": 0.9481,
+      "step": 657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904289750952928e-05,
+      "loss": 0.8052,
+      "step": 658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039400847271973e-05,
+      "loss": 0.9542,
+      "step": 659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035898131562123e-05,
+      "loss": 1.0356,
+      "step": 660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032389364745402e-05,
+      "loss": 0.9877,
+      "step": 661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028874549171547e-05,
+      "loss": 0.941,
+      "step": 662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902535368719433e-05,
+      "loss": 0.8768,
+      "step": 663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021826781171588e-05,
+      "loss": 0.9253,
+      "step": 664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90182938334652e-05,
+      "loss": 0.9403,
+      "step": 665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901475484644108e-05,
+      "loss": 0.9773,
+      "step": 666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90112098224692e-05,
+      "loss": 0.9875,
+      "step": 667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007658763923578e-05,
+      "loss": 1.0129,
+      "step": 668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900410167318226e-05,
+      "loss": 0.9099,
+      "step": 669
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000538552627337e-05,
+      "loss": 0.9665,
+      "step": 670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996969404644942e-05,
+      "loss": 0.9079,
+      "step": 671
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993394231625232e-05,
+      "loss": 1.0232,
+      "step": 672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989813035962418e-05,
+      "loss": 0.9059,
+      "step": 673
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986225820054727e-05,
+      "loss": 0.8933,
+      "step": 674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898263258630443e-05,
+      "loss": 0.9049,
+      "step": 675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897903333711782e-05,
+      "loss": 1.073,
+      "step": 676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975428074905222e-05,
+      "loss": 0.9406,
+      "step": 677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971816802080987e-05,
+      "loss": 0.9552,
+      "step": 678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968199521063484e-05,
+      "loss": 0.9201,
+      "step": 679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964576234275123e-05,
+      "loss": 1.0264,
+      "step": 680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896094694414232e-05,
+      "loss": 0.9051,
+      "step": 681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957311653095517e-05,
+      "loss": 0.9367,
+      "step": 682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895367036356917e-05,
+      "loss": 0.868,
+      "step": 683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950023078001766e-05,
+      "loss": 0.9012,
+      "step": 684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946369798835787e-05,
+      "loss": 1.0091,
+      "step": 685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942710528517745e-05,
+      "loss": 0.9713,
+      "step": 686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893904526949816e-05,
+      "loss": 0.9221,
+      "step": 687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935374024231557e-05,
+      "loss": 0.9033,
+      "step": 688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893169679517648e-05,
+      "loss": 0.9005,
+      "step": 689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928013584795465e-05,
+      "loss": 0.9542,
+      "step": 690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924324395555066e-05,
+      "loss": 1.0048,
+      "step": 691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920629229925846e-05,
+      "loss": 0.8786,
+      "step": 692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916928090382347e-05,
+      "loss": 0.8874,
+      "step": 693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913220979403138e-05,
+      "loss": 0.9049,
+      "step": 694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890950789947077e-05,
+      "loss": 0.9362,
+      "step": 695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905788853071792e-05,
+      "loss": 0.9848,
+      "step": 696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902063842696767e-05,
+      "loss": 1.044,
+      "step": 697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898332870840227e-05,
+      "loss": 0.886,
+      "step": 698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894595940000705e-05,
+      "loss": 0.9674,
+      "step": 699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890853052680737e-05,
+      "loss": 1.0261,
+      "step": 700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887104211386828e-05,
+      "loss": 0.9911,
+      "step": 701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883349418629487e-05,
+      "loss": 0.8957,
+      "step": 702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879588676923194e-05,
+      "loss": 1.0421,
+      "step": 703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875821988786426e-05,
+      "loss": 0.9257,
+      "step": 704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872049356741636e-05,
+      "loss": 1.0081,
+      "step": 705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868270783315258e-05,
+      "loss": 0.9289,
+      "step": 706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88644862710377e-05,
+      "loss": 0.9552,
+      "step": 707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860695822443357e-05,
+      "loss": 0.9553,
+      "step": 708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856899440070594e-05,
+      "loss": 0.9464,
+      "step": 709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853097126461747e-05,
+      "loss": 0.9035,
+      "step": 710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884928888416313e-05,
+      "loss": 1.1029,
+      "step": 711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845474715725022e-05,
+      "loss": 0.9602,
+      "step": 712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841654623701673e-05,
+      "loss": 0.9348,
+      "step": 713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88378286106513e-05,
+      "loss": 0.9415,
+      "step": 714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8833996679136084e-05,
+      "loss": 0.9362,
+      "step": 715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830158831722173e-05,
+      "loss": 0.9827,
+      "step": 716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826315070979674e-05,
+      "loss": 0.8924,
+      "step": 717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822465399482648e-05,
+      "loss": 0.9711,
+      "step": 718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818609819809123e-05,
+      "loss": 1.0053,
+      "step": 719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881474833454109e-05,
+      "loss": 0.9968,
+      "step": 720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810880946264468e-05,
+      "loss": 0.9441,
+      "step": 721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807007657569164e-05,
+      "loss": 1.006,
+      "step": 722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880312847104901e-05,
+      "loss": 0.8653,
+      "step": 723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799243389301796e-05,
+      "loss": 0.9786,
+      "step": 724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795352414929265e-05,
+      "loss": 0.9686,
+      "step": 725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791455550537102e-05,
+      "loss": 1.0142,
+      "step": 726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878755279873493e-05,
+      "loss": 0.9092,
+      "step": 727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783644162136324e-05,
+      "loss": 1.0149,
+      "step": 728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8779729643358798e-05,
+      "loss": 0.923,
+      "step": 729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775809245023804e-05,
+      "loss": 0.9808,
+      "step": 730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771882969756732e-05,
+      "loss": 0.9055,
+      "step": 731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767950820186903e-05,
+      "loss": 0.8868,
+      "step": 732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764012798947576e-05,
+      "loss": 0.9326,
+      "step": 733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760068908675945e-05,
+      "loss": 0.995,
+      "step": 734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756119152013134e-05,
+      "loss": 0.9418,
+      "step": 735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875216353160418e-05,
+      "loss": 0.897,
+      "step": 736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748202050098075e-05,
+      "loss": 0.8383,
+      "step": 737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744234710147712e-05,
+      "loss": 1.0276,
+      "step": 738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740261514409913e-05,
+      "loss": 0.9105,
+      "step": 739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873628246554543e-05,
+      "loss": 0.8913,
+      "step": 740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732297566218932e-05,
+      "loss": 0.9235,
+      "step": 741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728306819098995e-05,
+      "loss": 0.9085,
+      "step": 742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724310226858117e-05,
+      "loss": 0.868,
+      "step": 743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720307792172724e-05,
+      "loss": 0.9925,
+      "step": 744
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716299517723135e-05,
+      "loss": 0.9701,
+      "step": 745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712285406193585e-05,
+      "loss": 0.9006,
+      "step": 746
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870826546027223e-05,
+      "loss": 0.9669,
+      "step": 747
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704239682651116e-05,
+      "loss": 0.9493,
+      "step": 748
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700208076026204e-05,
+      "loss": 0.9757,
+      "step": 749
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696170643097355e-05,
+      "loss": 0.9623,
+      "step": 750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869212738656833e-05,
+      "loss": 0.9766,
+      "step": 751
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688078309146807e-05,
+      "loss": 1.0148,
+      "step": 752
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684023413544333e-05,
+      "loss": 0.8782,
+      "step": 753
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679962702476377e-05,
+      "loss": 1.026,
+      "step": 754
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675896178662282e-05,
+      "loss": 0.9036,
+      "step": 755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671823844825304e-05,
+      "loss": 0.9353,
+      "step": 756
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866774570369257e-05,
+      "loss": 1.0334,
+      "step": 757
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663661757995113e-05,
+      "loss": 0.9098,
+      "step": 758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659572010467843e-05,
+      "loss": 0.9583,
+      "step": 759
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865547646384956e-05,
+      "loss": 0.9583,
+      "step": 760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651375120882936e-05,
+      "loss": 0.9839,
+      "step": 761
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647267984314545e-05,
+      "loss": 0.9383,
+      "step": 762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864315505689483e-05,
+      "loss": 1.0026,
+      "step": 763
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86390363413781e-05,
+      "loss": 0.9684,
+      "step": 764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634911840522564e-05,
+      "loss": 0.9285,
+      "step": 765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630781557090292e-05,
+      "loss": 0.9226,
+      "step": 766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626645493847224e-05,
+      "loss": 0.9128,
+      "step": 767
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622503653563173e-05,
+      "loss": 0.8964,
+      "step": 768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618356039011828e-05,
+      "loss": 0.8563,
+      "step": 769
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614202652970734e-05,
+      "loss": 0.9623,
+      "step": 770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610043498221307e-05,
+      "loss": 0.8784,
+      "step": 771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860587857754883e-05,
+      "loss": 0.788,
+      "step": 772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8601707893742434e-05,
+      "loss": 0.8744,
+      "step": 773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597531449595125e-05,
+      "loss": 0.977,
+      "step": 774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593349247903756e-05,
+      "loss": 0.8745,
+      "step": 775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858916129146904e-05,
+      "loss": 0.9115,
+      "step": 776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858496758309554e-05,
+      "loss": 1.135,
+      "step": 777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858076812559168e-05,
+      "loss": 0.9357,
+      "step": 778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576562921769727e-05,
+      "loss": 0.8805,
+      "step": 779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572351974445792e-05,
+      "loss": 0.9889,
+      "step": 780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568135286439843e-05,
+      "loss": 0.9838,
+      "step": 781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856391286057568e-05,
+      "loss": 0.9184,
+      "step": 782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559684699680955e-05,
+      "loss": 0.9236,
+      "step": 783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855545080658716e-05,
+      "loss": 0.8754,
+      "step": 784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551211184129624e-05,
+      "loss": 0.8619,
+      "step": 785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546965835147504e-05,
+      "loss": 0.9381,
+      "step": 786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542714762483815e-05,
+      "loss": 0.9459,
+      "step": 787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538457968985378e-05,
+      "loss": 0.9195,
+      "step": 788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853419545750286e-05,
+      "loss": 0.9452,
+      "step": 789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529927230890757e-05,
+      "loss": 0.9562,
+      "step": 790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852565329200739e-05,
+      "loss": 1.0372,
+      "step": 791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521373643714904e-05,
+      "loss": 0.8632,
+      "step": 792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851708828887927e-05,
+      "loss": 0.8921,
+      "step": 793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512797230370273e-05,
+      "loss": 0.94,
+      "step": 794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508500471061528e-05,
+      "loss": 0.9858,
+      "step": 795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504198013830465e-05,
+      "loss": 0.9192,
+      "step": 796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499889861558328e-05,
+      "loss": 0.9937,
+      "step": 797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849557601713017e-05,
+      "loss": 1.0403,
+      "step": 798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491256483434866e-05,
+      "loss": 0.846,
+      "step": 799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486931263365097e-05,
+      "loss": 0.9122,
+      "step": 800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482600359817344e-05,
+      "loss": 0.8773,
+      "step": 801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478263775691907e-05,
+      "loss": 0.89,
+      "step": 802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473921513892884e-05,
+      "loss": 1.014,
+      "step": 803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846957357732817e-05,
+      "loss": 0.9986,
+      "step": 804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465219968909468e-05,
+      "loss": 0.9712,
+      "step": 805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846086069155228e-05,
+      "loss": 1.015,
+      "step": 806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456495748175896e-05,
+      "loss": 0.977,
+      "step": 807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845212514170341e-05,
+      "loss": 1.002,
+      "step": 808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447748875061702e-05,
+      "loss": 0.9413,
+      "step": 809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443366951181442e-05,
+      "loss": 0.8742,
+      "step": 810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438979372997098e-05,
+      "loss": 0.9596,
+      "step": 811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843458614344691e-05,
+      "loss": 0.8343,
+      "step": 812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843018726547291e-05,
+      "loss": 0.9725,
+      "step": 813
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425782742020916e-05,
+      "loss": 0.895,
+      "step": 814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421372576040524e-05,
+      "loss": 0.9387,
+      "step": 815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416956770485106e-05,
+      "loss": 0.8353,
+      "step": 816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.822,
+      "step": 817
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408108252481568e-05,
+      "loss": 0.87,
+      "step": 818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403675545959077e-05,
+      "loss": 0.8805,
+      "step": 819
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399237211712798e-05,
+      "loss": 0.9459,
+      "step": 820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394793252714976e-05,
+      "loss": 0.8933,
+      "step": 821
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390343671941608e-05,
+      "loss": 1.0439,
+      "step": 822
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385888472372474e-05,
+      "loss": 0.8611,
+      "step": 823
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381427656991094e-05,
+      "loss": 0.9741,
+      "step": 824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837696122878477e-05,
+      "loss": 0.9758,
+      "step": 825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372489190744546e-05,
+      "loss": 1.0536,
+      "step": 826
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836801154586523e-05,
+      "loss": 0.9328,
+      "step": 827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363528297145393e-05,
+      "loss": 0.9494,
+      "step": 828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359039447587346e-05,
+      "loss": 0.8936,
+      "step": 829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835454500019715e-05,
+      "loss": 0.8844,
+      "step": 830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350044957984627e-05,
+      "loss": 0.8328,
+      "step": 831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834553932396334e-05,
+      "loss": 0.9338,
+      "step": 832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341028101150583e-05,
+      "loss": 1.0228,
+      "step": 833
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833651129256742e-05,
+      "loss": 0.8585,
+      "step": 834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833198890123863e-05,
+      "loss": 1.0294,
+      "step": 835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832746093019275e-05,
+      "loss": 0.9114,
+      "step": 836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322927382462038e-05,
+      "loss": 0.935,
+      "step": 837
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318388261082495e-05,
+      "loss": 1.0116,
+      "step": 838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313843569093853e-05,
+      "loss": 0.9204,
+      "step": 839
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830929330953958e-05,
+      "loss": 0.8373,
+      "step": 840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830473748546686e-05,
+      "loss": 0.8432,
+      "step": 841
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830017609992661e-05,
+      "loss": 0.9973,
+      "step": 842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295609155973476e-05,
+      "loss": 1.0305,
+      "step": 843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291036656665823e-05,
+      "loss": 0.9986,
+      "step": 844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828645860506573e-05,
+      "loss": 0.9169,
+      "step": 845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281875004239004e-05,
+      "loss": 0.9215,
+      "step": 846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827728585725516e-05,
+      "loss": 0.93,
+      "step": 847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272691167187435e-05,
+      "loss": 0.8963,
+      "step": 848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826809093711277e-05,
+      "loss": 0.8922,
+      "step": 849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826348517011182e-05,
+      "loss": 0.8271,
+      "step": 850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258873869268953e-05,
+      "loss": 0.8621,
+      "step": 851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254257037672232e-05,
+      "loss": 0.9062,
+      "step": 852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249634678413433e-05,
+      "loss": 0.9521,
+      "step": 853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245006794588027e-05,
+      "loss": 0.9173,
+      "step": 854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240373389295192e-05,
+      "loss": 0.8995,
+      "step": 855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235734465637794e-05,
+      "loss": 0.9163,
+      "step": 856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231090026722407e-05,
+      "loss": 0.9221,
+      "step": 857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226440075659285e-05,
+      "loss": 1.0205,
+      "step": 858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221784615562382e-05,
+      "loss": 0.9989,
+      "step": 859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821712364954934e-05,
+      "loss": 0.8204,
+      "step": 860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821245718074149e-05,
+      "loss": 0.945,
+      "step": 861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820778521226384e-05,
+      "loss": 0.9243,
+      "step": 862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203107747245088e-05,
+      "loss": 0.9791,
+      "step": 863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819842478881761e-05,
+      "loss": 0.9064,
+      "step": 864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193736340117464e-05,
+      "loss": 0.8714,
+      "step": 865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818904240428438e-05,
+      "loss": 0.965,
+      "step": 866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184342984461766e-05,
+      "loss": 0.9142,
+      "step": 867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179638083796707e-05,
+      "loss": 0.9631,
+      "step": 868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174927705439947e-05,
+      "loss": 0.9387,
+      "step": 869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170211852545906e-05,
+      "loss": 0.8621,
+      "step": 870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816549052827267e-05,
+      "loss": 0.9189,
+      "step": 871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160763735781983e-05,
+      "loss": 0.9468,
+      "step": 872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815603147823926e-05,
+      "loss": 0.9186,
+      "step": 873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151293758813572e-05,
+      "loss": 0.9312,
+      "step": 874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814655058067764e-05,
+      "loss": 0.9283,
+      "step": 875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141801947007855e-05,
+      "loss": 1.0742,
+      "step": 876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137047860984252e-05,
+      "loss": 0.926,
+      "step": 877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132288325790518e-05,
+      "loss": 0.9187,
+      "step": 878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812752334461399e-05,
+      "loss": 0.9786,
+      "step": 879
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122752920645657e-05,
+      "loss": 0.9219,
+      "step": 880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811797705708014e-05,
+      "loss": 0.902,
+      "step": 881
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811319575711572e-05,
+      "loss": 1.0024,
+      "step": 882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108409023954297e-05,
+      "loss": 0.9657,
+      "step": 883
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103616860801442e-05,
+      "loss": 0.8897,
+      "step": 884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098819270866322e-05,
+      "loss": 1.0176,
+      "step": 885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094016257361776e-05,
+      "loss": 0.9877,
+      "step": 886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808920782350424e-05,
+      "loss": 1.0839,
+      "step": 887
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8084393972513818e-05,
+      "loss": 0.9583,
+      "step": 888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079574707614202e-05,
+      "loss": 0.964,
+      "step": 889
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074750032032738e-05,
+      "loss": 0.8968,
+      "step": 890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806991994900039e-05,
+      "loss": 0.9617,
+      "step": 891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806508446175173e-05,
+      "loss": 0.9666,
+      "step": 892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806024357352497e-05,
+      "loss": 0.9796,
+      "step": 893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055397287561914e-05,
+      "loss": 0.9559,
+      "step": 894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050545607108005e-05,
+      "loss": 0.9686,
+      "step": 895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045688535412283e-05,
+      "loss": 0.8449,
+      "step": 896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80408260757274e-05,
+      "loss": 0.988,
+      "step": 897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803595823130963e-05,
+      "loss": 0.9725,
+      "step": 898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031085005418832e-05,
+      "loss": 0.8803,
+      "step": 899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802620640131848e-05,
+      "loss": 1.0452,
+      "step": 900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021322422275655e-05,
+      "loss": 0.8775,
+      "step": 901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016433071561026e-05,
+      "loss": 0.9099,
+      "step": 902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011538352448867e-05,
+      "loss": 1.0608,
+      "step": 903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006638268217044e-05,
+      "loss": 0.8666,
+      "step": 904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001732822147013e-05,
+      "loss": 0.8489,
+      "step": 905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996822017523828e-05,
+      "loss": 1.0228,
+      "step": 906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991905857636126e-05,
+      "loss": 0.994,
+      "step": 907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7986984345776134e-05,
+      "loss": 0.9066,
+      "step": 908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982057485239664e-05,
+      "loss": 0.9314,
+      "step": 909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79771252793261e-05,
+      "loss": 0.9697,
+      "step": 910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797218773133841e-05,
+      "loss": 0.854,
+      "step": 911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967244844583152e-05,
+      "loss": 0.8551,
+      "step": 912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796229662237044e-05,
+      "loss": 1.0164,
+      "step": 913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957343068013977e-05,
+      "loss": 1.0465,
+      "step": 914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795238418483103e-05,
+      "loss": 0.9244,
+      "step": 915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947419976142428e-05,
+      "loss": 0.938,
+      "step": 916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942450445272575e-05,
+      "loss": 0.9276,
+      "step": 917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937475595549445e-05,
+      "loss": 1.0016,
+      "step": 918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932495430304556e-05,
+      "loss": 1.0442,
+      "step": 919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7927509952873003e-05,
+      "loss": 1.0545,
+      "step": 920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922519166593426e-05,
+      "loss": 0.9352,
+      "step": 921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917523074808024e-05,
+      "loss": 0.8757,
+      "step": 922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912521680862555e-05,
+      "loss": 0.8821,
+      "step": 923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790751498810632e-05,
+      "loss": 0.8796,
+      "step": 924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902502999892178e-05,
+      "loss": 0.9398,
+      "step": 925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789748571957652e-05,
+      "loss": 1.0154,
+      "step": 926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892463150519287e-05,
+      "loss": 0.737,
+      "step": 927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788743529608397e-05,
+      "loss": 0.9608,
+      "step": 928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882402159637588e-05,
+      "loss": 0.9447,
+      "step": 929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877363744550703e-05,
+      "loss": 0.9102,
+      "step": 930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872320054197416e-05,
+      "loss": 0.9581,
+      "step": 931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867271091955352e-05,
+      "loss": 1.0145,
+      "step": 932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786221686120567e-05,
+      "loss": 0.9331,
+      "step": 933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857157365333053e-05,
+      "loss": 0.8354,
+      "step": 934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852092607725722e-05,
+      "loss": 0.96,
+      "step": 935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847022591775412e-05,
+      "loss": 0.89,
+      "step": 936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784194732087738e-05,
+      "loss": 0.9181,
+      "step": 937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7836866798430402e-05,
+      "loss": 0.9119,
+      "step": 938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783178102783677e-05,
+      "loss": 1.0369,
+      "step": 939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7826690012502304e-05,
+      "loss": 0.8767,
+      "step": 940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782159375583631e-05,
+      "loss": 0.8637,
+      "step": 941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816492261251635e-05,
+      "loss": 0.999,
+      "step": 942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811385532164603e-05,
+      "loss": 0.8567,
+      "step": 943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7806273571995066e-05,
+      "loss": 0.8944,
+      "step": 944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780115638416637e-05,
+      "loss": 0.8495,
+      "step": 945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796033972105365e-05,
+      "loss": 0.9549,
+      "step": 946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790906339242397e-05,
+      "loss": 1.0772,
+      "step": 947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785773489011302e-05,
+      "loss": 0.9509,
+      "step": 948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780635424849425e-05,
+      "loss": 0.802,
+      "step": 949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777549215019759e-05,
+      "loss": 0.9226,
+      "step": 950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777034366850012e-05,
+      "loss": 0.9139,
+      "step": 951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765189983204814e-05,
+      "loss": 0.8335,
+      "step": 952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776003109776296e-05,
+      "loss": 0.8795,
+      "step": 953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775486701562934e-05,
+      "loss": 1.0147,
+      "step": 954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7749697740262197e-05,
+      "loss": 0.9443,
+      "step": 955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744523275123263e-05,
+      "loss": 0.9051,
+      "step": 956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739343623677748e-05,
+      "loss": 0.8985,
+      "step": 957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773415878939432e-05,
+      "loss": 0.929,
+      "step": 958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728968775745138e-05,
+      "loss": 0.898,
+      "step": 959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723773586205815e-05,
+      "loss": 0.9834,
+      "step": 960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771857322425544e-05,
+      "loss": 0.9471,
+      "step": 961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713367693376548e-05,
+      "loss": 0.9181,
+      "step": 962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770815699705516e-05,
+      "loss": 0.8341,
+      "step": 963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702941138780746e-05,
+      "loss": 0.9797,
+      "step": 964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769772012204622e-05,
+      "loss": 0.9654,
+      "step": 965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769249395034797e-05,
+      "loss": 0.9471,
+      "step": 966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687262627185826e-05,
+      "loss": 0.942,
+      "step": 967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768202615606307e-05,
+      "loss": 0.839,
+      "step": 968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676784540486427e-05,
+      "loss": 0.9783,
+      "step": 969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767153778396608e-05,
+      "loss": 0.925,
+      "step": 970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666285890015634e-05,
+      "loss": 0.8832,
+      "step": 971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661028862152156e-05,
+      "loss": 0.9967,
+      "step": 972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655766703896134e-05,
+      "loss": 0.8394,
+      "step": 973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765049941877151e-05,
+      "loss": 0.935,
+      "step": 974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645227010305635e-05,
+      "loss": 0.8496,
+      "step": 975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639949482029314e-05,
+      "loss": 0.8854,
+      "step": 976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7634666837476765e-05,
+      "loss": 0.9538,
+      "step": 977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762937908018564e-05,
+      "loss": 0.9815,
+      "step": 978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762408621369702e-05,
+      "loss": 0.9028,
+      "step": 979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618788241555387e-05,
+      "loss": 0.9407,
+      "step": 980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761348516730867e-05,
+      "loss": 0.9942,
+      "step": 981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760817699450819e-05,
+      "loss": 0.9912,
+      "step": 982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7602863726708698e-05,
+      "loss": 0.9611,
+      "step": 983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597545367468346e-05,
+      "loss": 0.7632,
+      "step": 984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7592221920348705e-05,
+      "loss": 0.9746,
+      "step": 985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586893388914758e-05,
+      "loss": 0.8674,
+      "step": 986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758155977673487e-05,
+      "loss": 0.9234,
+      "step": 987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757622108738083e-05,
+      "loss": 0.8566,
+      "step": 988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757087732442782e-05,
+      "loss": 0.9862,
+      "step": 989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756552849145442e-05,
+      "loss": 0.9194,
+      "step": 990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560174592042603e-05,
+      "loss": 0.9251,
+      "step": 991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755481562977774e-05,
+      "loss": 0.9741,
+      "step": 992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754945160824858e-05,
+      "loss": 0.9675,
+      "step": 993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544082531047284e-05,
+      "loss": 0.9696,
+      "step": 994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753870840176937e-05,
+      "loss": 0.9048,
+      "step": 995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533329224013763e-05,
+      "loss": 0.8881,
+      "step": 996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7527945001382756e-05,
+      "loss": 0.9126,
+      "step": 997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752255573748202e-05,
+      "loss": 1.0191,
+      "step": 998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7517161435920606e-05,
+      "loss": 0.9338,
+      "step": 999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751176210031094e-05,
+      "loss": 0.8297,
+      "step": 1000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750635773426882e-05,
+      "loss": 0.9299,
+      "step": 1001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500948341413402e-05,
+      "loss": 0.9103,
+      "step": 1002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495533925367224e-05,
+      "loss": 0.8967,
+      "step": 1003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749011448975618e-05,
+      "loss": 0.8996,
+      "step": 1004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484690038209522e-05,
+      "loss": 0.8634,
+      "step": 1005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479260574359863e-05,
+      "loss": 0.9287,
+      "step": 1006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747382610184318e-05,
+      "loss": 0.9054,
+      "step": 1007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746838662429879e-05,
+      "loss": 0.9242,
+      "step": 1008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746294214536938e-05,
+      "loss": 0.9833,
+      "step": 1009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457492668700967e-05,
+      "loss": 0.9643,
+      "step": 1010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452038197942935e-05,
+      "loss": 0.8654,
+      "step": 1011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446578736747994e-05,
+      "loss": 0.8709,
+      "step": 1012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441114288772207e-05,
+      "loss": 0.9152,
+      "step": 1013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435644857674967e-05,
+      "loss": 0.8979,
+      "step": 1014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743017044711902e-05,
+      "loss": 0.8718,
+      "step": 1015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742469106077043e-05,
+      "loss": 0.9364,
+      "step": 1016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419206702298602e-05,
+      "loss": 0.9539,
+      "step": 1017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741371737537627e-05,
+      "loss": 0.9609,
+      "step": 1018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408223083679493e-05,
+      "loss": 0.9262,
+      "step": 1019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402723830887657e-05,
+      "loss": 0.9171,
+      "step": 1020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397219620683465e-05,
+      "loss": 0.9504,
+      "step": 1021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739171045675295e-05,
+      "loss": 1.0122,
+      "step": 1022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738619634278545e-05,
+      "loss": 0.8665,
+      "step": 1023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738067728247363e-05,
+      "loss": 1.0275,
+      "step": 1024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737515327951346e-05,
+      "loss": 0.9405,
+      "step": 1025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369624337604215e-05,
+      "loss": 0.8148,
+      "step": 1026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364090460448497e-05,
+      "loss": 0.8279,
+      "step": 1027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358551651752187e-05,
+      "loss": 0.8665,
+      "step": 1028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7353007915224497e-05,
+      "loss": 0.924,
+      "step": 1029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347459254577908e-05,
+      "loss": 0.8001,
+      "step": 1030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734190567352822e-05,
+      "loss": 0.96,
+      "step": 1031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336347175794523e-05,
+      "loss": 1.045,
+      "step": 1032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330783765099203e-05,
+      "loss": 1.0026,
+      "step": 1033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325215445167926e-05,
+      "loss": 0.9193,
+      "step": 1034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319642219729655e-05,
+      "loss": 0.883,
+      "step": 1035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731406409251663e-05,
+      "loss": 0.9882,
+      "step": 1036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308481067264384e-05,
+      "loss": 0.9148,
+      "step": 1037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7302893147711722e-05,
+      "loss": 0.9277,
+      "step": 1038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297300337600727e-05,
+      "loss": 0.9317,
+      "step": 1039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291702640676763e-05,
+      "loss": 0.925,
+      "step": 1040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728610006068846e-05,
+      "loss": 0.8742,
+      "step": 1041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728049260138772e-05,
+      "loss": 0.8935,
+      "step": 1042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7274880266529716e-05,
+      "loss": 0.8589,
+      "step": 1043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7269263059872883e-05,
+      "loss": 0.9588,
+      "step": 1044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7263640985178916e-05,
+      "loss": 1.0371,
+      "step": 1045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258014046212774e-05,
+      "loss": 0.9375,
+      "step": 1046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7252382246742675e-05,
+      "loss": 0.9794,
+      "step": 1047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7246745590540083e-05,
+      "loss": 0.9934,
+      "step": 1048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7241104081379725e-05,
+      "loss": 1.0111,
+      "step": 1049
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235457723039576e-05,
+      "loss": 0.8417,
+      "step": 1050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7229806519300845e-05,
+      "loss": 0.8985,
+      "step": 1051
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7224150473948012e-05,
+      "loss": 0.9276,
+      "step": 1052
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721848959076877e-05,
+      "loss": 0.9187,
+      "step": 1053
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721282387355408e-05,
+      "loss": 0.9525,
+      "step": 1054
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207153326098115e-05,
+      "loss": 0.8522,
+      "step": 1055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201477952198296e-05,
+      "loss": 0.9257,
+      "step": 1056
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195797755655274e-05,
+      "loss": 0.9099,
+      "step": 1057
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719011274027293e-05,
+      "loss": 1.0047,
+      "step": 1058
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184422909858375e-05,
+      "loss": 1.0144,
+      "step": 1059
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7178728268221933e-05,
+      "loss": 0.8621,
+      "step": 1060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173028819177167e-05,
+      "loss": 0.8386,
+      "step": 1061
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167324566540843e-05,
+      "loss": 0.958,
+      "step": 1062
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161615514132956e-05,
+      "loss": 0.8713,
+      "step": 1063
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715590166577671e-05,
+      "loss": 0.9462,
+      "step": 1064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715018302529852e-05,
+      "loss": 0.946,
+      "step": 1065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144459596528008e-05,
+      "loss": 1.0506,
+      "step": 1066
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138731383298015e-05,
+      "loss": 0.945,
+      "step": 1067
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132998389444564e-05,
+      "loss": 0.8079,
+      "step": 1068
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127260618806908e-05,
+      "loss": 0.9101,
+      "step": 1069
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712151807522747e-05,
+      "loss": 0.8885,
+      "step": 1070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7115770762551888e-05,
+      "loss": 0.9396,
+      "step": 1071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711001868462899e-05,
+      "loss": 0.9675,
+      "step": 1072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710426184531079e-05,
+      "loss": 0.9194,
+      "step": 1073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7098500248452496e-05,
+      "loss": 0.8886,
+      "step": 1074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70927338979125e-05,
+      "loss": 0.9357,
+      "step": 1075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086962797552376e-05,
+      "loss": 0.9086,
+      "step": 1076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081186951236882e-05,
+      "loss": 0.8857,
+      "step": 1077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075406362833958e-05,
+      "loss": 0.9453,
+      "step": 1078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70696210362147e-05,
+      "loss": 0.8814,
+      "step": 1079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706383097525341e-05,
+      "loss": 1.0207,
+      "step": 1080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058036183827525e-05,
+      "loss": 1.0282,
+      "step": 1081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705223666581768e-05,
+      "loss": 0.8923,
+      "step": 1082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704643242510765e-05,
+      "loss": 0.952,
+      "step": 1083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704062346558439e-05,
+      "loss": 0.9379,
+      "step": 1084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034809791138012e-05,
+      "loss": 0.842,
+      "step": 1085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7028991405661783e-05,
+      "loss": 0.9571,
+      "step": 1086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7023168313052118e-05,
+      "loss": 0.9134,
+      "step": 1087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7017340517208604e-05,
+      "loss": 0.825,
+      "step": 1088
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701150802203395e-05,
+      "loss": 0.8672,
+      "step": 1089
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005670831434036e-05,
+      "loss": 0.8662,
+      "step": 1090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999828949317877e-05,
+      "loss": 0.863,
+      "step": 1091
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6993982379597627e-05,
+      "loss": 0.9376,
+      "step": 1092
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988131126188582e-05,
+      "loss": 1.0558,
+      "step": 1093
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982275193009183e-05,
+      "loss": 0.9412,
+      "step": 1094
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976414583980984e-05,
+      "loss": 0.8665,
+      "step": 1095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697054930302869e-05,
+      "loss": 0.9361,
+      "step": 1096
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696467935408013e-05,
+      "loss": 0.9032,
+      "step": 1097
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958804741066254e-05,
+      "loss": 0.9417,
+      "step": 1098
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695292546792114e-05,
+      "loss": 0.9089,
+      "step": 1099
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6947041538581977e-05,
+      "loss": 0.8171,
+      "step": 1100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941152956989088e-05,
+      "loss": 0.9296,
+      "step": 1101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69352597270859e-05,
+      "loss": 0.9175,
+      "step": 1102
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692936185281896e-05,
+      "loss": 1.0023,
+      "step": 1103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923459338137917e-05,
+      "loss": 0.9671,
+      "step": 1104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6917552186995538e-05,
+      "loss": 0.8421,
+      "step": 1105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6911640403347686e-05,
+      "loss": 0.9166,
+      "step": 1106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905723991153326e-05,
+      "loss": 0.9108,
+      "step": 1107
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6899802954374528e-05,
+      "loss": 1.0226,
+      "step": 1108
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689387729697646e-05,
+      "loss": 0.9603,
+      "step": 1109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688794702292738e-05,
+      "loss": 0.8759,
+      "step": 1110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882012136198637e-05,
+      "loss": 0.8865,
+      "step": 1111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687607264076467e-05,
+      "loss": 0.9618,
+      "step": 1112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870128540603006e-05,
+      "loss": 0.9328,
+      "step": 1113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864179839694257e-05,
+      "loss": 0.9654,
+      "step": 1114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858226542022117e-05,
+      "loss": 0.938,
+      "step": 1115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852268651573346e-05,
+      "loss": 0.9513,
+      "step": 1116
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846306172337796e-05,
+      "loss": 0.9649,
+      "step": 1117
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684033910830838e-05,
+      "loss": 0.8958,
+      "step": 1118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6834367463481093e-05,
+      "loss": 0.9022,
+      "step": 1119
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828391241854983e-05,
+      "loss": 0.9747,
+      "step": 1120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822410447432175e-05,
+      "loss": 0.8397,
+      "step": 1121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6816425084217856e-05,
+      "loss": 0.9215,
+      "step": 1122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810435156220258e-05,
+      "loss": 0.8946,
+      "step": 1123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680444066745069e-05,
+      "loss": 0.9521,
+      "step": 1124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798441621923503e-05,
+      "loss": 0.8654,
+      "step": 1125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.67924380236561e-05,
+      "loss": 0.932,
+      "step": 1126
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678642987666894e-05,
+      "loss": 0.9916,
+      "step": 1127
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6780417184985517e-05,
+      "loss": 0.9553,
+      "step": 1128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774399952632383e-05,
+      "loss": 0.9319,
+      "step": 1129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768378183639115e-05,
+      "loss": 0.9149,
+      "step": 1130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762351882038342e-05,
+      "loss": 0.8964,
+      "step": 1131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756321051865713e-05,
+      "loss": 0.851,
+      "step": 1132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675028569715993e-05,
+      "loss": 1.0092,
+      "step": 1133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744245821962704e-05,
+      "loss": 0.9844,
+      "step": 1134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6738201430318786e-05,
+      "loss": 0.8813,
+      "step": 1135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732152526275954e-05,
+      "loss": 0.8766,
+      "step": 1136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726099113884996e-05,
+      "loss": 0.8962,
+      "step": 1137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6720041197199727e-05,
+      "loss": 0.8883,
+      "step": 1138
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713978780276976e-05,
+      "loss": 0.8654,
+      "step": 1139
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670791186717659e-05,
+      "loss": 0.9435,
+      "step": 1140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670184046196142e-05,
+      "loss": 0.9082,
+      "step": 1141
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669576456869733e-05,
+      "loss": 0.8732,
+      "step": 1142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668968419145319e-05,
+      "loss": 0.9836,
+      "step": 1143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668359933430087e-05,
+      "loss": 0.9472,
+      "step": 1144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677510001315237e-05,
+      "loss": 0.9156,
+      "step": 1145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671416196574163e-05,
+      "loss": 0.8097,
+      "step": 1146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665317924158513e-05,
+      "loss": 1.0016,
+      "step": 1147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665921518815214e-05,
+      "loss": 1.1316,
+      "step": 1148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665310799264189e-05,
+      "loss": 0.9693,
+      "step": 1149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646996341717588e-05,
+      "loss": 0.8843,
+      "step": 1150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640880239472058e-05,
+      "loss": 0.8066,
+      "step": 1151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634759690001082e-05,
+      "loss": 1.0056,
+      "step": 1152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628634697403447e-05,
+      "loss": 0.8474,
+      "step": 1153
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6622505265780896e-05,
+      "loss": 0.8691,
+      "step": 1154
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616371399238148e-05,
+      "loss": 0.903,
+      "step": 1155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610233101882894e-05,
+      "loss": 0.9281,
+      "step": 1156
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6604090377825804e-05,
+      "loss": 0.8899,
+      "step": 1157
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597943231180486e-05,
+      "loss": 1.0569,
+      "step": 1158
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6591791666063533e-05,
+      "loss": 0.901,
+      "step": 1159
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6585635686594487e-05,
+      "loss": 1.0731,
+      "step": 1160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579475296895847e-05,
+      "loss": 0.9878,
+      "step": 1161
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573310501093074e-05,
+      "loss": 0.9571,
+      "step": 1162
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656714130331456e-05,
+      "loss": 0.8544,
+      "step": 1163
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560967707691663e-05,
+      "loss": 0.8676,
+      "step": 1164
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655478971835868e-05,
+      "loss": 0.8981,
+      "step": 1165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.8897,
+      "step": 1166
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542420575114352e-05,
+      "loss": 1.0359,
+      "step": 1167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536229429486303e-05,
+      "loss": 0.9961,
+      "step": 1168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530033906714753e-05,
+      "loss": 0.879,
+      "step": 1169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652383401094868e-05,
+      "loss": 0.8674,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517629746339997e-05,
+      "loss": 0.9246,
+      "step": 1171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511421117043534e-05,
+      "loss": 1.0032,
+      "step": 1172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6505208127217055e-05,
+      "loss": 0.9122,
+      "step": 1173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498990781021232e-05,
+      "loss": 0.8587,
+      "step": 1174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649276908261967e-05,
+      "loss": 0.9976,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6486543036178874e-05,
+      "loss": 0.9031,
+      "step": 1176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480312645868265e-05,
+      "loss": 0.8551,
+      "step": 1177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647407791586017e-05,
+      "loss": 0.8973,
+      "step": 1178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467838850329836e-05,
+      "loss": 0.8917,
+      "step": 1179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461595453455397e-05,
+      "loss": 0.9252,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6455347729417893e-05,
+      "loss": 0.8705,
+      "step": 1181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644909568240127e-05,
+      "loss": 0.9747,
+      "step": 1182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644283931659236e-05,
+      "loss": 1.0201,
+      "step": 1183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643657863618088e-05,
+      "loss": 0.9465,
+      "step": 1184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430313645359454e-05,
+      "loss": 0.7953,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642404434832358e-05,
+      "loss": 0.9192,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417770749271652e-05,
+      "loss": 0.9194,
+      "step": 1187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411492852404927e-05,
+      "loss": 0.9221,
+      "step": 1188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640521066192755e-05,
+      "loss": 0.9353,
+      "step": 1189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398924182046553e-05,
+      "loss": 0.9127,
+      "step": 1190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392633416971816e-05,
+      "loss": 0.9832,
+      "step": 1191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386338370916104e-05,
+      "loss": 0.8399,
+      "step": 1192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638003904809505e-05,
+      "loss": 0.8258,
+      "step": 1193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6373735452727145e-05,
+      "loss": 0.89,
+      "step": 1194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6367427589033744e-05,
+      "loss": 0.9194,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6361115461239058e-05,
+      "loss": 0.8542,
+      "step": 1196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635479907357016e-05,
+      "loss": 0.9095,
+      "step": 1197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348478430256966e-05,
+      "loss": 0.9004,
+      "step": 1198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6342153535532253e-05,
+      "loss": 0.938,
+      "step": 1199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335824393631633e-05,
+      "loss": 0.835,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632949100879357e-05,
+      "loss": 0.9003,
+      "step": 1201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6323153385259368e-05,
+      "loss": 0.9475,
+      "step": 1202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631681152727317e-05,
+      "loss": 0.9678,
+      "step": 1203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6310465439081958e-05,
+      "loss": 0.8745,
+      "step": 1204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304115124935537e-05,
+      "loss": 0.9428,
+      "step": 1205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297760589086545e-05,
+      "loss": 0.7988,
+      "step": 1206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629140183579046e-05,
+      "loss": 0.8698,
+      "step": 1207
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285038869305565e-05,
+      "loss": 0.8663,
+      "step": 1208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627867169389298e-05,
+      "loss": 0.9254,
+      "step": 1209
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6272300313816634e-05,
+      "loss": 1.0048,
+      "step": 1210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626592473334328e-05,
+      "loss": 0.8083,
+      "step": 1211
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625954495674247e-05,
+      "loss": 0.8986,
+      "step": 1212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625316098828658e-05,
+      "loss": 0.9067,
+      "step": 1213
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624677283225079e-05,
+      "loss": 0.9335,
+      "step": 1214
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6240380492913087e-05,
+      "loss": 0.9652,
+      "step": 1215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6233983974554246e-05,
+      "loss": 0.8462,
+      "step": 1216
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622758328145785e-05,
+      "loss": 0.8935,
+      "step": 1217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221178417910284e-05,
+      "loss": 0.8778,
+      "step": 1218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621476938820071e-05,
+      "loss": 0.9491,
+      "step": 1219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62083561966211e-05,
+      "loss": 0.953,
+      "step": 1220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6201938847466193e-05,
+      "loss": 0.8266,
+      "step": 1221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619551734503353e-05,
+      "loss": 0.9644,
+      "step": 1222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189091693623418e-05,
+      "loss": 0.9244,
+      "step": 1223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182661897538947e-05,
+      "loss": 0.8459,
+      "step": 1224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6176227961085995e-05,
+      "loss": 0.9122,
+      "step": 1225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169789888573197e-05,
+      "loss": 0.8755,
+      "step": 1226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616334768431196e-05,
+      "loss": 0.8557,
+      "step": 1227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6156901352616462e-05,
+      "loss": 0.8476,
+      "step": 1228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6150450897803652e-05,
+      "loss": 0.8925,
+      "step": 1229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6143996324193227e-05,
+      "loss": 0.8671,
+      "step": 1230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137537636107642e-05,
+      "loss": 0.8884,
+      "step": 1231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131074837872124e-05,
+      "loss": 0.9483,
+      "step": 1232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124607933814638e-05,
+      "loss": 0.846,
+      "step": 1233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118136928265896e-05,
+      "loss": 0.8923,
+      "step": 1234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611166182555937e-05,
+      "loss": 0.9218,
+      "step": 1235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105182630031265e-05,
+      "loss": 0.9321,
+      "step": 1236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098699346020534e-05,
+      "loss": 0.8839,
+      "step": 1237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092211977868856e-05,
+      "loss": 0.8947,
+      "step": 1238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085720529920667e-05,
+      "loss": 0.9186,
+      "step": 1239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079225006523108e-05,
+      "loss": 0.8689,
+      "step": 1240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6072725412026066e-05,
+      "loss": 0.8404,
+      "step": 1241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6066221750782157e-05,
+      "loss": 0.8838,
+      "step": 1242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059714027146708e-05,
+      "loss": 0.873,
+      "step": 1243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605320224547778e-05,
+      "loss": 0.9127,
+      "step": 1244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6046686410136136e-05,
+      "loss": 0.8097,
+      "step": 1245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604016652548527e-05,
+      "loss": 0.8405,
+      "step": 1246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603364259589137e-05,
+      "loss": 0.9944,
+      "step": 1247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6027114625723353e-05,
+      "loss": 0.8257,
+      "step": 1248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020582619352827e-05,
+      "loss": 0.962,
+      "step": 1249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6014046581154108e-05,
+      "loss": 0.9242,
+      "step": 1250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600750651550421e-05,
+      "loss": 1.0372,
+      "step": 1251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000962426782844e-05,
+      "loss": 0.8681,
+      "step": 1252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5994414319372423e-05,
+      "loss": 0.9658,
+      "step": 1253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598786219765804e-05,
+      "loss": 0.9037,
+      "step": 1254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598130606602748e-05,
+      "loss": 0.9743,
+      "step": 1255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597474592887122e-05,
+      "loss": 0.9002,
+      "step": 1256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5968181790582408e-05,
+      "loss": 0.9254,
+      "step": 1257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596161365555688e-05,
+      "loss": 0.8292,
+      "step": 1258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595504152819315e-05,
+      "loss": 0.8896,
+      "step": 1259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948465412892393e-05,
+      "loss": 0.8623,
+      "step": 1260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5941885314058464e-05,
+      "loss": 0.9321,
+      "step": 1261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935301236097894e-05,
+      "loss": 0.9926,
+      "step": 1262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592871318341986e-05,
+      "loss": 0.9717,
+      "step": 1263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922121160436214e-05,
+      "loss": 0.8985,
+      "step": 1264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915525171561467e-05,
+      "loss": 0.9964,
+      "step": 1265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590892522121277e-05,
+      "loss": 0.9746,
+      "step": 1266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5902321313809948e-05,
+      "loss": 1.0236,
+      "step": 1267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589571345377547e-05,
+      "loss": 0.9774,
+      "step": 1268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588910164553444e-05,
+      "loss": 0.9143,
+      "step": 1269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882485893514617e-05,
+      "loss": 0.9403,
+      "step": 1270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58758662021464e-05,
+      "loss": 0.9975,
+      "step": 1271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586924257586283e-05,
+      "loss": 0.9704,
+      "step": 1272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862615019099564e-05,
+      "loss": 0.9463,
+      "step": 1273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585598353629492e-05,
+      "loss": 0.8812,
+      "step": 1274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849348131889822e-05,
+      "loss": 0.946,
+      "step": 1275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584270881032783e-05,
+      "loss": 0.9913,
+      "step": 1276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836065576055127e-05,
+      "loss": 0.9815,
+      "step": 1277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5829418433520513e-05,
+      "loss": 0.8639,
+      "step": 1278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582276738717541e-05,
+      "loss": 0.9325,
+      "step": 1279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816112441473842e-05,
+      "loss": 0.9119,
+      "step": 1280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5809453600872467e-05,
+      "loss": 1.0213,
+      "step": 1281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802790869830534e-05,
+      "loss": 0.8906,
+      "step": 1282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796124252809894e-05,
+      "loss": 0.9616,
+      "step": 1283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789453754275016e-05,
+      "loss": 0.9704,
+      "step": 1284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782779378692957e-05,
+      "loss": 0.9047,
+      "step": 1285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5776101130533373e-05,
+      "loss": 0.925,
+      "step": 1286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769419014268518e-05,
+      "loss": 0.9254,
+      "step": 1287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576273303437322e-05,
+      "loss": 0.9459,
+      "step": 1288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5756043195324924e-05,
+      "loss": 0.9434,
+      "step": 1289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5749349501603627e-05,
+      "loss": 1.0824,
+      "step": 1290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574265195769193e-05,
+      "loss": 0.9844,
+      "step": 1291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735950568075002e-05,
+      "loss": 0.8324,
+      "step": 1292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729245337240596e-05,
+      "loss": 0.8361,
+      "step": 1293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572253626967902e-05,
+      "loss": 0.8336,
+      "step": 1294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715823369883173e-05,
+      "loss": 0.885,
+      "step": 1295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57091066423485e-05,
+      "loss": 0.884,
+      "step": 1296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702386091573032e-05,
+      "loss": 0.8752,
+      "step": 1297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695661722057335e-05,
+      "loss": 0.8571,
+      "step": 1298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688933538304558e-05,
+      "loss": 0.9438,
+      "step": 1299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568220154482038e-05,
+      "loss": 0.8306,
+      "step": 1300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675465746113044e-05,
+      "loss": 0.9688,
+      "step": 1301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668726146693346e-05,
+      "loss": 0.9559,
+      "step": 1302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5661982751074622e-05,
+      "loss": 0.9346,
+      "step": 1303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565523556377274e-05,
+      "loss": 0.9397,
+      "step": 1304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648484589306125e-05,
+      "loss": 0.9767,
+      "step": 1305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641729832195728e-05,
+      "loss": 0.9295,
+      "step": 1306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5634971296965027e-05,
+      "loss": 0.8808,
+      "step": 1307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562820898814005e-05,
+      "loss": 0.8978,
+      "step": 1308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562144291024933e-05,
+      "loss": 0.9262,
+      "step": 1309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614673067823945e-05,
+      "loss": 0.9139,
+      "step": 1310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607899465397473e-05,
+      "loss": 0.8154,
+      "step": 1311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560112210750602e-05,
+      "loss": 0.8997,
+      "step": 1312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5594340998688213e-05,
+      "loss": 0.8915,
+      "step": 1313
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587556143485185e-05,
+      "loss": 0.9199,
+      "step": 1314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580767546440573e-05,
+      "loss": 0.9599,
+      "step": 1315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5573975212100528e-05,
+      "loss": 0.9999,
+      "step": 1316
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55671791450137e-05,
+      "loss": 0.9573,
+      "step": 1317
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560379349731234e-05,
+      "loss": 0.9477,
+      "step": 1318
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553575830806786e-05,
+      "loss": 0.8685,
+      "step": 1319
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546768592796487e-05,
+      "loss": 0.8181,
+      "step": 1320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539957640258973e-05,
+      "loss": 0.9919,
+      "step": 1321
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533142977755363e-05,
+      "loss": 0.9306,
+      "step": 1322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5526324609849254e-05,
+      "loss": 0.9625,
+      "step": 1323
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551950254110674e-05,
+      "loss": 0.8075,
+      "step": 1324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5512676776096376e-05,
+      "loss": 0.8304,
+      "step": 1325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.55058473193892e-05,
+      "loss": 0.941,
+      "step": 1326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549901417555873e-05,
+      "loss": 0.9718,
+      "step": 1327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5492177349180938e-05,
+      "loss": 0.9433,
+      "step": 1328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5485336844834274e-05,
+      "loss": 0.9239,
+      "step": 1329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478492667099638e-05,
+      "loss": 0.9208,
+      "step": 1330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547164482056041e-05,
+      "loss": 0.9916,
+      "step": 1331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5464793309802408e-05,
+      "loss": 0.9679,
+      "step": 1332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5457938139413916e-05,
+      "loss": 0.8016,
+      "step": 1333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545107931398565e-05,
+      "loss": 0.876,
+      "step": 1334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544421683811081e-05,
+      "loss": 0.913,
+      "step": 1335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543735071638501e-05,
+      "loss": 0.9551,
+      "step": 1336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430480953406305e-05,
+      "loss": 0.931,
+      "step": 1337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54236075537752e-05,
+      "loss": 0.8507,
+      "step": 1338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5416730522094642e-05,
+      "loss": 0.947,
+      "step": 1339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5409849862969994e-05,
+      "loss": 0.8655,
+      "step": 1340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5402965581009052e-05,
+      "loss": 0.9795,
+      "step": 1341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539607768082205e-05,
+      "loss": 0.9612,
+      "step": 1342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538918616702163e-05,
+      "loss": 0.8544,
+      "step": 1343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538229104422286e-05,
+      "loss": 0.963,
+      "step": 1344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375392317043233e-05,
+      "loss": 0.7957,
+      "step": 1345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368489990102636e-05,
+      "loss": 0.9216,
+      "step": 1346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5361584068023385e-05,
+      "loss": 0.9795,
+      "step": 1347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5354674555430196e-05,
+      "loss": 0.9971,
+      "step": 1348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534776145695019e-05,
+      "loss": 0.8431,
+      "step": 1349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5340844777212896e-05,
+      "loss": 1.0051,
+      "step": 1350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333924520850227e-05,
+      "loss": 0.8291,
+      "step": 1351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5327000692496503e-05,
+      "loss": 0.871,
+      "step": 1352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532007329678843e-05,
+      "loss": 0.8863,
+      "step": 1353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531314233836511e-05,
+      "loss": 0.9312,
+      "step": 1354
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306207821868024e-05,
+      "loss": 0.8881,
+      "step": 1355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299269751941037e-05,
+      "loss": 0.9453,
+      "step": 1356
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5292328133230398e-05,
+      "loss": 0.9143,
+      "step": 1357
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5285382970384717e-05,
+      "loss": 0.8687,
+      "step": 1358
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527843426805501e-05,
+      "loss": 0.8857,
+      "step": 1359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5271482030894627e-05,
+      "loss": 0.8458,
+      "step": 1360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526452626355931e-05,
+      "loss": 0.9794,
+      "step": 1361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257566970707147e-05,
+      "loss": 0.896,
+      "step": 1362
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5250604156998602e-05,
+      "loss": 1.0368,
+      "step": 1363
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243637827096498e-05,
+      "loss": 1.0455,
+      "step": 1364
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5236667985665992e-05,
+      "loss": 0.9873,
+      "step": 1365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229694637374613e-05,
+      "loss": 0.8605,
+      "step": 1366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522271778689223e-05,
+      "loss": 0.9562,
+      "step": 1367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521573743889106e-05,
+      "loss": 0.9265,
+      "step": 1368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5208753598045665e-05,
+      "loss": 0.8708,
+      "step": 1369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201766269032934e-05,
+      "loss": 1.0098,
+      "step": 1370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194775456532107e-05,
+      "loss": 0.836,
+      "step": 1371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187781165224746e-05,
+      "loss": 0.9277,
+      "step": 1372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180783399794749e-05,
+      "loss": 0.835,
+      "step": 1373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517378216492834e-05,
+      "loss": 0.8315,
+      "step": 1374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166777465314062e-05,
+      "loss": 0.9259,
+      "step": 1375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159769305642776e-05,
+      "loss": 0.7811,
+      "step": 1376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152757690607674e-05,
+      "loss": 0.9696,
+      "step": 1377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145742624904252e-05,
+      "loss": 0.9022,
+      "step": 1378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513872411323031e-05,
+      "loss": 0.8358,
+      "step": 1379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131702160285966e-05,
+      "loss": 0.9834,
+      "step": 1380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124676770773645e-05,
+      "loss": 0.8953,
+      "step": 1381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117647949398066e-05,
+      "loss": 0.9195,
+      "step": 1382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110615700866248e-05,
+      "loss": 0.931,
+      "step": 1383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103580029887504e-05,
+      "loss": 0.8529,
+      "step": 1384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096540941173446e-05,
+      "loss": 0.9095,
+      "step": 1385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089498439437962e-05,
+      "loss": 0.9811,
+      "step": 1386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082452529397236e-05,
+      "loss": 1.0183,
+      "step": 1387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075403215769734e-05,
+      "loss": 1.0085,
+      "step": 1388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068350503276194e-05,
+      "loss": 0.932,
+      "step": 1389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061294396639637e-05,
+      "loss": 0.8947,
+      "step": 1390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054234900585359e-05,
+      "loss": 0.8973,
+      "step": 1391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047172019840917e-05,
+      "loss": 0.9251,
+      "step": 1392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040105759136139e-05,
+      "loss": 0.9358,
+      "step": 1393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5033036123203116e-05,
+      "loss": 0.9991,
+      "step": 1394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5025963116776203e-05,
+      "loss": 0.8728,
+      "step": 1395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018886744592008e-05,
+      "loss": 0.941,
+      "step": 1396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5011807011389395e-05,
+      "loss": 0.8699,
+      "step": 1397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5004723921909476e-05,
+      "loss": 1.0477,
+      "step": 1398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4997637480895613e-05,
+      "loss": 0.8583,
+      "step": 1399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990547693093413e-05,
+      "loss": 0.9376,
+      "step": 1400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498345456325072e-05,
+      "loss": 0.8838,
+      "step": 1401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976358096117627e-05,
+      "loss": 0.8944,
+      "step": 1402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4969258296446447e-05,
+      "loss": 0.884,
+      "step": 1403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496215516899173e-05,
+      "loss": 0.9689,
+      "step": 1404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4955048718510263e-05,
+      "loss": 0.9669,
+      "step": 1405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947938949761054e-05,
+      "loss": 0.8718,
+      "step": 1406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494082586750532e-05,
+      "loss": 0.8479,
+      "step": 1407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933709476506515e-05,
+      "loss": 0.8385,
+      "step": 1408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49265897815303e-05,
+      "loss": 0.8729,
+      "step": 1409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4919466787344552e-05,
+      "loss": 0.9684,
+      "step": 1410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4912340498719358e-05,
+      "loss": 0.774,
+      "step": 1411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4905210920427e-05,
+      "loss": 0.9078,
+      "step": 1412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4898078057241979e-05,
+      "loss": 0.8672,
+      "step": 1413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890941913940987e-05,
+      "loss": 0.8692,
+      "step": 1414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883802495302914e-05,
+      "loss": 0.913,
+      "step": 1415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4876659806108844e-05,
+      "loss": 0.894,
+      "step": 1416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869513851142051e-05,
+      "loss": 0.7948,
+      "step": 1417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4862364635187996e-05,
+      "loss": 0.9475,
+      "step": 1418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485521216303432e-05,
+      "loss": 0.918,
+      "step": 1419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848056439470857e-05,
+      "loss": 0.8309,
+      "step": 1420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840897469289601e-05,
+      "loss": 0.9303,
+      "step": 1421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833735257284732e-05,
+      "loss": 0.9763,
+      "step": 1422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48265698082526e-05,
+      "loss": 0.8938,
+      "step": 1423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4819401126991716e-05,
+      "loss": 0.9316,
+      "step": 1424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4812229218302767e-05,
+      "loss": 0.8763,
+      "step": 1425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805054086988585e-05,
+      "loss": 0.9849,
+      "step": 1426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797875737854177e-05,
+      "loss": 0.9095,
+      "step": 1427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790694175706698e-05,
+      "loss": 0.894,
+      "step": 1428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783509405355445e-05,
+      "loss": 0.8576,
+      "step": 1429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776321431611881e-05,
+      "loss": 0.8722,
+      "step": 1430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769130259289605e-05,
+      "loss": 1.0558,
+      "step": 1431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761935893204358e-05,
+      "loss": 0.9353,
+      "step": 1432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4754738338174016e-05,
+      "loss": 0.8349,
+      "step": 1433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747537599018604e-05,
+      "loss": 0.8887,
+      "step": 1434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740333680560264e-05,
+      "loss": 0.9289,
+      "step": 1435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733126587623272e-05,
+      "loss": 0.9336,
+      "step": 1436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725916325034036e-05,
+      "loss": 0.8973,
+      "step": 1437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718702897621083e-05,
+      "loss": 0.9489,
+      "step": 1438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711486310215053e-05,
+      "loss": 0.9232,
+      "step": 1439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470426656764871e-05,
+      "loss": 0.8694,
+      "step": 1440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469704367475693e-05,
+      "loss": 0.8283,
+      "step": 1441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689817636376698e-05,
+      "loss": 0.8437,
+      "step": 1442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.46825884573471e-05,
+      "loss": 0.8721,
+      "step": 1443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675356142509336e-05,
+      "loss": 0.8039,
+      "step": 1444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4668120696706694e-05,
+      "loss": 0.8988,
+      "step": 1445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660882124784568e-05,
+      "loss": 0.7792,
+      "step": 1446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653640431590442e-05,
+      "loss": 0.8251,
+      "step": 1447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464639562197389e-05,
+      "loss": 0.9398,
+      "step": 1448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639147700786573e-05,
+      "loss": 0.817,
+      "step": 1449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4631896672882235e-05,
+      "loss": 0.8555,
+      "step": 1450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624642543116703e-05,
+      "loss": 0.8217,
+      "step": 1451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617385316347879e-05,
+      "loss": 0.8635,
+      "step": 1452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4610124997435743e-05,
+      "loss": 0.874,
+      "step": 1453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602861591242335e-05,
+      "loss": 0.8818,
+      "step": 1454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595595102631775e-05,
+      "loss": 0.9304,
+      "step": 1455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4588325536470242e-05,
+      "loss": 0.882,
+      "step": 1456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581052897625978e-05,
+      "loss": 0.7676,
+      "step": 1457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573777190969276e-05,
+      "loss": 0.8685,
+      "step": 1458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566498421372492e-05,
+      "loss": 0.8537,
+      "step": 1459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559216593710028e-05,
+      "loss": 0.8294,
+      "step": 1460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551931712858334e-05,
+      "loss": 0.872,
+      "step": 1461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454464378369591e-05,
+      "loss": 0.8649,
+      "step": 1462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537352811103291e-05,
+      "loss": 0.8942,
+      "step": 1463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530058799963052e-05,
+      "loss": 0.8542,
+      "step": 1464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522761755159799e-05,
+      "loss": 0.915,
+      "step": 1465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515461681580182e-05,
+      "loss": 0.9403,
+      "step": 1466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450815858411287e-05,
+      "loss": 0.9778,
+      "step": 1467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500852467648551e-05,
+      "loss": 0.8825,
+      "step": 1468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493543337079952e-05,
+      "loss": 0.7959,
+      "step": 1469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486231197301799e-05,
+      "loss": 0.8816,
+      "step": 1470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478916053210847e-05,
+      "loss": 0.9036,
+      "step": 1471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471597909705858e-05,
+      "loss": 0.9805,
+      "step": 1472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464276771687602e-05,
+      "loss": 0.9174,
+      "step": 1473
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445695264405886e-05,
+      "loss": 1.0216,
+      "step": 1474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4449625531724402e-05,
+      "loss": 0.8688,
+      "step": 1475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442295439591015e-05,
+      "loss": 0.9266,
+      "step": 1476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434962372567465e-05,
+      "loss": 1.0451,
+      "step": 1477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442762633556452e-05,
+      "loss": 0.8607,
+      "step": 1478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420287333494931e-05,
+      "loss": 0.8986,
+      "step": 1479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412945371273442e-05,
+      "loss": 0.8087,
+      "step": 1480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4405600453816773e-05,
+      "loss": 0.9518,
+      "step": 1481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439825258604362e-05,
+      "loss": 1.0199,
+      "step": 1482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4390901772874668e-05,
+      "loss": 0.9321,
+      "step": 1483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438354801923256e-05,
+      "loss": 0.8861,
+      "step": 1484
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376191330041917e-05,
+      "loss": 0.8305,
+      "step": 1485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368831710229324e-05,
+      "loss": 0.8353,
+      "step": 1486
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361469164723325e-05,
+      "loss": 0.8723,
+      "step": 1487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354103698454427e-05,
+      "loss": 0.8466,
+      "step": 1488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346735316355093e-05,
+      "loss": 0.9261,
+      "step": 1489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339364023359736e-05,
+      "loss": 0.8746,
+      "step": 1490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4331989824404722e-05,
+      "loss": 0.9671,
+      "step": 1491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432461272442836e-05,
+      "loss": 0.8603,
+      "step": 1492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43172327283709e-05,
+      "loss": 0.9107,
+      "step": 1493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4309849841174538e-05,
+      "loss": 0.918,
+      "step": 1494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302464067783404e-05,
+      "loss": 1.007,
+      "step": 1495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429507541314355e-05,
+      "loss": 0.9395,
+      "step": 1496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4287683882202974e-05,
+      "loss": 0.9007,
+      "step": 1497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280289479911593e-05,
+      "loss": 0.8814,
+      "step": 1498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272892211221245e-05,
+      "loss": 0.866,
+      "step": 1499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265492081085687e-05,
+      "loss": 0.8221,
+      "step": 1500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42580890944606e-05,
+      "loss": 0.8366,
+      "step": 1501
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4250683256303568e-05,
+      "loss": 0.9845,
+      "step": 1502
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243274571574089e-05,
+      "loss": 0.9138,
+      "step": 1503
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235863045233566e-05,
+      "loss": 0.8511,
+      "step": 1504
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422844868224531e-05,
+      "loss": 1.0433,
+      "step": 1505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4221031487574527e-05,
+      "loss": 0.8821,
+      "step": 1506
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213611466188317e-05,
+      "loss": 0.9471,
+      "step": 1507
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420618862305568e-05,
+      "loss": 0.8394,
+      "step": 1508
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198762963147501e-05,
+      "loss": 0.8731,
+      "step": 1509
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191334491436548e-05,
+      "loss": 0.9117,
+      "step": 1510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183903212897479e-05,
+      "loss": 0.9107,
+      "step": 1511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417646913250683e-05,
+      "loss": 0.8319,
+      "step": 1512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416903225524301e-05,
+      "loss": 0.9303,
+      "step": 1513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161592586086304e-05,
+      "loss": 0.9277,
+      "step": 1514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.9006,
+      "step": 1515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146704892024714e-05,
+      "loss": 0.8275,
+      "step": 1516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139256877089733e-05,
+      "loss": 0.8179,
+      "step": 1517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131806090201667e-05,
+      "loss": 0.8139,
+      "step": 1518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124352536350112e-05,
+      "loss": 0.8913,
+      "step": 1519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411689622052652e-05,
+      "loss": 0.897,
+      "step": 1520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4109437147724191e-05,
+      "loss": 0.9308,
+      "step": 1521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101975322938277e-05,
+      "loss": 0.8979,
+      "step": 1522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094510751165771e-05,
+      "loss": 0.7866,
+      "step": 1523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4087043437405497e-05,
+      "loss": 1.0717,
+      "step": 1524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079573386658119e-05,
+      "loss": 0.913,
+      "step": 1525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072100603926147e-05,
+      "loss": 0.9579,
+      "step": 1526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40646250942139e-05,
+      "loss": 0.8486,
+      "step": 1527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057146862527542e-05,
+      "loss": 0.8393,
+      "step": 1528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049665913875044e-05,
+      "loss": 0.9865,
+      "step": 1529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042182253266204e-05,
+      "loss": 0.8693,
+      "step": 1530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403469588571264e-05,
+      "loss": 0.8704,
+      "step": 1531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027206816227773e-05,
+      "loss": 0.8174,
+      "step": 1532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4019715049826845e-05,
+      "loss": 0.8003,
+      "step": 1533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4012220591526894e-05,
+      "loss": 0.9033,
+      "step": 1534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004723446346766e-05,
+      "loss": 1.0323,
+      "step": 1535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3997223619307101e-05,
+      "loss": 0.9199,
+      "step": 1536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3989721115430347e-05,
+      "loss": 0.9774,
+      "step": 1537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982215939740726e-05,
+      "loss": 0.9558,
+      "step": 1538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974708097264265e-05,
+      "loss": 0.9678,
+      "step": 1539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967197593028772e-05,
+      "loss": 0.8786,
+      "step": 1540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959684432063838e-05,
+      "loss": 0.7983,
+      "step": 1541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3952168619400834e-05,
+      "loss": 0.9819,
+      "step": 1542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944650160072897e-05,
+      "loss": 0.8202,
+      "step": 1543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937129059114951e-05,
+      "loss": 0.8986,
+      "step": 1544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929605321563682e-05,
+      "loss": 0.8732,
+      "step": 1545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392207895245754e-05,
+      "loss": 0.9013,
+      "step": 1546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914549956836738e-05,
+      "loss": 0.8938,
+      "step": 1547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907018339743255e-05,
+      "loss": 0.8942,
+      "step": 1548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899484106220816e-05,
+      "loss": 0.9401,
+      "step": 1549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891947261314899e-05,
+      "loss": 0.8721,
+      "step": 1550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884407810072742e-05,
+      "loss": 0.8842,
+      "step": 1551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876865757543313e-05,
+      "loss": 0.9327,
+      "step": 1552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869321108777332e-05,
+      "loss": 0.8571,
+      "step": 1553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861773868827254e-05,
+      "loss": 0.8476,
+      "step": 1554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854224042747269e-05,
+      "loss": 0.8827,
+      "step": 1555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846671635593305e-05,
+      "loss": 0.9703,
+      "step": 1556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839116652423006e-05,
+      "loss": 0.8605,
+      "step": 1557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831559098295755e-05,
+      "loss": 0.8167,
+      "step": 1558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3823998978272645e-05,
+      "loss": 0.8652,
+      "step": 1559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816436297416496e-05,
+      "loss": 0.9459,
+      "step": 1560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808871060791837e-05,
+      "loss": 0.8816,
+      "step": 1561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801303273464917e-05,
+      "loss": 0.9279,
+      "step": 1562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793732940503677e-05,
+      "loss": 0.967,
+      "step": 1563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786160066977778e-05,
+      "loss": 0.9336,
+      "step": 1564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778584657958583e-05,
+      "loss": 0.918,
+      "step": 1565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771006718519137e-05,
+      "loss": 1.0151,
+      "step": 1566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763426253734195e-05,
+      "loss": 0.9328,
+      "step": 1567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755843268680195e-05,
+      "loss": 0.8332,
+      "step": 1568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3748257768435272e-05,
+      "loss": 0.7861,
+      "step": 1569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740669758079235e-05,
+      "loss": 1.0385,
+      "step": 1570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733079242693572e-05,
+      "loss": 0.8887,
+      "step": 1571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3725486227361464e-05,
+      "loss": 0.8366,
+      "step": 1572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717890717167754e-05,
+      "loss": 0.8564,
+      "step": 1573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3710292717198956e-05,
+      "loss": 0.8967,
+      "step": 1574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702692232543253e-05,
+      "loss": 0.7928,
+      "step": 1575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3695089268290498e-05,
+      "loss": 0.8167,
+      "step": 1576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3687483829532194e-05,
+      "loss": 0.8701,
+      "step": 1577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3679875921361506e-05,
+      "loss": 0.9475,
+      "step": 1578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672265548873257e-05,
+      "loss": 0.936,
+      "step": 1579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3664652717163915e-05,
+      "loss": 0.8481,
+      "step": 1580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657037431331597e-05,
+      "loss": 0.8572,
+      "step": 1581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3649419696476057e-05,
+      "loss": 0.9023,
+      "step": 1582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36417995176987e-05,
+      "loss": 0.8157,
+      "step": 1583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634176900102563e-05,
+      "loss": 0.8273,
+      "step": 1584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3626551848792311e-05,
+      "loss": 0.8675,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618924368874246e-05,
+      "loss": 0.8345,
+      "step": 1586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611294465456295e-05,
+      "loss": 0.8628,
+      "step": 1587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603662143648007e-05,
+      "loss": 0.9016,
+      "step": 1588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596027408560544e-05,
+      "loss": 1.0243,
+      "step": 1589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588390265306697e-05,
+      "loss": 0.905,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580750719000862e-05,
+      "loss": 1.0982,
+      "step": 1591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3573108774759038e-05,
+      "loss": 0.8488,
+      "step": 1592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356546443769885e-05,
+      "loss": 0.9595,
+      "step": 1593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3557817712939503e-05,
+      "loss": 0.9282,
+      "step": 1594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3550168605601812e-05,
+      "loss": 0.8207,
+      "step": 1595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542517120808185e-05,
+      "loss": 0.8987,
+      "step": 1596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534863263682627e-05,
+      "loss": 0.8839,
+      "step": 1597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3527207039350727e-05,
+      "loss": 0.9244,
+      "step": 1598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519548452939657e-05,
+      "loss": 0.8902,
+      "step": 1599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511887509578175e-05,
+      "loss": 0.8987,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350422421439662e-05,
+      "loss": 0.863,
+      "step": 1601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496558572526897e-05,
+      "loss": 0.9023,
+      "step": 1602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348889058910249e-05,
+      "loss": 0.8423,
+      "step": 1603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3481220269258449e-05,
+      "loss": 0.8702,
+      "step": 1604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473547618131388e-05,
+      "loss": 0.9366,
+      "step": 1605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465872640859483e-05,
+      "loss": 0.8664,
+      "step": 1606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458195342582468e-05,
+      "loss": 0.9348,
+      "step": 1607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450515728441632e-05,
+      "loss": 0.9942,
+      "step": 1608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344283380357981e-05,
+      "loss": 0.9018,
+      "step": 1609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435149573141386e-05,
+      "loss": 0.8551,
+      "step": 1610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427463042272298e-05,
+      "loss": 0.8587,
+      "step": 1611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419774216120015e-05,
+      "loss": 0.7996,
+      "step": 1612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3412083099833538e-05,
+      "loss": 0.8908,
+      "step": 1613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340438969856341e-05,
+      "loss": 0.7814,
+      "step": 1614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3396694017461708e-05,
+      "loss": 0.9213,
+      "step": 1615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388996061682028e-05,
+      "loss": 0.8749,
+      "step": 1616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381295836379488e-05,
+      "loss": 0.9502,
+      "step": 1617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373593346710736e-05,
+      "loss": 0.8504,
+      "step": 1618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365888597833922e-05,
+      "loss": 0.9084,
+      "step": 1619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3358181594908723e-05,
+      "loss": 0.9522,
+      "step": 1620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350472343096317e-05,
+      "loss": 0.8691,
+      "step": 1621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342760847559388e-05,
+      "loss": 0.7945,
+      "step": 1622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335047113462127e-05,
+      "loss": 0.8194,
+      "step": 1623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327331145970221e-05,
+      "loss": 0.8263,
+      "step": 1624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319612950250856e-05,
+      "loss": 0.8064,
+      "step": 1625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3311892531472705e-05,
+      "loss": 0.8464,
+      "step": 1626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304169894805933e-05,
+      "loss": 0.9147,
+      "step": 1627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296445045422188e-05,
+      "loss": 0.8763,
+      "step": 1628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288717988494605e-05,
+      "loss": 0.8674,
+      "step": 1629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280988729197795e-05,
+      "loss": 0.9612,
+      "step": 1630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273257272707835e-05,
+      "loss": 0.9212,
+      "step": 1631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326552362420229e-05,
+      "loss": 0.8409,
+      "step": 1632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3257787788860178e-05,
+      "loss": 0.9489,
+      "step": 1633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325004977186199e-05,
+      "loss": 0.883,
+      "step": 1634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242309578389673e-05,
+      "loss": 0.8843,
+      "step": 1635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3234567213626639e-05,
+      "loss": 0.8539,
+      "step": 1636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3226822682757745e-05,
+      "loss": 0.9557,
+      "step": 1637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219075990969304e-05,
+      "loss": 0.9167,
+      "step": 1638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3211327143449076e-05,
+      "loss": 0.7765,
+      "step": 1639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320357614538626e-05,
+      "loss": 0.9097,
+      "step": 1640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.31958230019715e-05,
+      "loss": 0.9124,
+      "step": 1641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3188067718396876e-05,
+      "loss": 0.8857,
+      "step": 1642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180310299855901e-05,
+      "loss": 0.9215,
+      "step": 1643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3172550751543516e-05,
+      "loss": 0.8282,
+      "step": 1644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3164789078656092e-05,
+      "loss": 0.9248,
+      "step": 1645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157025286391416e-05,
+      "loss": 0.9549,
+      "step": 1646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3149259379948703e-05,
+      "loss": 0.862,
+      "step": 1647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141491364528576e-05,
+      "loss": 0.8397,
+      "step": 1648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3133721245333073e-05,
+      "loss": 0.8867,
+      "step": 1649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125949027565645e-05,
+      "loss": 0.8598,
+      "step": 1650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118174716431144e-05,
+      "loss": 0.9006,
+      "step": 1651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110398317135822e-05,
+      "loss": 0.9481,
+      "step": 1652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102619834887336e-05,
+      "loss": 0.8439,
+      "step": 1653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3094839274894731e-05,
+      "loss": 0.8851,
+      "step": 1654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087056642368447e-05,
+      "loss": 0.9665,
+      "step": 1655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307927194252031e-05,
+      "loss": 0.8457,
+      "step": 1656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071485180563537e-05,
+      "loss": 0.9754,
+      "step": 1657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063696361712717e-05,
+      "loss": 0.9767,
+      "step": 1658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055905491183822e-05,
+      "loss": 0.9077,
+      "step": 1659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048112574194191e-05,
+      "loss": 0.8381,
+      "step": 1660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040317615962543e-05,
+      "loss": 0.8253,
+      "step": 1661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032520621708962e-05,
+      "loss": 0.8728,
+      "step": 1662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3024721596654889e-05,
+      "loss": 0.886,
+      "step": 1663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016920546023127e-05,
+      "loss": 0.8924,
+      "step": 1664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009117475037843e-05,
+      "loss": 0.861,
+      "step": 1665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3001312388924547e-05,
+      "loss": 0.9747,
+      "step": 1666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993505292910108e-05,
+      "loss": 0.9389,
+      "step": 1667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298569619222273e-05,
+      "loss": 0.957,
+      "step": 1668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977885092091969e-05,
+      "loss": 0.8578,
+      "step": 1669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970071997748712e-05,
+      "loss": 0.8784,
+      "step": 1670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962256914425195e-05,
+      "loss": 0.9349,
+      "step": 1671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954439847354971e-05,
+      "loss": 0.8573,
+      "step": 1672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946620801772926e-05,
+      "loss": 0.9385,
+      "step": 1673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293879978291527e-05,
+      "loss": 0.8732,
+      "step": 1674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930976796019546e-05,
+      "loss": 0.9556,
+      "step": 1675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923151846324597e-05,
+      "loss": 0.8187,
+      "step": 1676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915324939070594e-05,
+      "loss": 0.8726,
+      "step": 1677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907496079499011e-05,
+      "loss": 0.9507,
+      "step": 1678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2899665272852635e-05,
+      "loss": 0.8131,
+      "step": 1679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891832524375551e-05,
+      "loss": 0.8657,
+      "step": 1680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288399783931315e-05,
+      "loss": 0.9276,
+      "step": 1681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876161222912118e-05,
+      "loss": 0.9049,
+      "step": 1682
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868322680420433e-05,
+      "loss": 0.8185,
+      "step": 1683
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860482217087364e-05,
+      "loss": 0.8118,
+      "step": 1684
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852639838163465e-05,
+      "loss": 0.8441,
+      "step": 1685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284479554890058e-05,
+      "loss": 0.863,
+      "step": 1686
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836949354551818e-05,
+      "loss": 0.7957,
+      "step": 1687
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829101260371572e-05,
+      "loss": 0.8626,
+      "step": 1688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2821251271615515e-05,
+      "loss": 0.9501,
+      "step": 1689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813399393540578e-05,
+      "loss": 0.9669,
+      "step": 1690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2805545631404957e-05,
+      "loss": 0.8449,
+      "step": 1691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2797689990468113e-05,
+      "loss": 0.8984,
+      "step": 1692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789832475990771e-05,
+      "loss": 0.8256,
+      "step": 1693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27819730932349e-05,
+      "loss": 0.8812,
+      "step": 1694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774111847463728e-05,
+      "loss": 0.9359,
+      "step": 1695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2766248743941725e-05,
+      "loss": 0.9418,
+      "step": 1696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275838378793461e-05,
+      "loss": 0.9231,
+      "step": 1697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750516984709335e-05,
+      "loss": 1.0035,
+      "step": 1698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27426483395341e-05,
+      "loss": 0.8052,
+      "step": 1699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2734777857678333e-05,
+      "loss": 0.9131,
+      "step": 1700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2726905544412689e-05,
+      "loss": 0.9245,
+      "step": 1701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271903140500905e-05,
+      "loss": 0.8785,
+      "step": 1702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711155444740529e-05,
+      "loss": 0.9351,
+      "step": 1703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270327766888145e-05,
+      "loss": 0.902,
+      "step": 1704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2695398082707355e-05,
+      "loss": 0.8967,
+      "step": 1705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687516691494998e-05,
+      "loss": 0.9383,
+      "step": 1706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679633500522343e-05,
+      "loss": 0.9213,
+      "step": 1707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267174851506856e-05,
+      "loss": 0.8508,
+      "step": 1708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266386174041402e-05,
+      "loss": 0.9171,
+      "step": 1709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655973181840286e-05,
+      "loss": 0.9308,
+      "step": 1710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648082844630127e-05,
+      "loss": 0.756,
+      "step": 1711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26401907340675e-05,
+      "loss": 0.8298,
+      "step": 1712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2632296855437535e-05,
+      "loss": 0.8661,
+      "step": 1713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624401214026574e-05,
+      "loss": 0.8657,
+      "step": 1714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2616503815122113e-05,
+      "loss": 0.8501,
+      "step": 1715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2608604664012833e-05,
+      "loss": 0.8976,
+      "step": 1716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26007037659886e-05,
+      "loss": 0.8492,
+      "step": 1717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2592801126340434e-05,
+      "loss": 0.9702,
+      "step": 1718
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584896750360536e-05,
+      "loss": 0.8313,
+      "step": 1719
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576990643342248e-05,
+      "loss": 0.8987,
+      "step": 1720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2569082810580097e-05,
+      "loss": 0.9396,
+      "step": 1721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.256117325736975e-05,
+      "loss": 0.9574,
+      "step": 1722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553261989008031e-05,
+      "loss": 0.7936,
+      "step": 1723
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2545349010792908e-05,
+      "loss": 0.9046,
+      "step": 1724
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537434328023501e-05,
+      "loss": 0.882,
+      "step": 1725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529517946000067e-05,
+      "loss": 0.8833,
+      "step": 1726
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521599870023998e-05,
+      "loss": 0.964,
+      "step": 1727
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513680105397835e-05,
+      "loss": 0.8708,
+      "step": 1728
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505758657425227e-05,
+      "loss": 0.9753,
+      "step": 1729
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249783553141097e-05,
+      "loss": 0.9389,
+      "step": 1730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489910732660968e-05,
+      "loss": 0.8734,
+      "step": 1731
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481984266482264e-05,
+      "loss": 0.871,
+      "step": 1732
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474056138183004e-05,
+      "loss": 0.9091,
+      "step": 1733
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466126353072443e-05,
+      "loss": 0.8559,
+      "step": 1734
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458194916460962e-05,
+      "loss": 0.8542,
+      "step": 1735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2450261833660033e-05,
+      "loss": 0.8567,
+      "step": 1736
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2442327109982237e-05,
+      "loss": 0.8536,
+      "step": 1737
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434390750741256e-05,
+      "loss": 0.9742,
+      "step": 1738
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2426452761251861e-05,
+      "loss": 0.8731,
+      "step": 1739
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418513146829918e-05,
+      "loss": 0.8049,
+      "step": 1740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410571912792379e-05,
+      "loss": 0.9407,
+      "step": 1741
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240262906445729e-05,
+      "loss": 0.8279,
+      "step": 1742
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2394684607143763e-05,
+      "loss": 0.8089,
+      "step": 1743
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386738546172e-05,
+      "loss": 0.9183,
+      "step": 1744
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378790886863267e-05,
+      "loss": 0.8503,
+      "step": 1745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370841634539912e-05,
+      "loss": 0.7985,
+      "step": 1746
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362890794525342e-05,
+      "loss": 0.9632,
+      "step": 1747
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235493837214402e-05,
+      "loss": 0.9111,
+      "step": 1748
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346984372721492e-05,
+      "loss": 0.8221,
+      "step": 1749
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339028801584332e-05,
+      "loss": 0.9008,
+      "step": 1750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331071664060189e-05,
+      "loss": 0.9133,
+      "step": 1751
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2323112965477744e-05,
+      "loss": 0.9407,
+      "step": 1752
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2315152711166737e-05,
+      "loss": 0.8675,
+      "step": 1753
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307190906457944e-05,
+      "loss": 0.8471,
+      "step": 1754
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299227556683175e-05,
+      "loss": 0.916,
+      "step": 1755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2291262667175286e-05,
+      "loss": 0.8921,
+      "step": 1756
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283296243268151e-05,
+      "loss": 0.8586,
+      "step": 1757
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2275328290296677e-05,
+      "loss": 0.8404,
+      "step": 1758
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.22673588135968e-05,
+      "loss": 0.8996,
+      "step": 1759
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2259387818505468e-05,
+      "loss": 0.8673,
+      "step": 1760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251415310360656e-05,
+      "loss": 0.8163,
+      "step": 1761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243441294501344e-05,
+      "loss": 0.8336,
+      "step": 1762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2235465776267516e-05,
+      "loss": 0.7749,
+      "step": 1763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227488761000178e-05,
+      "loss": 0.931,
+      "step": 1764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2219510254041332e-05,
+      "loss": 0.8696,
+      "step": 1765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221153026073397e-05,
+      "loss": 0.7996,
+      "step": 1766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203548786422094e-05,
+      "loss": 0.8641,
+      "step": 1767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195565836450687e-05,
+      "loss": 0.8736,
+      "step": 1768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187581416165721e-05,
+      "loss": 0.8989,
+      "step": 1769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2179595530914162e-05,
+      "loss": 0.8458,
+      "step": 1770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2171608186043944e-05,
+      "loss": 0.8936,
+      "step": 1771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163619386903988e-05,
+      "loss": 0.8218,
+      "step": 1772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155629138844184e-05,
+      "loss": 0.8409,
+      "step": 1773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147637447215402e-05,
+      "loss": 0.8441,
+      "step": 1774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139644317369461e-05,
+      "loss": 0.8836,
+      "step": 1775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131649754659156e-05,
+      "loss": 0.8679,
+      "step": 1776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212365376443824e-05,
+      "loss": 0.9322,
+      "step": 1777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115656352061417e-05,
+      "loss": 0.9057,
+      "step": 1778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107657522884352e-05,
+      "loss": 0.8066,
+      "step": 1779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099657282263651e-05,
+      "loss": 0.9196,
+      "step": 1780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091655635556866e-05,
+      "loss": 0.9085,
+      "step": 1781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083652588122495e-05,
+      "loss": 0.8761,
+      "step": 1782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207564814531997e-05,
+      "loss": 0.8618,
+      "step": 1783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067642312509663e-05,
+      "loss": 0.8192,
+      "step": 1784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059635095052868e-05,
+      "loss": 0.8079,
+      "step": 1785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051626498311813e-05,
+      "loss": 0.9299,
+      "step": 1786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2043616527649643e-05,
+      "loss": 0.857,
+      "step": 1787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2035605188430437e-05,
+      "loss": 0.9762,
+      "step": 1788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202759248601918e-05,
+      "loss": 0.8874,
+      "step": 1789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2019578425781764e-05,
+      "loss": 0.9121,
+      "step": 1790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2011563013084996e-05,
+      "loss": 0.8402,
+      "step": 1791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2003546253296602e-05,
+      "loss": 0.8823,
+      "step": 1792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1995528151785191e-05,
+      "loss": 0.8431,
+      "step": 1793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198750871392028e-05,
+      "loss": 0.9657,
+      "step": 1794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979487945072275e-05,
+      "loss": 0.8632,
+      "step": 1795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971465850612481e-05,
+      "loss": 0.8714,
+      "step": 1796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1963442435913083e-05,
+      "loss": 0.9943,
+      "step": 1797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1955417706347159e-05,
+      "loss": 0.9099,
+      "step": 1798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1947391667288657e-05,
+      "loss": 0.893,
+      "step": 1799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1939364324112412e-05,
+      "loss": 0.9382,
+      "step": 1800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1931335682194125e-05,
+      "loss": 0.9483,
+      "step": 1801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923305746910372e-05,
+      "loss": 0.7751,
+      "step": 1802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1915274523638592e-05,
+      "loss": 0.8036,
+      "step": 1803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907242017757084e-05,
+      "loss": 0.8416,
+      "step": 1804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189920823464501e-05,
+      "loss": 0.9181,
+      "step": 1805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189117317968239e-05,
+      "loss": 0.9257,
+      "step": 1806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1883136858250088e-05,
+      "loss": 0.9843,
+      "step": 1807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1875099275729824e-05,
+      "loss": 0.7971,
+      "step": 1808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1867060437504156e-05,
+      "loss": 0.9211,
+      "step": 1809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859020348956487e-05,
+      "loss": 0.9437,
+      "step": 1810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1850979015471053e-05,
+      "loss": 0.9232,
+      "step": 1811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842936442432933e-05,
+      "loss": 0.8939,
+      "step": 1812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834892635228024e-05,
+      "loss": 0.9343,
+      "step": 1813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826847599243056e-05,
+      "loss": 0.8607,
+      "step": 1814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1818801339865584e-05,
+      "loss": 0.8071,
+      "step": 1815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1810753862483977e-05,
+      "loss": 0.8511,
+      "step": 1816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1802705172487424e-05,
+      "loss": 0.8266,
+      "step": 1817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794655275265926e-05,
+      "loss": 0.8539,
+      "step": 1818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786604176210281e-05,
+      "loss": 0.8943,
+      "step": 1819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778551880712113e-05,
+      "loss": 0.8607,
+      "step": 1820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770498394163827e-05,
+      "loss": 0.9075,
+      "step": 1821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762443721958643e-05,
+      "loss": 0.9509,
+      "step": 1822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1754387869490557e-05,
+      "loss": 1.0182,
+      "step": 1823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746330842154371e-05,
+      "loss": 0.8574,
+      "step": 1824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738272645345665e-05,
+      "loss": 0.8607,
+      "step": 1825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730213284460808e-05,
+      "loss": 0.9657,
+      "step": 1826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172215276489694e-05,
+      "loss": 0.855,
+      "step": 1827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714091092051984e-05,
+      "loss": 0.8495,
+      "step": 1828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706028271324634e-05,
+      "loss": 0.8185,
+      "step": 1829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697964308114349e-05,
+      "loss": 0.889,
+      "step": 1830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689899207821367e-05,
+      "loss": 0.8704,
+      "step": 1831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1681832975846663e-05,
+      "loss": 0.924,
+      "step": 1832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167376561759199e-05,
+      "loss": 0.8282,
+      "step": 1833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1665697138459849e-05,
+      "loss": 0.9377,
+      "step": 1834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1657627543853491e-05,
+      "loss": 0.7785,
+      "step": 1835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649556839176917e-05,
+      "loss": 0.9085,
+      "step": 1836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164148502983486e-05,
+      "loss": 0.9344,
+      "step": 1837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633412121232813e-05,
+      "loss": 1.0096,
+      "step": 1838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1625338118776985e-05,
+      "loss": 0.8706,
+      "step": 1839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1617263027874333e-05,
+      "loss": 0.8265,
+      "step": 1840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609186853932531e-05,
+      "loss": 0.9231,
+      "step": 1841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1601109602359984e-05,
+      "loss": 0.8821,
+      "step": 1842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593031278565818e-05,
+      "loss": 0.8746,
+      "step": 1843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158495188795988e-05,
+      "loss": 0.9115,
+      "step": 1844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157687143595273e-05,
+      "loss": 0.9115,
+      "step": 1845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156878992795563e-05,
+      "loss": 0.8861,
+      "step": 1846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1560707369380563e-05,
+      "loss": 0.9358,
+      "step": 1847
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1552623765640206e-05,
+      "loss": 0.8305,
+      "step": 1848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544539122147939e-05,
+      "loss": 0.9267,
+      "step": 1849
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1536453444317846e-05,
+      "loss": 0.8628,
+      "step": 1850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152836673756468e-05,
+      "loss": 0.9214,
+      "step": 1851
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1520279007303912e-05,
+      "loss": 0.9079,
+      "step": 1852
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1512190258951683e-05,
+      "loss": 0.9196,
+      "step": 1853
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1504100497924815e-05,
+      "loss": 0.9326,
+      "step": 1854
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1496009729640814e-05,
+      "loss": 0.8856,
+      "step": 1855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1487917959517856e-05,
+      "loss": 0.9743,
+      "step": 1856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479825192974791e-05,
+      "loss": 0.9305,
+      "step": 1857
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471731435431134e-05,
+      "loss": 0.8934,
+      "step": 1858
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146363669230707e-05,
+      "loss": 0.8705,
+      "step": 1859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455540969023434e-05,
+      "loss": 0.9273,
+      "step": 1860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447444271001717e-05,
+      "loss": 0.9076,
+      "step": 1861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439346603664076e-05,
+      "loss": 0.9134,
+      "step": 1862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431247972433306e-05,
+      "loss": 0.9958,
+      "step": 1863
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.9331,
+      "step": 1864
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1415047839986797e-05,
+      "loss": 0.8444,
+      "step": 1865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1406946349619867e-05,
+      "loss": 0.9254,
+      "step": 1866
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1398843917057421e-05,
+      "loss": 0.8308,
+      "step": 1867
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1390740547725443e-05,
+      "loss": 0.7884,
+      "step": 1868
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382636247050556e-05,
+      "loss": 0.8831,
+      "step": 1869
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1374531020459994e-05,
+      "loss": 1.0115,
+      "step": 1870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136642487338162e-05,
+      "loss": 0.7961,
+      "step": 1871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135831781124391e-05,
+      "loss": 0.8777,
+      "step": 1872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350209839475961e-05,
+      "loss": 0.9777,
+      "step": 1873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342100963507458e-05,
+      "loss": 0.8523,
+      "step": 1874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333991188768716e-05,
+      "loss": 0.8896,
+      "step": 1875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325880520690632e-05,
+      "loss": 0.9548,
+      "step": 1876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131776896470472e-05,
+      "loss": 0.9579,
+      "step": 1877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309656526243075e-05,
+      "loss": 0.8249,
+      "step": 1878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301543210738383e-05,
+      "loss": 0.903,
+      "step": 1879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293429023623927e-05,
+      "loss": 0.7592,
+      "step": 1880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285313970333564e-05,
+      "loss": 0.9352,
+      "step": 1881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277198056301737e-05,
+      "loss": 0.8234,
+      "step": 1882
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269081286963461e-05,
+      "loss": 0.9005,
+      "step": 1883
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260963667754328e-05,
+      "loss": 0.9675,
+      "step": 1884
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252845204110496e-05,
+      "loss": 0.9176,
+      "step": 1885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244725901468695e-05,
+      "loss": 0.92,
+      "step": 1886
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236605765266203e-05,
+      "loss": 1.1006,
+      "step": 1887
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228484800940869e-05,
+      "loss": 0.9483,
+      "step": 1888
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1220363013931096e-05,
+      "loss": 1.0278,
+      "step": 1889
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212240409675825e-05,
+      "loss": 0.8139,
+      "step": 1890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1204116993614562e-05,
+      "loss": 0.8117,
+      "step": 1891
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119599277118735e-05,
+      "loss": 0.905,
+      "step": 1892
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1187867747834762e-05,
+      "loss": 0.8985,
+      "step": 1893
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179741928997918e-05,
+      "loss": 0.8754,
+      "step": 1894
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171615320118473e-05,
+      "loss": 0.8737,
+      "step": 1895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163487926638603e-05,
+      "loss": 1.0228,
+      "step": 1896
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1155359754001014e-05,
+      "loss": 0.9463,
+      "step": 1897
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114723080764893e-05,
+      "loss": 0.8383,
+      "step": 1898
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139101093026099e-05,
+      "loss": 0.7981,
+      "step": 1899
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1130970615576777e-05,
+      "loss": 0.9073,
+      "step": 1900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122839380745738e-05,
+      "loss": 0.998,
+      "step": 1901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114707393978258e-05,
+      "loss": 0.8763,
+      "step": 1902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1106574660720116e-05,
+      "loss": 0.92,
+      "step": 1903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109844118641759e-05,
+      "loss": 0.8798,
+      "step": 1904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090306976517464e-05,
+      "loss": 0.8586,
+      "step": 1905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082172036467007e-05,
+      "loss": 0.8581,
+      "step": 1906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074036371713973e-05,
+      "loss": 0.8881,
+      "step": 1907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065899987706606e-05,
+      "loss": 0.8225,
+      "step": 1908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1057762889893635e-05,
+      "loss": 0.8736,
+      "step": 1909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1049625083724262e-05,
+      "loss": 0.8847,
+      "step": 1910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1041486574648166e-05,
+      "loss": 0.773,
+      "step": 1911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033347368115494e-05,
+      "loss": 0.8475,
+      "step": 1912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1025207469576865e-05,
+      "loss": 0.9013,
+      "step": 1913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017066884483352e-05,
+      "loss": 0.8002,
+      "step": 1914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10089256182865e-05,
+      "loss": 0.8789,
+      "step": 1915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1000783676438301e-05,
+      "loss": 0.8508,
+      "step": 1916
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.09926410643912e-05,
+      "loss": 0.8777,
+      "step": 1917
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984497787598099e-05,
+      "loss": 0.9663,
+      "step": 1918
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0976353851512336e-05,
+      "loss": 0.9559,
+      "step": 1919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0968209261587698e-05,
+      "loss": 0.8007,
+      "step": 1920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09600640232784e-05,
+      "loss": 0.9398,
+      "step": 1921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0951918142039099e-05,
+      "loss": 0.8578,
+      "step": 1922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943771623324884e-05,
+      "loss": 0.8974,
+      "step": 1923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0935624472591262e-05,
+      "loss": 0.852,
+      "step": 1924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0927476695294177e-05,
+      "loss": 0.9001,
+      "step": 1925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0919328296889973e-05,
+      "loss": 0.8341,
+      "step": 1926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0911179282835432e-05,
+      "loss": 0.9378,
+      "step": 1927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903029658587728e-05,
+      "loss": 0.841,
+      "step": 1928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089487942960446e-05,
+      "loss": 0.8693,
+      "step": 1929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886728601343626e-05,
+      "loss": 0.9511,
+      "step": 1930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087857717926362e-05,
+      "loss": 0.962,
+      "step": 1931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870425168823241e-05,
+      "loss": 0.9307,
+      "step": 1932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862272575481677e-05,
+      "loss": 0.888,
+      "step": 1933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085411940469851e-05,
+      "loss": 0.822,
+      "step": 1934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845965661933709e-05,
+      "loss": 0.9556,
+      "step": 1935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0837811352647619e-05,
+      "loss": 0.9641,
+      "step": 1936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0829656482300972e-05,
+      "loss": 0.8218,
+      "step": 1937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821501056354876e-05,
+      "loss": 0.7804,
+      "step": 1938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813345080270812e-05,
+      "loss": 0.8495,
+      "step": 1939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805188559510613e-05,
+      "loss": 0.8716,
+      "step": 1940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0797031499536502e-05,
+      "loss": 0.9847,
+      "step": 1941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0788873905811044e-05,
+      "loss": 1.0058,
+      "step": 1942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780715783797168e-05,
+      "loss": 0.8258,
+      "step": 1943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0772557138958162e-05,
+      "loss": 0.8819,
+      "step": 1944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0764397976757658e-05,
+      "loss": 0.8162,
+      "step": 1945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075623830265963e-05,
+      "loss": 0.9294,
+      "step": 1946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0748078122128403e-05,
+      "loss": 0.8083,
+      "step": 1947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0739917440628645e-05,
+      "loss": 0.896,
+      "step": 1948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731756263625343e-05,
+      "loss": 0.8653,
+      "step": 1949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723594596583828e-05,
+      "loss": 0.8549,
+      "step": 1950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0715432444969765e-05,
+      "loss": 0.8803,
+      "step": 1951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707269814249127e-05,
+      "loss": 0.8311,
+      "step": 1952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069910670988822e-05,
+      "loss": 1.0068,
+      "step": 1953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069094313735366e-05,
+      "loss": 0.9073,
+      "step": 1954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682779102112386e-05,
+      "loss": 0.8636,
+      "step": 1955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674614609631634e-05,
+      "loss": 0.8774,
+      "step": 1956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066644966537896e-05,
+      "loss": 0.8713,
+      "step": 1957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065828427482221e-05,
+      "loss": 0.8578,
+      "step": 1958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0650118443429536e-05,
+      "loss": 0.9283,
+      "step": 1959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641952176669385e-05,
+      "loss": 0.8453,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0633785480010492e-05,
+      "loss": 0.7645,
+      "step": 1961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0625618358921888e-05,
+      "loss": 0.9534,
+      "step": 1962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617450818872876e-05,
+      "loss": 0.9665,
+      "step": 1963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0609282865333046e-05,
+      "loss": 0.9193,
+      "step": 1964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601114503772267e-05,
+      "loss": 0.925,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059294573966068e-05,
+      "loss": 0.795,
+      "step": 1966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0584776578468698e-05,
+      "loss": 0.9004,
+      "step": 1967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0576607025666984e-05,
+      "loss": 0.9033,
+      "step": 1968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056843708672649e-05,
+      "loss": 0.9006,
+      "step": 1969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560266767118406e-05,
+      "loss": 0.9361,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552096072314183e-05,
+      "loss": 0.9045,
+      "step": 1971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0543925007785525e-05,
+      "loss": 0.8271,
+      "step": 1972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0535753579004382e-05,
+      "loss": 0.9342,
+      "step": 1973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052758179144295e-05,
+      "loss": 0.8288,
+      "step": 1974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519409650573656e-05,
+      "loss": 0.7813,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0511237161869182e-05,
+      "loss": 0.799,
+      "step": 1976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0503064330802422e-05,
+      "loss": 0.9495,
+      "step": 1977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494891162846515e-05,
+      "loss": 0.8541,
+      "step": 1978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486717663474816e-05,
+      "loss": 0.8443,
+      "step": 1979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0478543838160911e-05,
+      "loss": 0.8672,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04703696923786e-05,
+      "loss": 0.8227,
+      "step": 1981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462195231601884e-05,
+      "loss": 0.8204,
+      "step": 1982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0454020461305002e-05,
+      "loss": 0.9549,
+      "step": 1983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044584538696238e-05,
+      "loss": 0.9046,
+      "step": 1984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437670014048651e-05,
+      "loss": 0.8618,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0429494348038656e-05,
+      "loss": 0.8715,
+      "step": 1986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0421318394407423e-05,
+      "loss": 0.8227,
+      "step": 1987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413142158630176e-05,
+      "loss": 0.8373,
+      "step": 1988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040496564618233e-05,
+      "loss": 0.8419,
+      "step": 1989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0396788862539484e-05,
+      "loss": 0.869,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388611813177416e-05,
+      "loss": 0.8861,
+      "step": 1991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380434503572086e-05,
+      "loss": 0.7734,
+      "step": 1992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372256939199622e-05,
+      "loss": 0.8745,
+      "step": 1993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0364079125536334e-05,
+      "loss": 0.8956,
+      "step": 1994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355901068058692e-05,
+      "loss": 0.846,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034772277224332e-05,
+      "loss": 0.8553,
+      "step": 1996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033954424356702e-05,
+      "loss": 0.8948,
+      "step": 1997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0331365487506735e-05,
+      "loss": 0.8042,
+      "step": 1998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0323186509539569e-05,
+      "loss": 0.9054,
+      "step": 1999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315007315142772e-05,
+      "loss": 0.8544,
+      "step": 2000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0306827909793735e-05,
+      "loss": 0.8471,
+      "step": 2001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029864829897e-05,
+      "loss": 0.8583,
+      "step": 2002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0290468488149235e-05,
+      "loss": 0.8152,
+      "step": 2003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0282288482809248e-05,
+      "loss": 0.8662,
+      "step": 2004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027410828842798e-05,
+      "loss": 0.9362,
+      "step": 2005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265927910483492e-05,
+      "loss": 0.7992,
+      "step": 2006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257747354453971e-05,
+      "loss": 0.8609,
+      "step": 2007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249566625817726e-05,
+      "loss": 0.8356,
+      "step": 2008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024138573005318e-05,
+      "loss": 0.8878,
+      "step": 2009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233204672638862e-05,
+      "loss": 0.8926,
+      "step": 2010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225023459053416e-05,
+      "loss": 0.9624,
+      "step": 2011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021684209477559e-05,
+      "loss": 0.8014,
+      "step": 2012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0208660585284232e-05,
+      "loss": 0.8761,
+      "step": 2013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0200478936058287e-05,
+      "loss": 0.8869,
+      "step": 2014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0192297152576791e-05,
+      "loss": 0.8602,
+      "step": 2015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184115240318876e-05,
+      "loss": 0.8502,
+      "step": 2016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175933204763753e-05,
+      "loss": 0.848,
+      "step": 2017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0167751051390723e-05,
+      "loss": 1.0448,
+      "step": 2018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0159568785679158e-05,
+      "loss": 0.8181,
+      "step": 2019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0151386413108513e-05,
+      "loss": 0.9384,
+      "step": 2020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143203939158305e-05,
+      "loss": 0.8898,
+      "step": 2021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135021369308138e-05,
+      "loss": 0.9996,
+      "step": 2022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0126838709037651e-05,
+      "loss": 0.9095,
+      "step": 2023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0118655963826572e-05,
+      "loss": 0.8699,
+      "step": 2024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110473139154662e-05,
+      "loss": 0.9185,
+      "step": 2025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0102290240501755e-05,
+      "loss": 0.964,
+      "step": 2026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094107273347727e-05,
+      "loss": 0.9192,
+      "step": 2027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0085924243172493e-05,
+      "loss": 0.8766,
+      "step": 2028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077741155456022e-05,
+      "loss": 0.8185,
+      "step": 2029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069558015678313e-05,
+      "loss": 0.913,
+      "step": 2030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0061374829319403e-05,
+      "loss": 0.8201,
+      "step": 2031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005319160185936e-05,
+      "loss": 0.8052,
+      "step": 2032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004500833877828e-05,
+      "loss": 0.8161,
+      "step": 2033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003682504555628e-05,
+      "loss": 0.8998,
+      "step": 2034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00286417276735e-05,
+      "loss": 0.8073,
+      "step": 2035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020458390610101e-05,
+      "loss": 0.9862,
+      "step": 2036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012275039846247e-05,
+      "loss": 0.9156,
+      "step": 2037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0004091680862113e-05,
+      "loss": 0.9174,
+      "step": 2038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995908319137888e-06,
+      "loss": 0.9579,
+      "step": 2039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98772496015376e-06,
+      "loss": 0.9367,
+      "step": 2040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979541609389902e-06,
+      "loss": 0.7691,
+      "step": 2041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971358272326501e-06,
+      "loss": 0.7937,
+      "step": 2042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963174954443725e-06,
+      "loss": 0.9004,
+      "step": 2043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954991661221724e-06,
+      "loss": 0.8871,
+      "step": 2044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.946808398140643e-06,
+      "loss": 0.8713,
+      "step": 2045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9386251706806e-06,
+      "loss": 0.9532,
+      "step": 2046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93044198432169e-06,
+      "loss": 0.9004,
+      "step": 2047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922258844543978e-06,
+      "loss": 0.9165,
+      "step": 2048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914075756827508e-06,
+      "loss": 0.8338,
+      "step": 2049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.905892726652277e-06,
+      "loss": 0.8867,
+      "step": 2050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897709759498245e-06,
+      "loss": 0.8709,
+      "step": 2051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889526860845341e-06,
+      "loss": 0.8969,
+      "step": 2052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881344036173433e-06,
+      "loss": 0.8268,
+      "step": 2053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873161290962354e-06,
+      "loss": 0.8839,
+      "step": 2054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864978630691865e-06,
+      "loss": 0.8649,
+      "step": 2055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856796060841695e-06,
+      "loss": 0.8706,
+      "step": 2056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848613586891492e-06,
+      "loss": 0.8002,
+      "step": 2057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.840431214320845e-06,
+      "loss": 0.8508,
+      "step": 2058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83224894860928e-06,
+      "loss": 0.8468,
+      "step": 2059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82406679523625e-06,
+      "loss": 0.9246,
+      "step": 2060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815884759681128e-06,
+      "loss": 0.8287,
+      "step": 2061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807702847423209e-06,
+      "loss": 0.8157,
+      "step": 2062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.799521063941717e-06,
+      "loss": 0.8819,
+      "step": 2063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791339414715771e-06,
+      "loss": 0.9214,
+      "step": 2064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783157905224413e-06,
+      "loss": 0.8029,
+      "step": 2065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774976540946589e-06,
+      "loss": 0.8965,
+      "step": 2066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766795327361142e-06,
+      "loss": 0.8936,
+      "step": 2067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758614269946825e-06,
+      "loss": 0.9234,
+      "step": 2068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750433374182277e-06,
+      "loss": 0.8241,
+      "step": 2069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74225264554603e-06,
+      "loss": 0.9107,
+      "step": 2070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734072089516513e-06,
+      "loss": 0.8079,
+      "step": 2071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725891711572023e-06,
+      "loss": 0.848,
+      "step": 2072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717711517190754e-06,
+      "loss": 0.8805,
+      "step": 2073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.709531511850769e-06,
+      "loss": 0.9655,
+      "step": 2074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701351701030004e-06,
+      "loss": 0.9084,
+      "step": 2075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.693172090206264e-06,
+      "loss": 0.8569,
+      "step": 2076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684992684857232e-06,
+      "loss": 0.7939,
+      "step": 2077
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.676813490460434e-06,
+      "loss": 0.8466,
+      "step": 2078
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668634512493269e-06,
+      "loss": 0.9254,
+      "step": 2079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660455756432986e-06,
+      "loss": 0.8969,
+      "step": 2080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652277227756681e-06,
+      "loss": 0.8781,
+      "step": 2081
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644098931941313e-06,
+      "loss": 0.8352,
+      "step": 2082
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.635920874463667e-06,
+      "loss": 0.8149,
+      "step": 2083
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627743060800378e-06,
+      "loss": 0.8886,
+      "step": 2084
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619565496427919e-06,
+      "loss": 0.936,
+      "step": 2085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611388186822586e-06,
+      "loss": 0.8883,
+      "step": 2086
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603211137460518e-06,
+      "loss": 0.8788,
+      "step": 2087
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595034353817673e-06,
+      "loss": 0.7318,
+      "step": 2088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586857841369826e-06,
+      "loss": 0.9245,
+      "step": 2089
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578681605592579e-06,
+      "loss": 0.8443,
+      "step": 2090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570505651961347e-06,
+      "loss": 0.8967,
+      "step": 2091
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562329985951352e-06,
+      "loss": 0.9521,
+      "step": 2092
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.554154613037624e-06,
+      "loss": 0.9201,
+      "step": 2093
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545979538695001e-06,
+      "loss": 0.8383,
+      "step": 2094
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537804768398116e-06,
+      "loss": 0.8768,
+      "step": 2095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529630307621407e-06,
+      "loss": 0.9262,
+      "step": 2096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52145616183909e-06,
+      "loss": 0.8552,
+      "step": 2097
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513282336525184e-06,
+      "loss": 0.8902,
+      "step": 2098
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505108837153489e-06,
+      "loss": 0.8996,
+      "step": 2099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49693566919758e-06,
+      "loss": 0.9026,
+      "step": 2100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488762838130823e-06,
+      "loss": 0.941,
+      "step": 2101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480590349426347e-06,
+      "loss": 0.8548,
+      "step": 2102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472418208557056e-06,
+      "loss": 0.8182,
+      "step": 2103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.464246420995623e-06,
+      "loss": 0.8273,
+      "step": 2104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456074992214477e-06,
+      "loss": 0.9184,
+      "step": 2105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447903927685818e-06,
+      "loss": 0.9405,
+      "step": 2106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439733232881596e-06,
+      "loss": 0.8502,
+      "step": 2107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.431562913273511e-06,
+      "loss": 0.9018,
+      "step": 2108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.423392974333014e-06,
+      "loss": 0.8416,
+      "step": 2109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415223421531308e-06,
+      "loss": 0.9001,
+      "step": 2110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407054260339323e-06,
+      "loss": 0.855,
+      "step": 2111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398885496227733e-06,
+      "loss": 0.8703,
+      "step": 2112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.390717134666957e-06,
+      "loss": 0.7917,
+      "step": 2113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382549181127128e-06,
+      "loss": 0.9384,
+      "step": 2114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374381641078115e-06,
+      "loss": 0.848,
+      "step": 2115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36621451998951e-06,
+      "loss": 0.9047,
+      "step": 2116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358047823330617e-06,
+      "loss": 0.8838,
+      "step": 2117
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.349881556570469e-06,
+      "loss": 0.8415,
+      "step": 2118
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.341715725177793e-06,
+      "loss": 0.8941,
+      "step": 2119
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333550334621043e-06,
+      "loss": 0.7615,
+      "step": 2120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325385390368367e-06,
+      "loss": 0.8009,
+      "step": 2121
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317220897887617e-06,
+      "loss": 0.8017,
+      "step": 2122
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30905686264634e-06,
+      "loss": 0.826,
+      "step": 2123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300893290111785e-06,
+      "loss": 0.8403,
+      "step": 2124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292730185750877e-06,
+      "loss": 0.8381,
+      "step": 2125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284567555030237e-06,
+      "loss": 0.9123,
+      "step": 2126
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276405403416174e-06,
+      "loss": 0.9088,
+      "step": 2127
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26824373637466e-06,
+      "loss": 0.8817,
+      "step": 2128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26008255937136e-06,
+      "loss": 0.9228,
+      "step": 2129
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2519218778716e-06,
+      "loss": 0.9097,
+      "step": 2130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243761697340373e-06,
+      "loss": 0.8035,
+      "step": 2131
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23560202324235e-06,
+      "loss": 0.8329,
+      "step": 2132
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22744286104184e-06,
+      "loss": 0.9372,
+      "step": 2133
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219284216202833e-06,
+      "loss": 0.8476,
+      "step": 2134
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.211126094188963e-06,
+      "loss": 0.77,
+      "step": 2135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202968500463504e-06,
+      "loss": 0.9384,
+      "step": 2136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194811440489388e-06,
+      "loss": 0.8706,
+      "step": 2137
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.186654919729195e-06,
+      "loss": 0.8953,
+      "step": 2138
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178498943645125e-06,
+      "loss": 0.8984,
+      "step": 2139
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170343517699028e-06,
+      "loss": 0.8886,
+      "step": 2140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162188647352386e-06,
+      "loss": 0.9489,
+      "step": 2141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.154034338066295e-06,
+      "loss": 0.8091,
+      "step": 2142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145880595301495e-06,
+      "loss": 0.9293,
+      "step": 2143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137727424518328e-06,
+      "loss": 0.8404,
+      "step": 2144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129574831176762e-06,
+      "loss": 0.871,
+      "step": 2145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121422820736385e-06,
+      "loss": 0.8313,
+      "step": 2146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113271398656375e-06,
+      "loss": 0.8989,
+      "step": 2147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10512057039554e-06,
+      "loss": 0.904,
+      "step": 2148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096970341412275e-06,
+      "loss": 0.9116,
+      "step": 2149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.088820717164573e-06,
+      "loss": 0.8003,
+      "step": 2150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.080671703110028e-06,
+      "loss": 0.9254,
+      "step": 2151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072523304705828e-06,
+      "loss": 0.8415,
+      "step": 2152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06437552740874e-06,
+      "loss": 0.7938,
+      "step": 2153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056228376675118e-06,
+      "loss": 0.9016,
+      "step": 2154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048081857960904e-06,
+      "loss": 0.7548,
+      "step": 2155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.039935976721601e-06,
+      "loss": 0.8705,
+      "step": 2156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.031790738412307e-06,
+      "loss": 0.8539,
+      "step": 2157
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.023646148487668e-06,
+      "loss": 0.8679,
+      "step": 2158
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015502212401903e-06,
+      "loss": 0.8793,
+      "step": 2159
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007358935608805e-06,
+      "loss": 0.8702,
+      "step": 2160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999216323561704e-06,
+      "loss": 0.9293,
+      "step": 2161
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991074381713504e-06,
+      "loss": 0.8995,
+      "step": 2162
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982933115516653e-06,
+      "loss": 0.8553,
+      "step": 2163
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97479253042314e-06,
+      "loss": 0.7632,
+      "step": 2164
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966652631884506e-06,
+      "loss": 0.898,
+      "step": 2165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958513425351836e-06,
+      "loss": 0.8418,
+      "step": 2166
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95037491627574e-06,
+      "loss": 0.8033,
+      "step": 2167
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.942237110106365e-06,
+      "loss": 0.8154,
+      "step": 2168
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934100012293397e-06,
+      "loss": 0.9337,
+      "step": 2169
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925963628286029e-06,
+      "loss": 0.8112,
+      "step": 2170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917827963532998e-06,
+      "loss": 0.9957,
+      "step": 2171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90969302348254e-06,
+      "loss": 0.8255,
+      "step": 2172
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901558813582411e-06,
+      "loss": 0.9371,
+      "step": 2173
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.893425339279891e-06,
+      "loss": 0.8894,
+      "step": 2174
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885292606021746e-06,
+      "loss": 0.9534,
+      "step": 2175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877160619254264e-06,
+      "loss": 0.9912,
+      "step": 2176
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869029384423228e-06,
+      "loss": 0.8834,
+      "step": 2177
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860898906973904e-06,
+      "loss": 0.8738,
+      "step": 2178
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85276919235107e-06,
+      "loss": 0.847,
+      "step": 2179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.844640245998989e-06,
+      "loss": 0.8139,
+      "step": 2180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8365120733614e-06,
+      "loss": 0.9321,
+      "step": 2181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828384679881532e-06,
+      "loss": 0.8384,
+      "step": 2182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820258071002084e-06,
+      "loss": 0.8626,
+      "step": 2183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.812132252165241e-06,
+      "loss": 1.0075,
+      "step": 2184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804007228812656e-06,
+      "loss": 0.8947,
+      "step": 2185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795883006385441e-06,
+      "loss": 0.9208,
+      "step": 2186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787759590324177e-06,
+      "loss": 0.9037,
+      "step": 2187
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779636986068911e-06,
+      "loss": 0.824,
+      "step": 2188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771515199059133e-06,
+      "loss": 0.7805,
+      "step": 2189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7633942347338e-06,
+      "loss": 0.8232,
+      "step": 2190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755274098531312e-06,
+      "loss": 0.8958,
+      "step": 2191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747154795889505e-06,
+      "loss": 0.9078,
+      "step": 2192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739036332245672e-06,
+      "loss": 0.8391,
+      "step": 2193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730918713036542e-06,
+      "loss": 0.7847,
+      "step": 2194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.722801943698267e-06,
+      "loss": 0.834,
+      "step": 2195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714686029666441e-06,
+      "loss": 0.8445,
+      "step": 2196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706570976376076e-06,
+      "loss": 0.8105,
+      "step": 2197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698456789261617e-06,
+      "loss": 0.7812,
+      "step": 2198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.690343473756928e-06,
+      "loss": 0.8161,
+      "step": 2199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682231035295281e-06,
+      "loss": 0.8769,
+      "step": 2200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674119479309366e-06,
+      "loss": 0.8948,
+      "step": 2201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666008811231289e-06,
+      "loss": 0.8391,
+      "step": 2202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657899036492544e-06,
+      "loss": 0.8049,
+      "step": 2203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649790160524044e-06,
+      "loss": 0.829,
+      "step": 2204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641682188756091e-06,
+      "loss": 0.8818,
+      "step": 2205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633575126618381e-06,
+      "loss": 0.8398,
+      "step": 2206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.625468979540006e-06,
+      "loss": 0.8265,
+      "step": 2207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.617363752949447e-06,
+      "loss": 0.8912,
+      "step": 2208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.609259452274559e-06,
+      "loss": 0.8908,
+      "step": 2209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.601156082942584e-06,
+      "loss": 0.8982,
+      "step": 2210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.593053650380134e-06,
+      "loss": 0.8922,
+      "step": 2211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584952160013203e-06,
+      "loss": 0.8741,
+      "step": 2212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.8589,
+      "step": 2213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.568752027566696e-06,
+      "loss": 1.0073,
+      "step": 2214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560653396335926e-06,
+      "loss": 0.9452,
+      "step": 2215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552555728998286e-06,
+      "loss": 0.8117,
+      "step": 2216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.544459030976571e-06,
+      "loss": 0.8552,
+      "step": 2217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536363307692932e-06,
+      "loss": 0.8651,
+      "step": 2218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528268564568867e-06,
+      "loss": 0.9496,
+      "step": 2219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52017480702521e-06,
+      "loss": 0.8748,
+      "step": 2220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.512082040482148e-06,
+      "loss": 0.8329,
+      "step": 2221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503990270359188e-06,
+      "loss": 0.9652,
+      "step": 2222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.495899502075188e-06,
+      "loss": 1.0391,
+      "step": 2223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.487809741048322e-06,
+      "loss": 0.8823,
+      "step": 2224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47972099269609e-06,
+      "loss": 0.8076,
+      "step": 2225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47163326243532e-06,
+      "loss": 0.9296,
+      "step": 2226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46354655568216e-06,
+      "loss": 0.955,
+      "step": 2227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455460877852063e-06,
+      "loss": 0.9333,
+      "step": 2228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447376234359794e-06,
+      "loss": 0.8481,
+      "step": 2229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439292630619442e-06,
+      "loss": 0.865,
+      "step": 2230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.431210072044371e-06,
+      "loss": 0.9525,
+      "step": 2231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.423128564047273e-06,
+      "loss": 0.7374,
+      "step": 2232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415048112040122e-06,
+      "loss": 0.879,
+      "step": 2233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406968721434184e-06,
+      "loss": 0.7352,
+      "step": 2234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398890397640023e-06,
+      "loss": 0.8196,
+      "step": 2235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390813146067474e-06,
+      "loss": 0.9811,
+      "step": 2236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.382736972125672e-06,
+      "loss": 0.8719,
+      "step": 2237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37466188122302e-06,
+      "loss": 0.9366,
+      "step": 2238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36658787876719e-06,
+      "loss": 0.8952,
+      "step": 2239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358514970165139e-06,
+      "loss": 0.9127,
+      "step": 2240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350443160823088e-06,
+      "loss": 0.8901,
+      "step": 2241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342372456146512e-06,
+      "loss": 0.8011,
+      "step": 2242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.334302861540151e-06,
+      "loss": 0.7921,
+      "step": 2243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.326234382408014e-06,
+      "loss": 0.7762,
+      "step": 2244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31816702415334e-06,
+      "loss": 0.9425,
+      "step": 2245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310100792178636e-06,
+      "loss": 0.8551,
+      "step": 2246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302035691885653e-06,
+      "loss": 0.8747,
+      "step": 2247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29397172867537e-06,
+      "loss": 0.9009,
+      "step": 2248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285908907948021e-06,
+      "loss": 0.8135,
+      "step": 2249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277847235103064e-06,
+      "loss": 0.807,
+      "step": 2250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269786715539197e-06,
+      "loss": 0.8537,
+      "step": 2251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261727354654338e-06,
+      "loss": 0.8498,
+      "step": 2252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253669157845632e-06,
+      "loss": 0.8989,
+      "step": 2253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245612130509443e-06,
+      "loss": 0.8689,
+      "step": 2254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23755627804136e-06,
+      "loss": 0.7889,
+      "step": 2255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229501605836174e-06,
+      "loss": 0.8963,
+      "step": 2256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221448119287889e-06,
+      "loss": 0.8221,
+      "step": 2257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21339582378972e-06,
+      "loss": 0.7665,
+      "step": 2258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205344724734078e-06,
+      "loss": 0.8974,
+      "step": 2259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197294827512579e-06,
+      "loss": 0.9067,
+      "step": 2260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189246137516026e-06,
+      "loss": 0.9393,
+      "step": 2261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181198660134418e-06,
+      "loss": 0.8182,
+      "step": 2262
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173152400756948e-06,
+      "loss": 0.8668,
+      "step": 2263
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165107364771979e-06,
+      "loss": 0.8916,
+      "step": 2264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15706355756707e-06,
+      "loss": 0.8318,
+      "step": 2265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14902098452895e-06,
+      "loss": 0.9569,
+      "step": 2266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140979651043518e-06,
+      "loss": 0.8041,
+      "step": 2267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.132939562495845e-06,
+      "loss": 0.9764,
+      "step": 2268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12490072427018e-06,
+      "loss": 0.8499,
+      "step": 2269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116863141749914e-06,
+      "loss": 0.8499,
+      "step": 2270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108826820317612e-06,
+      "loss": 0.838,
+      "step": 2271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100791765354994e-06,
+      "loss": 0.9754,
+      "step": 2272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09275798224292e-06,
+      "loss": 0.8763,
+      "step": 2273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084725476361413e-06,
+      "loss": 0.8857,
+      "step": 2274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.076694253089632e-06,
+      "loss": 0.9643,
+      "step": 2275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068664317805877e-06,
+      "loss": 0.975,
+      "step": 2276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060635675887592e-06,
+      "loss": 0.9158,
+      "step": 2277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.052608332711345e-06,
+      "loss": 0.8262,
+      "step": 2278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.044582293652845e-06,
+      "loss": 0.7894,
+      "step": 2279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036557564086922e-06,
+      "loss": 0.8933,
+      "step": 2280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028534149387524e-06,
+      "loss": 0.9627,
+      "step": 2281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020512054927727e-06,
+      "loss": 0.8551,
+      "step": 2282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012491286079725e-06,
+      "loss": 0.9183,
+      "step": 2283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.004471848214812e-06,
+      "loss": 0.9773,
+      "step": 2284
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9964537467034e-06,
+      "loss": 0.8455,
+      "step": 2285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.988436986915005e-06,
+      "loss": 0.9444,
+      "step": 2286
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98042157421824e-06,
+      "loss": 0.9271,
+      "step": 2287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.972407513980827e-06,
+      "loss": 0.9405,
+      "step": 2288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964394811569565e-06,
+      "loss": 0.8978,
+      "step": 2289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.956383472350356e-06,
+      "loss": 0.8604,
+      "step": 2290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948373501688192e-06,
+      "loss": 0.8928,
+      "step": 2291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940364904947135e-06,
+      "loss": 0.8322,
+      "step": 2292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932357687490339e-06,
+      "loss": 0.8396,
+      "step": 2293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.924351854680033e-06,
+      "loss": 0.8353,
+      "step": 2294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916347411877507e-06,
+      "loss": 0.9134,
+      "step": 2295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908344364443136e-06,
+      "loss": 0.8399,
+      "step": 2296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900342717736354e-06,
+      "loss": 0.9303,
+      "step": 2297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892342477115652e-06,
+      "loss": 0.9171,
+      "step": 2298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884343647938588e-06,
+      "loss": 0.766,
+      "step": 2299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876346235561765e-06,
+      "loss": 0.893,
+      "step": 2300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868350245340846e-06,
+      "loss": 0.7591,
+      "step": 2301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860355682630544e-06,
+      "loss": 0.7963,
+      "step": 2302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852362552784603e-06,
+      "loss": 0.7928,
+      "step": 2303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844370861155814e-06,
+      "loss": 0.8373,
+      "step": 2304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836380613096015e-06,
+      "loss": 0.9061,
+      "step": 2305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828391813956058e-06,
+      "loss": 0.8286,
+      "step": 2306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.820404469085841e-06,
+      "loss": 0.8776,
+      "step": 2307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812418583834282e-06,
+      "loss": 0.8471,
+      "step": 2308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804434163549317e-06,
+      "loss": 0.9657,
+      "step": 2309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796451213577907e-06,
+      "loss": 0.9801,
+      "step": 2310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.788469739266032e-06,
+      "loss": 0.8711,
+      "step": 2311
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780489745958672e-06,
+      "loss": 0.8505,
+      "step": 2312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772511238999825e-06,
+      "loss": 0.8454,
+      "step": 2313
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764534223732487e-06,
+      "loss": 0.8195,
+      "step": 2314
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756558705498661e-06,
+      "loss": 0.7756,
+      "step": 2315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748584689639347e-06,
+      "loss": 0.8317,
+      "step": 2316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740612181494533e-06,
+      "loss": 0.8798,
+      "step": 2317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.732641186403202e-06,
+      "loss": 1.0025,
+      "step": 2318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724671709703328e-06,
+      "loss": 0.9043,
+      "step": 2319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716703756731854e-06,
+      "loss": 0.877,
+      "step": 2320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.708737332824718e-06,
+      "loss": 0.8295,
+      "step": 2321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700772443316829e-06,
+      "loss": 0.9021,
+      "step": 2322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692809093542058e-06,
+      "loss": 0.8544,
+      "step": 2323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.684847288833263e-06,
+      "loss": 0.8666,
+      "step": 2324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.676887034522258e-06,
+      "loss": 0.9427,
+      "step": 2325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668928335939816e-06,
+      "loss": 0.8306,
+      "step": 2326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660971198415673e-06,
+      "loss": 0.7974,
+      "step": 2327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653015627278513e-06,
+      "loss": 0.8622,
+      "step": 2328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.645061627855978e-06,
+      "loss": 0.8118,
+      "step": 2329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637109205474665e-06,
+      "loss": 0.8413,
+      "step": 2330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629158365460092e-06,
+      "loss": 0.8886,
+      "step": 2331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6212091131367335e-06,
+      "loss": 0.8068,
+      "step": 2332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613261453828005e-06,
+      "loss": 0.7973,
+      "step": 2333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.605315392856239e-06,
+      "loss": 0.8563,
+      "step": 2334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597370935542713e-06,
+      "loss": 0.9206,
+      "step": 2335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.589428087207624e-06,
+      "loss": 0.9706,
+      "step": 2336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.581486853170086e-06,
+      "loss": 0.7878,
+      "step": 2337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5735472387481444e-06,
+      "loss": 1.026,
+      "step": 2338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5656092492587466e-06,
+      "loss": 0.9693,
+      "step": 2339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.557672890017764e-06,
+      "loss": 0.8598,
+      "step": 2340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5497381663399716e-06,
+      "loss": 0.8519,
+      "step": 2341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541805083539041e-06,
+      "loss": 0.7625,
+      "step": 2342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533873646927557e-06,
+      "loss": 0.8268,
+      "step": 2343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525943861817001e-06,
+      "loss": 0.8875,
+      "step": 2344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51801573351774e-06,
+      "loss": 0.8606,
+      "step": 2345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510089267339032e-06,
+      "loss": 0.8446,
+      "step": 2346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.502164468589037e-06,
+      "loss": 0.9185,
+      "step": 2347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494241342574776e-06,
+      "loss": 0.8567,
+      "step": 2348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4863198946021695e-06,
+      "loss": 0.8524,
+      "step": 2349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478400129976005e-06,
+      "loss": 0.8222,
+      "step": 2350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470482053999937e-06,
+      "loss": 0.7935,
+      "step": 2351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462565671976504e-06,
+      "loss": 0.8594,
+      "step": 2352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4546509892070936e-06,
+      "loss": 0.8184,
+      "step": 2353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446738010991972e-06,
+      "loss": 0.8749,
+      "step": 2354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438826742630254e-06,
+      "loss": 0.7782,
+      "step": 2355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4309171894199054e-06,
+      "loss": 0.8955,
+      "step": 2356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423009356657751e-06,
+      "loss": 1.0468,
+      "step": 2357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415103249639471e-06,
+      "loss": 0.916,
+      "step": 2358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407198873659568e-06,
+      "loss": 0.8472,
+      "step": 2359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399296234011401e-06,
+      "loss": 0.9193,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39139533598717e-06,
+      "loss": 0.9093,
+      "step": 2361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383496184877892e-06,
+      "loss": 0.8198,
+      "step": 2362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375598785973429e-06,
+      "loss": 0.8558,
+      "step": 2363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.367703144562467e-06,
+      "loss": 0.7825,
+      "step": 2364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.359809265932505e-06,
+      "loss": 0.8278,
+      "step": 2365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351917155369877e-06,
+      "loss": 0.854,
+      "step": 2366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3440268181597165e-06,
+      "loss": 0.9137,
+      "step": 2367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336138259585985e-06,
+      "loss": 0.9024,
+      "step": 2368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328251484931445e-06,
+      "loss": 0.8368,
+      "step": 2369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3203664994776605e-06,
+      "loss": 0.8823,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312483308505003e-06,
+      "loss": 0.8953,
+      "step": 2371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.304601917292649e-06,
+      "loss": 0.8287,
+      "step": 2372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296722331118553e-06,
+      "loss": 0.8509,
+      "step": 2373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288844555259471e-06,
+      "loss": 0.8142,
+      "step": 2374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280968594990953e-06,
+      "loss": 0.7891,
+      "step": 2375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.273094455587314e-06,
+      "loss": 0.8846,
+      "step": 2376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.265222142321672e-06,
+      "loss": 0.8346,
+      "step": 2377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2573516604659034e-06,
+      "loss": 0.9145,
+      "step": 2378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.249483015290667e-06,
+      "loss": 0.9437,
+      "step": 2379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241616212065396e-06,
+      "loss": 0.7913,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.233751256058278e-06,
+      "loss": 0.8749,
+      "step": 2381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225888152536274e-06,
+      "loss": 0.8787,
+      "step": 2382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218026906765103e-06,
+      "loss": 0.8408,
+      "step": 2383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2101675240092305e-06,
+      "loss": 0.8232,
+      "step": 2384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.202310009531886e-06,
+      "loss": 0.8191,
+      "step": 2385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194454368595047e-06,
+      "loss": 0.8947,
+      "step": 2386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186600606459426e-06,
+      "loss": 0.8171,
+      "step": 2387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178748728384486e-06,
+      "loss": 0.8158,
+      "step": 2388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17089873962843e-06,
+      "loss": 0.8209,
+      "step": 2389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.163050645448185e-06,
+      "loss": 0.7698,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1552044510994254e-06,
+      "loss": 0.9138,
+      "step": 2391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147360161836536e-06,
+      "loss": 0.8901,
+      "step": 2392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139517782912637e-06,
+      "loss": 0.8122,
+      "step": 2393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131677319579571e-06,
+      "loss": 0.8703,
+      "step": 2394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123838777087884e-06,
+      "loss": 0.8893,
+      "step": 2395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116002160686851e-06,
+      "loss": 0.8399,
+      "step": 2396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.108167475624454e-06,
+      "loss": 0.9082,
+      "step": 2397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.100334727147369e-06,
+      "loss": 0.8387,
+      "step": 2398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.09250392050099e-06,
+      "loss": 0.8925,
+      "step": 2399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.084675060929409e-06,
+      "loss": 0.9043,
+      "step": 2400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076848153675405e-06,
+      "loss": 0.8379,
+      "step": 2401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.069023203980456e-06,
+      "loss": 0.7965,
+      "step": 2402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061200217084732e-06,
+      "loss": 0.8515,
+      "step": 2403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053379198227077e-06,
+      "loss": 0.8488,
+      "step": 2404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0455601526450345e-06,
+      "loss": 0.863,
+      "step": 2405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037743085574808e-06,
+      "loss": 0.8538,
+      "step": 2406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0299280022512875e-06,
+      "loss": 0.8359,
+      "step": 2407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022114907908036e-06,
+      "loss": 0.8682,
+      "step": 2408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014303807777274e-06,
+      "loss": 0.8381,
+      "step": 2409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006494707089896e-06,
+      "loss": 0.8364,
+      "step": 2410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9986876110754576e-06,
+      "loss": 0.9154,
+      "step": 2411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99088252496216e-06,
+      "loss": 0.8346,
+      "step": 2412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983079453976873e-06,
+      "loss": 1.0137,
+      "step": 2413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975278403345116e-06,
+      "loss": 0.8936,
+      "step": 2414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967479378291042e-06,
+      "loss": 0.7817,
+      "step": 2415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.95968238403746e-06,
+      "loss": 0.9036,
+      "step": 2416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951887425805812e-06,
+      "loss": 0.875,
+      "step": 2417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.944094508816182e-06,
+      "loss": 0.8656,
+      "step": 2418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936303638287286e-06,
+      "loss": 0.8843,
+      "step": 2419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9285148194364645e-06,
+      "loss": 0.8815,
+      "step": 2420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92072805747969e-06,
+      "loss": 0.8576,
+      "step": 2421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912943357631556e-06,
+      "loss": 0.8668,
+      "step": 2422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.905160725105271e-06,
+      "loss": 0.8207,
+      "step": 2423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897380165112667e-06,
+      "loss": 0.8888,
+      "step": 2424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889601682864182e-06,
+      "loss": 0.9111,
+      "step": 2425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881825283568859e-06,
+      "loss": 0.963,
+      "step": 2426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8740509724343555e-06,
+      "loss": 0.8344,
+      "step": 2427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.866278754666929e-06,
+      "loss": 0.9111,
+      "step": 2428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858508635471428e-06,
+      "loss": 0.8226,
+      "step": 2429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850740620051303e-06,
+      "loss": 0.7694,
+      "step": 2430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842974713608587e-06,
+      "loss": 0.8108,
+      "step": 2431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8352109213439084e-06,
+      "loss": 0.8911,
+      "step": 2432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8274492484564855e-06,
+      "loss": 0.8696,
+      "step": 2433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819689700144102e-06,
+      "loss": 0.85,
+      "step": 2434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811932281603125e-06,
+      "loss": 0.8706,
+      "step": 2435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.804176998028504e-06,
+      "loss": 0.9176,
+      "step": 2436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7964238546137425e-06,
+      "loss": 0.9293,
+      "step": 2437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.788672856550928e-06,
+      "loss": 0.8608,
+      "step": 2438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780924009030701e-06,
+      "loss": 0.8683,
+      "step": 2439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.773177317242257e-06,
+      "loss": 0.9191,
+      "step": 2440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76543278637336e-06,
+      "loss": 0.865,
+      "step": 2441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7576904216103275e-06,
+      "loss": 0.8692,
+      "step": 2442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.749950228138014e-06,
+      "loss": 0.7782,
+      "step": 2443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742212211139827e-06,
+      "loss": 0.8087,
+      "step": 2444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734476375797714e-06,
+      "loss": 0.8767,
+      "step": 2445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.726742727292164e-06,
+      "loss": 0.9145,
+      "step": 2446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.71901127080221e-06,
+      "loss": 0.8355,
+      "step": 2447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711282011505397e-06,
+      "loss": 0.9409,
+      "step": 2448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7035549545778115e-06,
+      "loss": 0.8528,
+      "step": 2449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695830105194071e-06,
+      "loss": 0.9156,
+      "step": 2450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688107468527297e-06,
+      "loss": 0.855,
+      "step": 2451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680387049749146e-06,
+      "loss": 0.8771,
+      "step": 2452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672668854029782e-06,
+      "loss": 0.833,
+      "step": 2453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6649528865378745e-06,
+      "loss": 0.9691,
+      "step": 2454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6572391524406165e-06,
+      "loss": 0.8346,
+      "step": 2455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.649527656903687e-06,
+      "loss": 0.8024,
+      "step": 2456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641818405091279e-06,
+      "loss": 0.9057,
+      "step": 2457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.634111402166081e-06,
+      "loss": 0.7551,
+      "step": 2458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626406653289268e-06,
+      "loss": 0.8422,
+      "step": 2459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618704163620511e-06,
+      "loss": 0.8498,
+      "step": 2460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611003938317975e-06,
+      "loss": 0.9107,
+      "step": 2461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603305982538295e-06,
+      "loss": 1.0446,
+      "step": 2462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5956103014365904e-06,
+      "loss": 0.8392,
+      "step": 2463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.587916900166469e-06,
+      "loss": 1.0212,
+      "step": 2464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58022578387999e-06,
+      "loss": 0.9558,
+      "step": 2465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572536957727704e-06,
+      "loss": 0.888,
+      "step": 2466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564850426858618e-06,
+      "loss": 0.8905,
+      "step": 2467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557166196420195e-06,
+      "loss": 0.8573,
+      "step": 2468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549484271558375e-06,
+      "loss": 0.7824,
+      "step": 2469
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541804657417534e-06,
+      "loss": 0.9105,
+      "step": 2470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.534127359140518e-06,
+      "loss": 0.8068,
+      "step": 2471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526452381868618e-06,
+      "loss": 0.8283,
+      "step": 2472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518779730741555e-06,
+      "loss": 0.9049,
+      "step": 2473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.511109410897511e-06,
+      "loss": 0.9207,
+      "step": 2474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503441427473108e-06,
+      "loss": 0.7676,
+      "step": 2475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.495775785603384e-06,
+      "loss": 0.9405,
+      "step": 2476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488112490421825e-06,
+      "loss": 0.8027,
+      "step": 2477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480451547060347e-06,
+      "loss": 0.8278,
+      "step": 2478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472792960649275e-06,
+      "loss": 0.8135,
+      "step": 2479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465136736317374e-06,
+      "loss": 0.8624,
+      "step": 2480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.457482879191819e-06,
+      "loss": 0.8266,
+      "step": 2481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.449831394398192e-06,
+      "loss": 0.9106,
+      "step": 2482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.442182287060503e-06,
+      "loss": 0.9092,
+      "step": 2483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434535562301153e-06,
+      "loss": 0.8181,
+      "step": 2484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426891225240962e-06,
+      "loss": 0.8773,
+      "step": 2485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419249280999144e-06,
+      "loss": 0.9057,
+      "step": 2486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411609734693305e-06,
+      "loss": 0.8051,
+      "step": 2487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.403972591439455e-06,
+      "loss": 0.8141,
+      "step": 2488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396337856351997e-06,
+      "loss": 0.9745,
+      "step": 2489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3887055345437065e-06,
+      "loss": 0.9803,
+      "step": 2490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381075631125754e-06,
+      "loss": 0.8458,
+      "step": 2491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.373448151207693e-06,
+      "loss": 0.9801,
+      "step": 2492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365823099897439e-06,
+      "loss": 0.949,
+      "step": 2493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358200482301303e-06,
+      "loss": 0.9099,
+      "step": 2494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350580303523947e-06,
+      "loss": 0.8861,
+      "step": 2495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342962568668407e-06,
+      "loss": 0.8354,
+      "step": 2496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335347282836089e-06,
+      "loss": 0.8703,
+      "step": 2497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327734451126743e-06,
+      "loss": 0.8947,
+      "step": 2498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3201240786384944e-06,
+      "loss": 0.8507,
+      "step": 2499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3125161704678105e-06,
+      "loss": 0.8691,
+      "step": 2500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.304910731709506e-06,
+      "loss": 0.9002,
+      "step": 2501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297307767456746e-06,
+      "loss": 0.7742,
+      "step": 2502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.289707282801048e-06,
+      "loss": 0.9294,
+      "step": 2503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.28210928283225e-06,
+      "loss": 0.9378,
+      "step": 2504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274513772638535e-06,
+      "loss": 1.0164,
+      "step": 2505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266920757306429e-06,
+      "loss": 0.8553,
+      "step": 2506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259330241920769e-06,
+      "loss": 0.8562,
+      "step": 2507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2517422315647315e-06,
+      "loss": 0.9431,
+      "step": 2508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.244156731319806e-06,
+      "loss": 0.8807,
+      "step": 2509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236573746265807e-06,
+      "loss": 0.7768,
+      "step": 2510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228993281480869e-06,
+      "loss": 0.8453,
+      "step": 2511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221415342041421e-06,
+      "loss": 0.9307,
+      "step": 2512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213839933022223e-06,
+      "loss": 0.8469,
+      "step": 2513
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206267059496327e-06,
+      "loss": 0.7712,
+      "step": 2514
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1986967265350875e-06,
+      "loss": 0.7827,
+      "step": 2515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191128939208162e-06,
+      "loss": 0.8742,
+      "step": 2516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183563702583506e-06,
+      "loss": 0.8761,
+      "step": 2517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.176001021727358e-06,
+      "loss": 0.8485,
+      "step": 2518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168440901704247e-06,
+      "loss": 0.9071,
+      "step": 2519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.160883347576996e-06,
+      "loss": 0.8122,
+      "step": 2520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153328364406697e-06,
+      "loss": 0.8178,
+      "step": 2521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145775957252733e-06,
+      "loss": 0.8753,
+      "step": 2522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13822613117275e-06,
+      "loss": 0.7719,
+      "step": 2523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13067889122267e-06,
+      "loss": 0.9306,
+      "step": 2524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123134242456691e-06,
+      "loss": 0.741,
+      "step": 2525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1155921899272596e-06,
+      "loss": 0.9041,
+      "step": 2526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108052738685102e-06,
+      "loss": 0.8359,
+      "step": 2527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100515893779188e-06,
+      "loss": 0.8817,
+      "step": 2528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0929816602567474e-06,
+      "loss": 0.8655,
+      "step": 2529
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085450043163261e-06,
+      "loss": 0.8614,
+      "step": 2530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077921047542463e-06,
+      "loss": 0.9394,
+      "step": 2531
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.070394678436322e-06,
+      "loss": 0.8506,
+      "step": 2532
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06287094088505e-06,
+      "loss": 0.7815,
+      "step": 2533
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.055349839927107e-06,
+      "loss": 0.8249,
+      "step": 2534
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04783138059917e-06,
+      "loss": 0.8805,
+      "step": 2535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0403155679361635e-06,
+      "loss": 0.8807,
+      "step": 2536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03280240697123e-06,
+      "loss": 0.8628,
+      "step": 2537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025291902735737e-06,
+      "loss": 0.9184,
+      "step": 2538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.01778406025928e-06,
+      "loss": 0.85,
+      "step": 2539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.010278884569658e-06,
+      "loss": 0.7838,
+      "step": 2540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002776380692901e-06,
+      "loss": 0.8822,
+      "step": 2541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995276553653239e-06,
+      "loss": 0.8668,
+      "step": 2542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.987779408473109e-06,
+      "loss": 0.8125,
+      "step": 2543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.980284950173154e-06,
+      "loss": 0.8544,
+      "step": 2544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972793183772227e-06,
+      "loss": 0.8254,
+      "step": 2545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9653041142873625e-06,
+      "loss": 0.8811,
+      "step": 2546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9578177467338005e-06,
+      "loss": 0.9182,
+      "step": 2547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95033408612496e-06,
+      "loss": 0.9154,
+      "step": 2548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94285313747246e-06,
+      "loss": 0.755,
+      "step": 2549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.935374905786102e-06,
+      "loss": 0.8986,
+      "step": 2550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927899396073857e-06,
+      "loss": 0.825,
+      "step": 2551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920426613341881e-06,
+      "loss": 0.8824,
+      "step": 2552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.912956562594509e-06,
+      "loss": 0.9225,
+      "step": 2553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905489248834232e-06,
+      "loss": 0.8904,
+      "step": 2554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898024677061722e-06,
+      "loss": 0.8992,
+      "step": 2555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.890562852275812e-06,
+      "loss": 0.8546,
+      "step": 2556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883103779473484e-06,
+      "loss": 0.9029,
+      "step": 2557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8756474636498894e-06,
+      "loss": 0.8109,
+      "step": 2558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868193909798337e-06,
+      "loss": 0.7997,
+      "step": 2559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860743122910268e-06,
+      "loss": 0.8958,
+      "step": 2560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8532951079752895e-06,
+      "loss": 0.7943,
+      "step": 2561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.8923,
+      "step": 2562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838407413913698e-06,
+      "loss": 0.7415,
+      "step": 2563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830967744756995e-06,
+      "loss": 0.8008,
+      "step": 2564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8235308674931736e-06,
+      "loss": 0.8361,
+      "step": 2565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816096787102522e-06,
+      "loss": 0.7932,
+      "step": 2566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.808665508563455e-06,
+      "loss": 0.7888,
+      "step": 2567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801237036852503e-06,
+      "loss": 0.8126,
+      "step": 2568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.793811376944321e-06,
+      "loss": 0.839,
+      "step": 2569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7863885338116846e-06,
+      "loss": 0.9384,
+      "step": 2570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7789685124254775e-06,
+      "loss": 0.9255,
+      "step": 2571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.771551317754691e-06,
+      "loss": 0.8481,
+      "step": 2572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764136954766437e-06,
+      "loss": 0.8294,
+      "step": 2573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756725428425914e-06,
+      "loss": 0.9216,
+      "step": 2574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.749316743696438e-06,
+      "loss": 0.9081,
+      "step": 2575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7419109055394025e-06,
+      "loss": 0.862,
+      "step": 2576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734507918914312e-06,
+      "loss": 0.7603,
+      "step": 2577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7271077887787575e-06,
+      "loss": 0.937,
+      "step": 2578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719710520088407e-06,
+      "loss": 0.7887,
+      "step": 2579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.712316117797028e-06,
+      "loss": 0.9163,
+      "step": 2580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704924586856455e-06,
+      "loss": 0.8132,
+      "step": 2581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697535932216602e-06,
+      "loss": 0.8596,
+      "step": 2582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690150158825462e-06,
+      "loss": 0.8689,
+      "step": 2583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682767271629104e-06,
+      "loss": 0.9092,
+      "step": 2584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6753872755716435e-06,
+      "loss": 0.8636,
+      "step": 2585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668010175595282e-06,
+      "loss": 0.8967,
+      "step": 2586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660635976640265e-06,
+      "loss": 0.889,
+      "step": 2587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653264683644908e-06,
+      "loss": 0.8582,
+      "step": 2588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645896301545574e-06,
+      "loss": 0.95,
+      "step": 2589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6385308352766795e-06,
+      "loss": 0.8382,
+      "step": 2590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631168289770679e-06,
+      "loss": 0.8569,
+      "step": 2591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623808669958088e-06,
+      "loss": 0.8345,
+      "step": 2592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616451980767443e-06,
+      "loss": 0.9671,
+      "step": 2593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609098227125334e-06,
+      "loss": 0.8929,
+      "step": 2594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601747413956382e-06,
+      "loss": 0.852,
+      "step": 2595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.594399546183231e-06,
+      "loss": 0.8226,
+      "step": 2596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587054628726558e-06,
+      "loss": 0.8459,
+      "step": 2597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579712666505071e-06,
+      "loss": 0.7965,
+      "step": 2598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572373664435485e-06,
+      "loss": 0.863,
+      "step": 2599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565037627432541e-06,
+      "loss": 0.812,
+      "step": 2600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55770456040899e-06,
+      "loss": 0.8517,
+      "step": 2601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.550374468275599e-06,
+      "loss": 0.8657,
+      "step": 2602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543047355941146e-06,
+      "loss": 0.8969,
+      "step": 2603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5357232283123995e-06,
+      "loss": 0.8249,
+      "step": 2604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528402090294142e-06,
+      "loss": 0.8357,
+      "step": 2605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521083946789156e-06,
+      "loss": 0.9481,
+      "step": 2606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.513768802698203e-06,
+      "loss": 0.841,
+      "step": 2607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5064566629200525e-06,
+      "loss": 0.8977,
+      "step": 2608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.499147532351452e-06,
+      "loss": 0.9011,
+      "step": 2609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491841415887134e-06,
+      "loss": 0.8381,
+      "step": 2610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484538318419817e-06,
+      "loss": 0.8169,
+      "step": 2611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477238244840202e-06,
+      "loss": 0.8843,
+      "step": 2612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469941200036951e-06,
+      "loss": 0.8622,
+      "step": 2613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.462647188896713e-06,
+      "loss": 0.8926,
+      "step": 2614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.455356216304091e-06,
+      "loss": 0.9537,
+      "step": 2615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448068287141663e-06,
+      "loss": 0.8687,
+      "step": 2616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440783406289973e-06,
+      "loss": 0.8618,
+      "step": 2617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433501578627511e-06,
+      "loss": 0.8948,
+      "step": 2618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426222809030726e-06,
+      "loss": 0.9054,
+      "step": 2619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4189471023740285e-06,
+      "loss": 0.9505,
+      "step": 2620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41167446352976e-06,
+      "loss": 0.8842,
+      "step": 2621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404404897368226e-06,
+      "loss": 0.8872,
+      "step": 2622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397138408757669e-06,
+      "loss": 0.8557,
+      "step": 2623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3898750025642624e-06,
+      "loss": 0.9682,
+      "step": 2624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3826146836521255e-06,
+      "loss": 0.8758,
+      "step": 2625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375357456883299e-06,
+      "loss": 0.7783,
+      "step": 2626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368103327117768e-06,
+      "loss": 0.8128,
+      "step": 2627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360852299213434e-06,
+      "loss": 0.89,
+      "step": 2628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3536043780261135e-06,
+      "loss": 0.9323,
+      "step": 2629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346359568409561e-06,
+      "loss": 0.8603,
+      "step": 2630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.339117875215436e-06,
+      "loss": 0.7945,
+      "step": 2631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331879303293308e-06,
+      "loss": 0.7858,
+      "step": 2632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324643857490666e-06,
+      "loss": 0.866,
+      "step": 2633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.317411542652901e-06,
+      "loss": 0.8574,
+      "step": 2634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310182363623304e-06,
+      "loss": 0.849,
+      "step": 2635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302956325243072e-06,
+      "loss": 0.7868,
+      "step": 2636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295733432351294e-06,
+      "loss": 0.8859,
+      "step": 2637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288513689784951e-06,
+      "loss": 0.8429,
+      "step": 2638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281297102378924e-06,
+      "loss": 0.9156,
+      "step": 2639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2740836749659664e-06,
+      "loss": 0.8185,
+      "step": 2640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26687341237673e-06,
+      "loss": 0.7992,
+      "step": 2641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25966631943974e-06,
+      "loss": 0.8212,
+      "step": 2642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252462400981397e-06,
+      "loss": 0.8419,
+      "step": 2643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245261661825982e-06,
+      "loss": 0.8173,
+      "step": 2644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238064106795644e-06,
+      "loss": 0.7901,
+      "step": 2645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230869740710398e-06,
+      "loss": 0.8099,
+      "step": 2646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22367856838812e-06,
+      "loss": 0.904,
+      "step": 2647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2164905946445585e-06,
+      "loss": 0.8602,
+      "step": 2648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209305824293307e-06,
+      "loss": 0.8476,
+      "step": 2649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2021242621458225e-06,
+      "loss": 0.8634,
+      "step": 2650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194945913011418e-06,
+      "loss": 0.8066,
+      "step": 2651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1877707816972355e-06,
+      "loss": 0.8255,
+      "step": 2652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180598873008285e-06,
+      "loss": 0.8391,
+      "step": 2653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173430191747402e-06,
+      "loss": 0.8779,
+      "step": 2654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16626474271527e-06,
+      "loss": 0.8301,
+      "step": 2655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159102530710405e-06,
+      "loss": 0.8699,
+      "step": 2656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.151943560529147e-06,
+      "loss": 0.8599,
+      "step": 2657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14478783696568e-06,
+      "loss": 0.9228,
+      "step": 2658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1376353648120095e-06,
+      "loss": 0.8959,
+      "step": 2659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130486148857952e-06,
+      "loss": 0.9194,
+      "step": 2660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.123340193891158e-06,
+      "loss": 0.8128,
+      "step": 2661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.116197504697089e-06,
+      "loss": 0.8675,
+      "step": 2662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109058086059014e-06,
+      "loss": 0.82,
+      "step": 2663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101921942758024e-06,
+      "loss": 0.8883,
+      "step": 2664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.094789079573006e-06,
+      "loss": 0.9634,
+      "step": 2665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.087659501280647e-06,
+      "loss": 0.8168,
+      "step": 2666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0805332126554515e-06,
+      "loss": 0.9453,
+      "step": 2667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.073410218469702e-06,
+      "loss": 0.8859,
+      "step": 2668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066290523493487e-06,
+      "loss": 0.9343,
+      "step": 2669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0591741324946846e-06,
+      "loss": 0.8327,
+      "step": 2670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05206105023895e-06,
+      "loss": 0.7998,
+      "step": 2671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044951281489735e-06,
+      "loss": 0.9994,
+      "step": 2672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03784483100827e-06,
+      "loss": 0.8211,
+      "step": 2673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030741703553559e-06,
+      "loss": 0.8295,
+      "step": 2674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023641903882376e-06,
+      "loss": 0.8614,
+      "step": 2675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016545436749283e-06,
+      "loss": 0.8901,
+      "step": 2676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00945230690659e-06,
+      "loss": 0.7922,
+      "step": 2677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0023625191043916e-06,
+      "loss": 0.83,
+      "step": 2678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.995276078090528e-06,
+      "loss": 0.8036,
+      "step": 2679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.988192988610606e-06,
+      "loss": 0.8932,
+      "step": 2680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9811132554079946e-06,
+      "loss": 0.9451,
+      "step": 2681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974036883223798e-06,
+      "loss": 0.924,
+      "step": 2682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966963876796888e-06,
+      "loss": 0.9829,
+      "step": 2683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959894240863869e-06,
+      "loss": 0.8341,
+      "step": 2684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9528279801590886e-06,
+      "loss": 0.9043,
+      "step": 2685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.945765099414644e-06,
+      "loss": 0.8397,
+      "step": 2686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938705603360365e-06,
+      "loss": 0.8938,
+      "step": 2687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931649496723809e-06,
+      "loss": 0.997,
+      "step": 2688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924596784230268e-06,
+      "loss": 0.8014,
+      "step": 2689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917547470602767e-06,
+      "loss": 0.7905,
+      "step": 2690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91050156056204e-06,
+      "loss": 0.7847,
+      "step": 2691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9034590588265585e-06,
+      "loss": 0.8553,
+      "step": 2692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.896419970112499e-06,
+      "loss": 0.8604,
+      "step": 2693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889384299133755e-06,
+      "loss": 0.8351,
+      "step": 2694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882352050601938e-06,
+      "loss": 0.9196,
+      "step": 2695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875323229226358e-06,
+      "loss": 0.929,
+      "step": 2696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868297839714034e-06,
+      "loss": 0.8587,
+      "step": 2697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861275886769695e-06,
+      "loss": 0.7969,
+      "step": 2698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854257375095752e-06,
+      "loss": 0.9687,
+      "step": 2699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8472423093923245e-06,
+      "loss": 0.8898,
+      "step": 2700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840230694357224e-06,
+      "loss": 0.8408,
+      "step": 2701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833222534685944e-06,
+      "loss": 0.8138,
+      "step": 2702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826217835071666e-06,
+      "loss": 0.8043,
+      "step": 2703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.819216600205254e-06,
+      "loss": 0.8562,
+      "step": 2704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.812218834775255e-06,
+      "loss": 0.803,
+      "step": 2705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8052245434678976e-06,
+      "loss": 0.8005,
+      "step": 2706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798233730967069e-06,
+      "loss": 0.8058,
+      "step": 2707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7912464019543385e-06,
+      "loss": 0.8571,
+      "step": 2708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784262561108942e-06,
+      "loss": 0.9647,
+      "step": 2709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777282213107771e-06,
+      "loss": 0.8332,
+      "step": 2710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770305362625391e-06,
+      "loss": 0.8061,
+      "step": 2711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.763332014334015e-06,
+      "loss": 0.8044,
+      "step": 2712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756362172903508e-06,
+      "loss": 0.8698,
+      "step": 2713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749395843001397e-06,
+      "loss": 0.7993,
+      "step": 2714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742433029292856e-06,
+      "loss": 0.8433,
+      "step": 2715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735473736440694e-06,
+      "loss": 0.8548,
+      "step": 2716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728517969105376e-06,
+      "loss": 0.7571,
+      "step": 2717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7215657319449924e-06,
+      "loss": 0.7587,
+      "step": 2718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714617029615279e-06,
+      "loss": 0.8232,
+      "step": 2719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707671866769607e-06,
+      "loss": 0.932,
+      "step": 2720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700730248058966e-06,
+      "loss": 0.8914,
+      "step": 2721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693792178131979e-06,
+      "loss": 0.8708,
+      "step": 2722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686857661634892e-06,
+      "loss": 0.825,
+      "step": 2723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67992670321157e-06,
+      "loss": 0.8763,
+      "step": 2724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672999307503498e-06,
+      "loss": 0.837,
+      "step": 2725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6660754791497755e-06,
+      "loss": 0.7873,
+      "step": 2726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659155222787105e-06,
+      "loss": 0.9093,
+      "step": 2727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652238543049806e-06,
+      "loss": 0.8256,
+      "step": 2728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645325444569805e-06,
+      "loss": 0.8448,
+      "step": 2729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.638415931976619e-06,
+      "loss": 0.8068,
+      "step": 2730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631510009897371e-06,
+      "loss": 0.784,
+      "step": 2731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624607682956773e-06,
+      "loss": 0.8556,
+      "step": 2732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.61770895577714e-06,
+      "loss": 0.9072,
+      "step": 2733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610813832978373e-06,
+      "loss": 0.8106,
+      "step": 2734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6039223191779526e-06,
+      "loss": 0.8803,
+      "step": 2735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597034418990947e-06,
+      "loss": 0.8399,
+      "step": 2736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590150137030009e-06,
+      "loss": 0.8096,
+      "step": 2737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.583269477905359e-06,
+      "loss": 0.936,
+      "step": 2738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576392446224801e-06,
+      "loss": 0.9234,
+      "step": 2739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569519046593702e-06,
+      "loss": 0.8765,
+      "step": 2740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.562649283614996e-06,
+      "loss": 0.7461,
+      "step": 2741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.555783161889192e-06,
+      "loss": 0.9425,
+      "step": 2742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548920686014348e-06,
+      "loss": 0.8227,
+      "step": 2743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.542061860586089e-06,
+      "loss": 0.9776,
+      "step": 2744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535206690197596e-06,
+      "loss": 0.8526,
+      "step": 2745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.528355179439593e-06,
+      "loss": 0.9349,
+      "step": 2746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.521507332900361e-06,
+      "loss": 0.7756,
+      "step": 2747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514663155165731e-06,
+      "loss": 0.9554,
+      "step": 2748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507822650819066e-06,
+      "loss": 0.8278,
+      "step": 2749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500985824441272e-06,
+      "loss": 0.9067,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.494152680610802e-06,
+      "loss": 1.0006,
+      "step": 2751
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487323223903627e-06,
+      "loss": 0.9426,
+      "step": 2752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480497458893262e-06,
+      "loss": 0.8355,
+      "step": 2753
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473675390150748e-06,
+      "loss": 0.8552,
+      "step": 2754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466857022244641e-06,
+      "loss": 0.8694,
+      "step": 2755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4600423597410315e-06,
+      "loss": 0.8315,
+      "step": 2756
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453231407203516e-06,
+      "loss": 0.85,
+      "step": 2757
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4464241691932194e-06,
+      "loss": 0.9228,
+      "step": 2758
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439620650268771e-06,
+      "loss": 0.8912,
+      "step": 2759
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432820854986306e-06,
+      "loss": 0.8796,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426024787899474e-06,
+      "loss": 0.8862,
+      "step": 2761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.41923245355943e-06,
+      "loss": 0.9053,
+      "step": 2762
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412443856514817e-06,
+      "loss": 0.9161,
+      "step": 2763
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405659001311785e-06,
+      "loss": 0.839,
+      "step": 2764
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398877892493981e-06,
+      "loss": 0.812,
+      "step": 2765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3921005346025296e-06,
+      "loss": 0.8956,
+      "step": 2766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385326932176058e-06,
+      "loss": 0.8976,
+      "step": 2767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3785570897506715e-06,
+      "loss": 0.9552,
+      "step": 2768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371791011859953e-06,
+      "loss": 0.8646,
+      "step": 2769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365028703034976e-06,
+      "loss": 0.9127,
+      "step": 2770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358270167804277e-06,
+      "loss": 0.8867,
+      "step": 2771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351515410693876e-06,
+      "loss": 0.7706,
+      "step": 2772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344764436227262e-06,
+      "loss": 0.9097,
+      "step": 2773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.338017248925381e-06,
+      "loss": 0.8074,
+      "step": 2774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331273853306651e-06,
+      "loss": 0.9507,
+      "step": 2775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3245342538869565e-06,
+      "loss": 0.8028,
+      "step": 2776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317798455179625e-06,
+      "loss": 0.8564,
+      "step": 2777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3110664616954466e-06,
+      "loss": 0.8352,
+      "step": 2778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3043382779426675e-06,
+      "loss": 0.758,
+      "step": 2779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297613908426972e-06,
+      "loss": 0.8851,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290893357651502e-06,
+      "loss": 0.7694,
+      "step": 2781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284176630116832e-06,
+      "loss": 0.8681,
+      "step": 2782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277463730320981e-06,
+      "loss": 0.9508,
+      "step": 2783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270754662759409e-06,
+      "loss": 0.924,
+      "step": 2784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264049431924997e-06,
+      "loss": 0.7687,
+      "step": 2785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257348042308071e-06,
+      "loss": 0.71,
+      "step": 2786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250650498396377e-06,
+      "loss": 0.8784,
+      "step": 2787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24395680467508e-06,
+      "loss": 0.824,
+      "step": 2788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237266965626779e-06,
+      "loss": 0.8885,
+      "step": 2789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230580985731488e-06,
+      "loss": 0.8622,
+      "step": 2790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.223898869466628e-06,
+      "loss": 0.9287,
+      "step": 2791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217220621307043e-06,
+      "loss": 0.9088,
+      "step": 2792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210546245724987e-06,
+      "loss": 0.8839,
+      "step": 2793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203875747190106e-06,
+      "loss": 0.8165,
+      "step": 2794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19720913016947e-06,
+      "loss": 0.8504,
+      "step": 2795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190546399127537e-06,
+      "loss": 0.8633,
+      "step": 2796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18388755852616e-06,
+      "loss": 0.8228,
+      "step": 2797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177232612824599e-06,
+      "loss": 0.7865,
+      "step": 2798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1705815664794916e-06,
+      "loss": 0.804,
+      "step": 2799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1639344239448755e-06,
+      "loss": 0.9674,
+      "step": 2800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157291189672173e-06,
+      "loss": 0.8019,
+      "step": 2801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150651868110181e-06,
+      "loss": 0.7841,
+      "step": 2802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144016463705081e-06,
+      "loss": 0.8941,
+      "step": 2803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137384980900436e-06,
+      "loss": 0.9063,
+      "step": 2804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130757424137176e-06,
+      "loss": 0.911,
+      "step": 2805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124133797853601e-06,
+      "loss": 0.8102,
+      "step": 2806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117514106485386e-06,
+      "loss": 0.8726,
+      "step": 2807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110898354465564e-06,
+      "loss": 0.8901,
+      "step": 2808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104286546224535e-06,
+      "loss": 0.7854,
+      "step": 2809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097678686190052e-06,
+      "loss": 0.8985,
+      "step": 2810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091074778787229e-06,
+      "loss": 0.8373,
+      "step": 2811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.084474828438538e-06,
+      "loss": 0.901,
+      "step": 2812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0778788395637855e-06,
+      "loss": 0.8492,
+      "step": 2813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071286816580142e-06,
+      "loss": 0.8684,
+      "step": 2814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064698763902112e-06,
+      "loss": 0.8802,
+      "step": 2815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058114685941538e-06,
+      "loss": 0.8799,
+      "step": 2816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051534587107611e-06,
+      "loss": 0.9684,
+      "step": 2817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0449584718068565e-06,
+      "loss": 0.8024,
+      "step": 2818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038386344443122e-06,
+      "loss": 0.9462,
+      "step": 2819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031818209417596e-06,
+      "loss": 0.793,
+      "step": 2820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025254071128784e-06,
+      "loss": 0.862,
+      "step": 2821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018693933972519e-06,
+      "loss": 0.8632,
+      "step": 2822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012137802341963e-06,
+      "loss": 0.8882,
+      "step": 2823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005585680627582e-06,
+      "loss": 0.9271,
+      "step": 2824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999037573217157e-06,
+      "loss": 0.9276,
+      "step": 2825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9924934844957955e-06,
+      "loss": 0.9077,
+      "step": 2826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985953418845896e-06,
+      "loss": 0.8124,
+      "step": 2827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979417380647173e-06,
+      "loss": 0.8265,
+      "step": 2828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97288537427665e-06,
+      "loss": 0.8,
+      "step": 2829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.966357404108629e-06,
+      "loss": 0.8777,
+      "step": 2830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959833474514732e-06,
+      "loss": 0.785,
+      "step": 2831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953313589863865e-06,
+      "loss": 0.9574,
+      "step": 2832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946797754522225e-06,
+      "loss": 0.831,
+      "step": 2833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9402859728532965e-06,
+      "loss": 0.9204,
+      "step": 2834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.933778249217846e-06,
+      "loss": 0.867,
+      "step": 2835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927274587973935e-06,
+      "loss": 0.9023,
+      "step": 2836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920774993476898e-06,
+      "loss": 0.8385,
+      "step": 2837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914279470079338e-06,
+      "loss": 0.8647,
+      "step": 2838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907788022131143e-06,
+      "loss": 0.9185,
+      "step": 2839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90130065397947e-06,
+      "loss": 0.9437,
+      "step": 2840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894817369968736e-06,
+      "loss": 0.8399,
+      "step": 2841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888338174440632e-06,
+      "loss": 0.845,
+      "step": 2842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881863071734108e-06,
+      "loss": 0.8147,
+      "step": 2843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875392066185366e-06,
+      "loss": 0.8007,
+      "step": 2844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8689251621278765e-06,
+      "loss": 0.8152,
+      "step": 2845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.86246236389236e-06,
+      "loss": 0.8045,
+      "step": 2846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856003675806777e-06,
+      "loss": 0.8215,
+      "step": 2847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849549102196351e-06,
+      "loss": 0.8532,
+      "step": 2848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843098647383539e-06,
+      "loss": 0.8895,
+      "step": 2849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836652315688042e-06,
+      "loss": 0.7982,
+      "step": 2850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830210111426807e-06,
+      "loss": 0.9673,
+      "step": 2851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823772038914009e-06,
+      "loss": 0.9167,
+      "step": 2852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817338102461054e-06,
+      "loss": 0.8774,
+      "step": 2853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8109083063765884e-06,
+      "loss": 0.9114,
+      "step": 2854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8044826549664736e-06,
+      "loss": 0.8281,
+      "step": 2855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7980611525338063e-06,
+      "loss": 0.9317,
+      "step": 2856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7916438033789037e-06,
+      "loss": 0.8672,
+      "step": 2857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78523061179929e-06,
+      "loss": 0.8327,
+      "step": 2858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778821582089721e-06,
+      "loss": 0.8617,
+      "step": 2859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.772416718542151e-06,
+      "loss": 0.8843,
+      "step": 2860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7660160254457587e-06,
+      "loss": 0.8497,
+      "step": 2861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.759619507086919e-06,
+      "loss": 0.8447,
+      "step": 2862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7532271677492103e-06,
+      "loss": 0.9,
+      "step": 2863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7468390117134203e-06,
+      "loss": 0.8344,
+      "step": 2864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7404550432575337e-06,
+      "loss": 0.7859,
+      "step": 2865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7340752666567248e-06,
+      "loss": 0.7784,
+      "step": 2866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.727699686183366e-06,
+      "loss": 0.8399,
+      "step": 2867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.721328306107023e-06,
+      "loss": 0.8919,
+      "step": 2868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7149611306944356e-06,
+      "loss": 0.839,
+      "step": 2869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7085981642095437e-06,
+      "loss": 0.7487,
+      "step": 2870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7022394109134585e-06,
+      "loss": 0.9097,
+      "step": 2871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6958848750644685e-06,
+      "loss": 0.8113,
+      "step": 2872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6895345609180478e-06,
+      "loss": 0.8372,
+      "step": 2873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683188472726832e-06,
+      "loss": 0.8748,
+      "step": 2874
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6768466147406333e-06,
+      "loss": 0.8908,
+      "step": 2875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.670508991206434e-06,
+      "loss": 0.8967,
+      "step": 2876
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6641756063683698e-06,
+      "loss": 0.7862,
+      "step": 2877
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.657846464467748e-06,
+      "loss": 0.8432,
+      "step": 2878
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.651521569743035e-06,
+      "loss": 0.8513,
+      "step": 2879
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645200926429844e-06,
+      "loss": 0.7961,
+      "step": 2880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6388845387609427e-06,
+      "loss": 0.9339,
+      "step": 2881
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63257241096626e-06,
+      "loss": 0.7812,
+      "step": 2882
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6262645472728574e-06,
+      "loss": 0.9407,
+      "step": 2883
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6199609519049505e-06,
+      "loss": 0.8848,
+      "step": 2884
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6136616290838976e-06,
+      "loss": 0.7421,
+      "step": 2885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.607366583028186e-06,
+      "loss": 0.8518,
+      "step": 2886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.601075817953451e-06,
+      "loss": 0.8728,
+      "step": 2887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5947893380724475e-06,
+      "loss": 0.9402,
+      "step": 2888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588507147595076e-06,
+      "loss": 0.8155,
+      "step": 2889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.582229250728353e-06,
+      "loss": 0.814,
+      "step": 2890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5759556516764205e-06,
+      "loss": 0.8706,
+      "step": 2891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.569686354640547e-06,
+      "loss": 0.862,
+      "step": 2892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5634213638191238e-06,
+      "loss": 0.8984,
+      "step": 2893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5571606834076446e-06,
+      "loss": 0.8707,
+      "step": 2894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.550904317598729e-06,
+      "loss": 0.8862,
+      "step": 2895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5446522705821064e-06,
+      "loss": 0.8746,
+      "step": 2896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5384045465446037e-06,
+      "loss": 0.9534,
+      "step": 2897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5321611496701668e-06,
+      "loss": 0.812,
+      "step": 2898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5259220841398333e-06,
+      "loss": 0.8295,
+      "step": 2899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.519687354131741e-06,
+      "loss": 0.9408,
+      "step": 2900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5134569638211325e-06,
+      "loss": 0.8068,
+      "step": 2901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.507230917380332e-06,
+      "loss": 0.8381,
+      "step": 2902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.501009218978767e-06,
+      "loss": 0.7936,
+      "step": 2903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4947918727829487e-06,
+      "loss": 0.8134,
+      "step": 2904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488578882956468e-06,
+      "loss": 0.8328,
+      "step": 2905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4823702536600046e-06,
+      "loss": 0.8631,
+      "step": 2906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.476165989051321e-06,
+      "loss": 0.7061,
+      "step": 2907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4699660932852497e-06,
+      "loss": 0.8038,
+      "step": 2908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.463770570513697e-06,
+      "loss": 0.8565,
+      "step": 2909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.457579424885649e-06,
+      "loss": 0.9403,
+      "step": 2910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.8503,
+      "step": 2911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4452102816413225e-06,
+      "loss": 0.8995,
+      "step": 2912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4390322923083385e-06,
+      "loss": 0.8477,
+      "step": 2913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4328586966854405e-06,
+      "loss": 0.7934,
+      "step": 2914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.42668949890693e-06,
+      "loss": 0.8878,
+      "step": 2915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4205247031041512e-06,
+      "loss": 0.7875,
+      "step": 2916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.414364313405515e-06,
+      "loss": 0.873,
+      "step": 2917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4082083339364715e-06,
+      "loss": 0.9063,
+      "step": 2918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4020567688195174e-06,
+      "loss": 0.9507,
+      "step": 2919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3959096221741995e-06,
+      "loss": 0.8242,
+      "step": 2920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3897668981171062e-06,
+      "loss": 0.728,
+      "step": 2921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3836286007618545e-06,
+      "loss": 0.832,
+      "step": 2922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3774947342191054e-06,
+      "loss": 0.8498,
+      "step": 2923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3713653025965544e-06,
+      "loss": 0.8575,
+      "step": 2924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.365240309998916e-06,
+      "loss": 0.8833,
+      "step": 2925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.359119760527947e-06,
+      "loss": 0.9062,
+      "step": 2926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3530036582824155e-06,
+      "loss": 0.8151,
+      "step": 2927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3468920073581137e-06,
+      "loss": 0.8776,
+      "step": 2928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3407848118478646e-06,
+      "loss": 0.8223,
+      "step": 2929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33468207584149e-06,
+      "loss": 0.7422,
+      "step": 2930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3285838034258378e-06,
+      "loss": 0.8815,
+      "step": 2931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.322489998684767e-06,
+      "loss": 0.8695,
+      "step": 2932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316400665699134e-06,
+      "loss": 0.781,
+      "step": 2933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3103158085468102e-06,
+      "loss": 0.8706,
+      "step": 2934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3042354313026702e-06,
+      "loss": 0.893,
+      "step": 2935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2981595380385824e-06,
+      "loss": 0.9538,
+      "step": 2936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2920881328234146e-06,
+      "loss": 0.9534,
+      "step": 2937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2860212197230256e-06,
+      "loss": 0.9828,
+      "step": 2938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.279958802800275e-06,
+      "loss": 0.9184,
+      "step": 2939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2739008861150077e-06,
+      "loss": 0.8538,
+      "step": 2940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2678474737240484e-06,
+      "loss": 0.7905,
+      "step": 2941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2617985696812127e-06,
+      "loss": 0.8887,
+      "step": 2942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2557541780372993e-06,
+      "loss": 0.9462,
+      "step": 2943
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2497143028400724e-06,
+      "loss": 0.8296,
+      "step": 2944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.243678948134289e-06,
+      "loss": 0.9328,
+      "step": 2945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.237648117961665e-06,
+      "loss": 0.8465,
+      "step": 2946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.231621816360887e-06,
+      "loss": 0.8369,
+      "step": 2947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2256000473676186e-06,
+      "loss": 0.887,
+      "step": 2948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2195828150144835e-06,
+      "loss": 0.7899,
+      "step": 2949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2135701233310613e-06,
+      "loss": 0.9742,
+      "step": 2950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2075619763439014e-06,
+      "loss": 0.7847,
+      "step": 2951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.201558378076498e-06,
+      "loss": 0.8526,
+      "step": 2952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1955593325493083e-06,
+      "loss": 0.7951,
+      "step": 2953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.189564843779742e-06,
+      "loss": 0.8088,
+      "step": 2954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.183574915782148e-06,
+      "loss": 0.8829,
+      "step": 2955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1775895525678247e-06,
+      "loss": 0.9057,
+      "step": 2956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1716087581450193e-06,
+      "loss": 0.8847,
+      "step": 2957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.16563253651891e-06,
+      "loss": 0.7801,
+      "step": 2958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1596608916916195e-06,
+      "loss": 0.7718,
+      "step": 2959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1536938276622075e-06,
+      "loss": 0.9224,
+      "step": 2960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1477313484266557e-06,
+      "loss": 0.9738,
+      "step": 2961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1417734579778846e-06,
+      "loss": 0.7978,
+      "step": 2962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1358201603057424e-06,
+      "loss": 0.8307,
+      "step": 2963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.129871459396997e-06,
+      "loss": 0.7313,
+      "step": 2964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1239273592353368e-06,
+      "loss": 0.9821,
+      "step": 2965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1179878638013696e-06,
+      "loss": 0.8862,
+      "step": 2966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1120529770726236e-06,
+      "loss": 0.9041,
+      "step": 2967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1061227030235442e-06,
+      "loss": 0.9226,
+      "step": 2968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.100197045625474e-06,
+      "loss": 0.8795,
+      "step": 2969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0942760088466763e-06,
+      "loss": 0.8041,
+      "step": 2970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.088359596652318e-06,
+      "loss": 0.7976,
+      "step": 2971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0824478130044635e-06,
+      "loss": 0.9077,
+      "step": 2972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.076540661862084e-06,
+      "loss": 0.7781,
+      "step": 2973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0706381471810443e-06,
+      "loss": 0.9426,
+      "step": 2974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.064740272914102e-06,
+      "loss": 0.8903,
+      "step": 2975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0588470430109173e-06,
+      "loss": 0.8111,
+      "step": 2976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0529584614180275e-06,
+      "loss": 0.7949,
+      "step": 2977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.047074532078865e-06,
+      "loss": 0.9272,
+      "step": 2978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.041195258933749e-06,
+      "loss": 0.9729,
+      "step": 2979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0353206459198704e-06,
+      "loss": 0.8967,
+      "step": 2980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0294506969713078e-06,
+      "loss": 0.8826,
+      "step": 2981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.023585416019017e-06,
+      "loss": 0.8654,
+      "step": 2982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0177248069908226e-06,
+      "loss": 0.9038,
+      "step": 2983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0118688738114177e-06,
+      "loss": 0.8753,
+      "step": 2984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.006017620402377e-06,
+      "loss": 0.812,
+      "step": 2985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.000171050682127e-06,
+      "loss": 0.8632,
+      "step": 2986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9943291685659648e-06,
+      "loss": 0.9049,
+      "step": 2987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9884919779660536e-06,
+      "loss": 0.8831,
+      "step": 2988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9826594827914003e-06,
+      "loss": 0.8627,
+      "step": 2989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.976831686947884e-06,
+      "loss": 0.7705,
+      "step": 2990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9710085943382207e-06,
+      "loss": 0.8517,
+      "step": 2991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.965190208861991e-06,
+      "loss": 0.792,
+      "step": 2992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9593765344156143e-06,
+      "loss": 0.8238,
+      "step": 2993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.953567574892354e-06,
+      "loss": 0.8263,
+      "step": 2994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.947763334182324e-06,
+      "loss": 0.8382,
+      "step": 2995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9419638161724775e-06,
+      "loss": 0.8411,
+      "step": 2996
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9361690247465924e-06,
+      "loss": 0.814,
+      "step": 2997
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9303789637852965e-06,
+      "loss": 0.9672,
+      "step": 2998
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9245936371660456e-06,
+      "loss": 0.8311,
+      "step": 2999
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.918813048763116e-06,
+      "loss": 0.9493,
+      "step": 3000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.913037202447625e-06,
+      "loss": 0.8315,
+      "step": 3001
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.907266102087505e-06,
+      "loss": 0.8647,
+      "step": 3002
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9014997515475074e-06,
+      "loss": 0.8798,
+      "step": 3003
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.895738154689216e-06,
+      "loss": 0.837,
+      "step": 3004
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8899813153710144e-06,
+      "loss": 0.7692,
+      "step": 3005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.884229237448114e-06,
+      "loss": 0.8474,
+      "step": 3006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8784819247725336e-06,
+      "loss": 0.8998,
+      "step": 3007
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8727393811930948e-06,
+      "loss": 0.8429,
+      "step": 3008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8670016105554333e-06,
+      "loss": 0.9406,
+      "step": 3009
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8612686167019876e-06,
+      "loss": 0.8839,
+      "step": 3010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8555404034719936e-06,
+      "loss": 0.8965,
+      "step": 3011
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8498169747014824e-06,
+      "loss": 0.767,
+      "step": 3012
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8440983342232932e-06,
+      "loss": 0.7343,
+      "step": 3013
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.838384485867044e-06,
+      "loss": 0.976,
+      "step": 3014
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.832675433459159e-06,
+      "loss": 0.8265,
+      "step": 3015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826971180822835e-06,
+      "loss": 0.9551,
+      "step": 3016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.821271731778066e-06,
+      "loss": 0.8275,
+      "step": 3017
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.815577090141629e-06,
+      "loss": 0.8919,
+      "step": 3018
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809887259727071e-06,
+      "loss": 0.9181,
+      "step": 3019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804202244344729e-06,
+      "loss": 0.7785,
+      "step": 3020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7985220478017103e-06,
+      "loss": 0.7894,
+      "step": 3021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.79284667390189e-06,
+      "loss": 0.8895,
+      "step": 3022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.787176126445923e-06,
+      "loss": 0.8514,
+      "step": 3023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7815104092312306e-06,
+      "loss": 0.7165,
+      "step": 3024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77584952605199e-06,
+      "loss": 0.7582,
+      "step": 3025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7701934806991535e-06,
+      "loss": 0.9003,
+      "step": 3026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764542276960429e-06,
+      "loss": 0.9121,
+      "step": 3027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7588959186202756e-06,
+      "loss": 0.8358,
+      "step": 3028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.75325440945992e-06,
+      "loss": 0.8686,
+      "step": 3029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7476177532573312e-06,
+      "loss": 0.7774,
+      "step": 3030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7419859537872285e-06,
+      "loss": 0.8248,
+      "step": 3031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7363590148210883e-06,
+      "loss": 0.7832,
+      "step": 3032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.730736940127119e-06,
+      "loss": 0.8507,
+      "step": 3033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.725119733470284e-06,
+      "loss": 0.8104,
+      "step": 3034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719507398612281e-06,
+      "loss": 0.8496,
+      "step": 3035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7138999393115407e-06,
+      "loss": 0.8046,
+      "step": 3036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7082973593232365e-06,
+      "loss": 0.8215,
+      "step": 3037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.702699662399274e-06,
+      "loss": 0.7988,
+      "step": 3038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6971068522882816e-06,
+      "loss": 0.8347,
+      "step": 3039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6915189327356185e-06,
+      "loss": 0.8406,
+      "step": 3040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6859359074833736e-06,
+      "loss": 0.9274,
+      "step": 3041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6803577802703496e-06,
+      "loss": 0.8239,
+      "step": 3042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6747845548320785e-06,
+      "loss": 0.7962,
+      "step": 3043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6692162349008e-06,
+      "loss": 0.9152,
+      "step": 3044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.663652824205476e-06,
+      "loss": 0.8899,
+      "step": 3045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.658094326471783e-06,
+      "loss": 0.8456,
+      "step": 3046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6525407454220955e-06,
+      "loss": 0.8615,
+      "step": 3047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.646992084775508e-06,
+      "loss": 0.8333,
+      "step": 3048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6414483482478138e-06,
+      "loss": 0.9741,
+      "step": 3049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6359095395515066e-06,
+      "loss": 0.9884,
+      "step": 3050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6303756623957845e-06,
+      "loss": 0.8896,
+      "step": 3051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6248467204865445e-06,
+      "loss": 0.8135,
+      "step": 3052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6193227175263715e-06,
+      "loss": 0.8509,
+      "step": 3053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613803657214552e-06,
+      "loss": 0.8746,
+      "step": 3054
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6082895432470524e-06,
+      "loss": 0.8248,
+      "step": 3055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6027803793165353e-06,
+      "loss": 0.9437,
+      "step": 3056
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597276169112346e-06,
+      "loss": 0.8889,
+      "step": 3057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5917769163205097e-06,
+      "loss": 0.8989,
+      "step": 3058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5862826246237315e-06,
+      "loss": 0.9065,
+      "step": 3059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5807932977014005e-06,
+      "loss": 0.8023,
+      "step": 3060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5753089392295727e-06,
+      "loss": 0.8404,
+      "step": 3061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569829552880981e-06,
+      "loss": 0.8719,
+      "step": 3062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5643551423250346e-06,
+      "loss": 0.847,
+      "step": 3063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5588857112277965e-06,
+      "loss": 0.9454,
+      "step": 3064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.553421263252006e-06,
+      "loss": 0.9309,
+      "step": 3065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547961802057066e-06,
+      "loss": 0.8189,
+      "step": 3066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5425073312990334e-06,
+      "loss": 0.987,
+      "step": 3067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.537057854630626e-06,
+      "loss": 0.9015,
+      "step": 3068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.531613375701213e-06,
+      "loss": 0.9291,
+      "step": 3069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526173898156824e-06,
+      "loss": 0.8655,
+      "step": 3070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5207394256401405e-06,
+      "loss": 0.7878,
+      "step": 3071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.515309961790482e-06,
+      "loss": 0.8485,
+      "step": 3072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.50988551024382e-06,
+      "loss": 0.8821,
+      "step": 3073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5044660746327754e-06,
+      "loss": 0.8874,
+      "step": 3074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4990516585865967e-06,
+      "loss": 0.8187,
+      "step": 3075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.493642265731183e-06,
+      "loss": 0.8857,
+      "step": 3076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.488237899689063e-06,
+      "loss": 0.9263,
+      "step": 3077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4828385640793974e-06,
+      "loss": 0.7469,
+      "step": 3078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.477444262517984e-06,
+      "loss": 0.8391,
+      "step": 3079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4720549986172495e-06,
+      "loss": 0.8822,
+      "step": 3080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4666707759862373e-06,
+      "loss": 0.8638,
+      "step": 3081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.461291598230631e-06,
+      "loss": 0.8277,
+      "step": 3082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4559174689527176e-06,
+      "loss": 0.8245,
+      "step": 3083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4505483917514173e-06,
+      "loss": 0.7896,
+      "step": 3084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4451843702222633e-06,
+      "loss": 0.8205,
+      "step": 3085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.439825407957399e-06,
+      "loss": 0.8803,
+      "step": 3086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4344715085455817e-06,
+      "loss": 0.9273,
+      "step": 3087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429122675572182e-06,
+      "loss": 0.8208,
+      "step": 3088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4237789126191715e-06,
+      "loss": 0.8323,
+      "step": 3089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.418440223265132e-06,
+      "loss": 0.881,
+      "step": 3090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4131066110852464e-06,
+      "loss": 0.8601,
+      "step": 3091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.407778079651293e-06,
+      "loss": 0.9298,
+      "step": 3092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.402454632531657e-06,
+      "loss": 0.7989,
+      "step": 3093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.397136273291306e-06,
+      "loss": 0.9471,
+      "step": 3094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3918230054918136e-06,
+      "loss": 0.8607,
+      "step": 3095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3865148326913355e-06,
+      "loss": 0.8826,
+      "step": 3096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3812117584446138e-06,
+      "loss": 0.9062,
+      "step": 3097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3759137863029822e-06,
+      "loss": 0.8399,
+      "step": 3098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3706209198143603e-06,
+      "loss": 0.8302,
+      "step": 3099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3653331625232367e-06,
+      "loss": 0.801,
+      "step": 3100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360050517970688e-06,
+      "loss": 0.8134,
+      "step": 3101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.354772989694367e-06,
+      "loss": 0.8917,
+      "step": 3102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3495005812284922e-06,
+      "loss": 0.8601,
+      "step": 3103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.344233296103866e-06,
+      "loss": 0.8733,
+      "step": 3104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3389711378478484e-06,
+      "loss": 0.812,
+      "step": 3105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3337141099843684e-06,
+      "loss": 0.9172,
+      "step": 3106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3284622160339266e-06,
+      "loss": 0.8771,
+      "step": 3107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3232154595135738e-06,
+      "loss": 0.8524,
+      "step": 3108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3179738439369313e-06,
+      "loss": 0.8324,
+      "step": 3109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3127373728141756e-06,
+      "loss": 0.9305,
+      "step": 3110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.307506049652031e-06,
+      "loss": 0.7627,
+      "step": 3111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.302279877953779e-06,
+      "loss": 0.7998,
+      "step": 3112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2970588612192568e-06,
+      "loss": 0.812,
+      "step": 3113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2918430029448413e-06,
+      "loss": 0.8313,
+      "step": 3114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.286632306623453e-06,
+      "loss": 0.8506,
+      "step": 3115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2814267757445662e-06,
+      "loss": 0.8635,
+      "step": 3116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.276226413794187e-06,
+      "loss": 0.9909,
+      "step": 3117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2710312242548626e-06,
+      "loss": 0.8214,
+      "step": 3118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2658412106056817e-06,
+      "loss": 0.8524,
+      "step": 3119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2606563763222545e-06,
+      "loss": 0.8502,
+      "step": 3120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2554767248767385e-06,
+      "loss": 0.8567,
+      "step": 3121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.250302259737803e-06,
+      "loss": 0.823,
+      "step": 3122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2451329843706626e-06,
+      "loss": 0.8527,
+      "step": 3123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239968902237042e-06,
+      "loss": 0.8676,
+      "step": 3124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2348100167951904e-06,
+      "loss": 0.8553,
+      "step": 3125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229656331499883e-06,
+      "loss": 0.8736,
+      "step": 3126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.224507849802413e-06,
+      "loss": 0.903,
+      "step": 3127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2193645751505777e-06,
+      "loss": 0.862,
+      "step": 3128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2142265109886994e-06,
+      "loss": 0.8489,
+      "step": 3129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2090936607576087e-06,
+      "loss": 0.7741,
+      "step": 3130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2039660278946363e-06,
+      "loss": 0.7941,
+      "step": 3131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1988436158336303e-06,
+      "loss": 0.8333,
+      "step": 3132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1937264280049365e-06,
+      "loss": 0.8839,
+      "step": 3133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1886144678353983e-06,
+      "loss": 0.8794,
+      "step": 3134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1835077387483704e-06,
+      "loss": 0.8218,
+      "step": 3135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17840624416369e-06,
+      "loss": 0.7311,
+      "step": 3136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1733099874976983e-06,
+      "loss": 0.7861,
+      "step": 3137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1682189721632297e-06,
+      "loss": 0.8379,
+      "step": 3138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1631332015696006e-06,
+      "loss": 0.9164,
+      "step": 3139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1580526791226242e-06,
+      "loss": 0.8157,
+      "step": 3140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1529774082245893e-06,
+      "loss": 0.8544,
+      "step": 3141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1479073922742798e-06,
+      "loss": 0.8462,
+      "step": 3142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1428426346669473e-06,
+      "loss": 0.888,
+      "step": 3143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.137783138794335e-06,
+      "loss": 0.8752,
+      "step": 3144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1327289080446503e-06,
+      "loss": 0.8307,
+      "step": 3145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1276799458025853e-06,
+      "loss": 0.8838,
+      "step": 3146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1226362554492965e-06,
+      "loss": 0.8787,
+      "step": 3147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.117597840362412e-06,
+      "loss": 0.9015,
+      "step": 3148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1125647039160334e-06,
+      "loss": 0.9322,
+      "step": 3149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.107536849480715e-06,
+      "loss": 0.8289,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102514280423484e-06,
+      "loss": 0.8702,
+      "step": 3151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0974970001078264e-06,
+      "loss": 0.7751,
+      "step": 3152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092485011893679e-06,
+      "loss": 0.7985,
+      "step": 3153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087478319137445e-06,
+      "loss": 0.8176,
+      "step": 3154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0824769251919775e-06,
+      "loss": 0.8981,
+      "step": 3155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0774808334065765e-06,
+      "loss": 0.9085,
+      "step": 3156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0724900471269983e-06,
+      "loss": 0.7883,
+      "step": 3157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067504569695444e-06,
+      "loss": 0.8332,
+      "step": 3158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0625244044505577e-06,
+      "loss": 0.8555,
+      "step": 3159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0575495547274236e-06,
+      "loss": 0.8018,
+      "step": 3160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0525800238575757e-06,
+      "loss": 0.9244,
+      "step": 3161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047615815168974e-06,
+      "loss": 0.8733,
+      "step": 3162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0426569319860256e-06,
+      "loss": 0.8526,
+      "step": 3163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0377033776295617e-06,
+      "loss": 0.8579,
+      "step": 3164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.032755155416851e-06,
+      "loss": 0.9173,
+      "step": 3165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027812268661592e-06,
+      "loss": 0.8383,
+      "step": 3166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.022874720673904e-06,
+      "loss": 0.7755,
+      "step": 3167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0179425147603404e-06,
+      "loss": 0.8644,
+      "step": 3168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0130156542238656e-06,
+      "loss": 0.8886,
+      "step": 3169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0080941423638744e-06,
+      "loss": 0.8532,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0031779824761754e-06,
+      "loss": 0.9058,
+      "step": 3171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9982671778529904e-06,
+      "loss": 0.8088,
+      "step": 3172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99336173178296e-06,
+      "loss": 0.8518,
+      "step": 3173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9884616475511364e-06,
+      "loss": 0.8362,
+      "step": 3174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983566928438976e-06,
+      "loss": 0.9043,
+      "step": 3175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978677577724345e-06,
+      "loss": 0.898,
+      "step": 3176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9737935986815205e-06,
+      "loss": 0.8185,
+      "step": 3177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.968914994581169e-06,
+      "loss": 0.8534,
+      "step": 3178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9640417686903734e-06,
+      "loss": 0.8466,
+      "step": 3179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9591739242726003e-06,
+      "loss": 0.9677,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9543114645877203e-06,
+      "loss": 0.9075,
+      "step": 3181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9494543928919963e-06,
+      "loss": 0.8193,
+      "step": 3182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9446027124380872e-06,
+      "loss": 0.8412,
+      "step": 3183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9397564264750336e-06,
+      "loss": 0.8932,
+      "step": 3184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.934915538248271e-06,
+      "loss": 0.8874,
+      "step": 3185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.930080050999612e-06,
+      "loss": 0.9271,
+      "step": 3186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9252499679672632e-06,
+      "loss": 0.8624,
+      "step": 3187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9204252923858003e-06,
+      "loss": 0.8465,
+      "step": 3188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9156060274861875e-06,
+      "loss": 0.8722,
+      "step": 3189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9107921764957595e-06,
+      "loss": 0.8038,
+      "step": 3190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.90598374263823e-06,
+      "loss": 0.7869,
+      "step": 3191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.901180729133678e-06,
+      "loss": 0.8831,
+      "step": 3192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.896383139198561e-06,
+      "loss": 0.8733,
+      "step": 3193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8915909760457018e-06,
+      "loss": 0.8919,
+      "step": 3194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8868042428842836e-06,
+      "loss": 0.8383,
+      "step": 3195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8820229429198633e-06,
+      "loss": 0.8084,
+      "step": 3196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8772470793543463e-06,
+      "loss": 0.9102,
+      "step": 3197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.872476655386012e-06,
+      "loss": 0.8769,
+      "step": 3198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8677116742094858e-06,
+      "loss": 0.8462,
+      "step": 3199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8629521390157511e-06,
+      "loss": 0.8996,
+      "step": 3200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8581980529921462e-06,
+      "loss": 0.8142,
+      "step": 3201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8534494193223629e-06,
+      "loss": 0.8125,
+      "step": 3202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.848706241186432e-06,
+      "loss": 0.8664,
+      "step": 3203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8439685217607405e-06,
+      "loss": 0.8528,
+      "step": 3204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8392362642180194e-06,
+      "loss": 0.8375,
+      "step": 3205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8345094717273359e-06,
+      "loss": 0.7945,
+      "step": 3206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8297881474540968e-06,
+      "loss": 0.8705,
+      "step": 3207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8250722945600563e-06,
+      "loss": 0.8488,
+      "step": 3208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8203619162032948e-06,
+      "loss": 0.9159,
+      "step": 3209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8156570155382357e-06,
+      "loss": 0.8277,
+      "step": 3210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8109575957156223e-06,
+      "loss": 0.8525,
+      "step": 3211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8062636598825388e-06,
+      "loss": 0.7468,
+      "step": 3212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8015752111823937e-06,
+      "loss": 0.9591,
+      "step": 3213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7968922527549149e-06,
+      "loss": 0.8603,
+      "step": 3214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7922147877361639e-06,
+      "loss": 0.9284,
+      "step": 3215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787542819258512e-06,
+      "loss": 0.8019,
+      "step": 3216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7828763504506597e-06,
+      "loss": 0.8591,
+      "step": 3217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7782153844376182e-06,
+      "loss": 0.7571,
+      "step": 3218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7735599243407175e-06,
+      "loss": 0.7592,
+      "step": 3219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.768909973277595e-06,
+      "loss": 0.8953,
+      "step": 3220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.764265534362205e-06,
+      "loss": 0.8788,
+      "step": 3221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.759626610704811e-06,
+      "loss": 0.9838,
+      "step": 3222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7549932054119735e-06,
+      "loss": 0.8215,
+      "step": 3223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7503653215865702e-06,
+      "loss": 0.8903,
+      "step": 3224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7457429623277711e-06,
+      "loss": 0.7726,
+      "step": 3225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.74112613073105e-06,
+      "loss": 0.8296,
+      "step": 3226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7365148298881828e-06,
+      "loss": 0.848,
+      "step": 3227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7319090628872337e-06,
+      "loss": 0.872,
+      "step": 3228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7273088328125676e-06,
+      "loss": 0.7862,
+      "step": 3229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7227141427448423e-06,
+      "loss": 0.7302,
+      "step": 3230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.718124995760999e-06,
+      "loss": 0.9094,
+      "step": 3231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7135413949342706e-06,
+      "loss": 0.7555,
+      "step": 3232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708963343334179e-06,
+      "loss": 0.8604,
+      "step": 3233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043908440265257e-06,
+      "loss": 0.6638,
+      "step": 3234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6998239000733906e-06,
+      "loss": 0.9481,
+      "step": 3235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6952625145331446e-06,
+      "loss": 0.8521,
+      "step": 3236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.690706690460423e-06,
+      "loss": 0.9148,
+      "step": 3237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6861564309061484e-06,
+      "loss": 0.8359,
+      "step": 3238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.681611738917508e-06,
+      "loss": 0.8367,
+      "step": 3239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6770726175379647e-06,
+      "loss": 0.825,
+      "step": 3240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6725390698072542e-06,
+      "loss": 0.7729,
+      "step": 3241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6680110987613708e-06,
+      "loss": 0.9081,
+      "step": 3242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6634887074325844e-06,
+      "loss": 0.7814,
+      "step": 3243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6589718988494174e-06,
+      "loss": 0.9055,
+      "step": 3244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6544606760366667e-06,
+      "loss": 0.8367,
+      "step": 3245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6499550420153743e-06,
+      "loss": 0.8408,
+      "step": 3246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6454549998028524e-06,
+      "loss": 0.8616,
+      "step": 3247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.640960552412657e-06,
+      "loss": 0.8353,
+      "step": 3248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6364717028546096e-06,
+      "loss": 0.8581,
+      "step": 3249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6319884541347698e-06,
+      "loss": 0.7917,
+      "step": 3250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6275108092554558e-06,
+      "loss": 0.9058,
+      "step": 3251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.623038771215234e-06,
+      "loss": 0.8522,
+      "step": 3252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6185723430089085e-06,
+      "loss": 0.8722,
+      "step": 3253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6141115276275298e-06,
+      "loss": 0.7961,
+      "step": 3254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6096563280583932e-06,
+      "loss": 0.7802,
+      "step": 3255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6052067472850276e-06,
+      "loss": 0.8185,
+      "step": 3256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6007627882872034e-06,
+      "loss": 0.8558,
+      "step": 3257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5963244540409272e-06,
+      "loss": 0.8408,
+      "step": 3258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5918917475184314e-06,
+      "loss": 0.9022,
+      "step": 3259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.8872,
+      "step": 3260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5830432295148956e-06,
+      "loss": 0.8263,
+      "step": 3261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.578627423959479e-06,
+      "loss": 0.876,
+      "step": 3262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5742172579790849e-06,
+      "loss": 0.8238,
+      "step": 3263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5698127345270942e-06,
+      "loss": 0.8505,
+      "step": 3264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.565413856553095e-06,
+      "loss": 0.8446,
+      "step": 3265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5610206270029072e-06,
+      "loss": 0.851,
+      "step": 3266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5566330488185587e-06,
+      "loss": 0.8289,
+      "step": 3267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5522511249382998e-06,
+      "loss": 0.8905,
+      "step": 3268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5478748582965919e-06,
+      "loss": 0.86,
+      "step": 3269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.543504251824104e-06,
+      "loss": 0.8315,
+      "step": 3270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5391393084477224e-06,
+      "loss": 0.826,
+      "step": 3271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5347800310905347e-06,
+      "loss": 0.8307,
+      "step": 3272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5304264226718334e-06,
+      "loss": 0.8281,
+      "step": 3273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5260784861071187e-06,
+      "loss": 0.8661,
+      "step": 3274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5217362243080946e-06,
+      "loss": 0.8216,
+      "step": 3275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5173996401826563e-06,
+      "loss": 0.7865,
+      "step": 3276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.513068736634906e-06,
+      "loss": 0.8376,
+      "step": 3277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5087435165651344e-06,
+      "loss": 0.7923,
+      "step": 3278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5044239828698293e-06,
+      "loss": 0.7459,
+      "step": 3279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.500110138441675e-06,
+      "loss": 0.8046,
+      "step": 3280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4958019861695382e-06,
+      "loss": 0.8484,
+      "step": 3281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4914995289384738e-06,
+      "loss": 0.8979,
+      "step": 3282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.487202769629732e-06,
+      "loss": 0.9437,
+      "step": 3283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4829117111207346e-06,
+      "loss": 0.8727,
+      "step": 3284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4786263562850967e-06,
+      "loss": 0.8059,
+      "step": 3285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.474346707992611e-06,
+      "loss": 0.9741,
+      "step": 3286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.470072769109242e-06,
+      "loss": 0.8043,
+      "step": 3287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4658045424971412e-06,
+      "loss": 0.8453,
+      "step": 3288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4615420310146234e-06,
+      "loss": 0.9792,
+      "step": 3289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4572852375161884e-06,
+      "loss": 0.8822,
+      "step": 3290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.453034164852497e-06,
+      "loss": 0.8847,
+      "step": 3291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4487888158703812e-06,
+      "loss": 0.768,
+      "step": 3292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4445491934128419e-06,
+      "loss": 0.9011,
+      "step": 3293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4403153003190485e-06,
+      "loss": 0.9515,
+      "step": 3294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4360871394243247e-06,
+      "loss": 0.8369,
+      "step": 3295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4318647135601605e-06,
+      "loss": 0.8101,
+      "step": 3296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4276480255542102e-06,
+      "loss": 0.7557,
+      "step": 3297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4234370782302742e-06,
+      "loss": 0.9255,
+      "step": 3298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.419231874408319e-06,
+      "loss": 0.8454,
+      "step": 3299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4150324169044604e-06,
+      "loss": 0.977,
+      "step": 3300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4108387085309627e-06,
+      "loss": 0.8981,
+      "step": 3301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.406650752096248e-06,
+      "loss": 0.8769,
+      "step": 3302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4024685504048775e-06,
+      "loss": 0.7905,
+      "step": 3303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3982921062575672e-06,
+      "loss": 0.8373,
+      "step": 3304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3941214224511746e-06,
+      "loss": 0.7491,
+      "step": 3305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3899565017786942e-06,
+      "loss": 0.8617,
+      "step": 3306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3857973470292675e-06,
+      "loss": 0.8722,
+      "step": 3307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.381643960988175e-06,
+      "loss": 0.7956,
+      "step": 3308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3774963464368295e-06,
+      "loss": 0.755,
+      "step": 3309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3733545061527797e-06,
+      "loss": 0.9309,
+      "step": 3310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3692184429097123e-06,
+      "loss": 0.8889,
+      "step": 3311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3650881594774367e-06,
+      "loss": 0.8442,
+      "step": 3312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3609636586219e-06,
+      "loss": 0.8875,
+      "step": 3313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3568449431051756e-06,
+      "loss": 0.8337,
+      "step": 3314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3527320156854563e-06,
+      "loss": 0.7874,
+      "step": 3315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3486248791170663e-06,
+      "loss": 0.8754,
+      "step": 3316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.344523536150445e-06,
+      "loss": 0.8276,
+      "step": 3317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3404279895321591e-06,
+      "loss": 0.8887,
+      "step": 3318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3363382420048898e-06,
+      "loss": 0.8653,
+      "step": 3319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3322542963074314e-06,
+      "loss": 0.6865,
+      "step": 3320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3281761551746985e-06,
+      "loss": 0.8375,
+      "step": 3321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.32410382133772e-06,
+      "loss": 0.8722,
+      "step": 3322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3200372975236264e-06,
+      "loss": 0.8578,
+      "step": 3323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3159765864556662e-06,
+      "loss": 0.762,
+      "step": 3324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3119216908531951e-06,
+      "loss": 0.8333,
+      "step": 3325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3078726134316667e-06,
+      "loss": 0.9313,
+      "step": 3326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3038293569026484e-06,
+      "loss": 0.7431,
+      "step": 3327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.299791923973801e-06,
+      "loss": 0.8967,
+      "step": 3328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2957603173488864e-06,
+      "loss": 0.8516,
+      "step": 3329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2917345397277736e-06,
+      "loss": 0.7504,
+      "step": 3330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.287714593806415e-06,
+      "loss": 0.805,
+      "step": 3331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2837004822768673e-06,
+      "loss": 0.8116,
+      "step": 3332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2796922078272789e-06,
+      "loss": 0.7499,
+      "step": 3333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2756897731418827e-06,
+      "loss": 0.7582,
+      "step": 3334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.271693180901008e-06,
+      "loss": 0.748,
+      "step": 3335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2677024337810717e-06,
+      "loss": 0.8019,
+      "step": 3336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2637175344545716e-06,
+      "loss": 0.9353,
+      "step": 3337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2597384855900884e-06,
+      "loss": 0.8221,
+      "step": 3338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2557652898522931e-06,
+      "loss": 0.8924,
+      "step": 3339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2517979499019284e-06,
+      "loss": 0.888,
+      "step": 3340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2478364683958211e-06,
+      "loss": 0.7782,
+      "step": 3341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2438808479868715e-06,
+      "loss": 0.9307,
+      "step": 3342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2399310913240547e-06,
+      "loss": 0.8308,
+      "step": 3343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2359872010524254e-06,
+      "loss": 0.9896,
+      "step": 3344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2320491798130995e-06,
+      "loss": 0.8495,
+      "step": 3345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.228117030243271e-06,
+      "loss": 0.8665,
+      "step": 3346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.224190754976199e-06,
+      "loss": 0.859,
+      "step": 3347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2202703566412033e-06,
+      "loss": 0.8251,
+      "step": 3348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2163558378636775e-06,
+      "loss": 0.8831,
+      "step": 3349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2124472012650735e-06,
+      "loss": 0.7832,
+      "step": 3350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2085444494629017e-06,
+      "loss": 0.7942,
+      "step": 3351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.204647585070735e-06,
+      "loss": 0.8694,
+      "step": 3352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.200756610698205e-06,
+      "loss": 0.8504,
+      "step": 3353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1968715289509914e-06,
+      "loss": 0.8533,
+      "step": 3354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1929923424308376e-06,
+      "loss": 0.8581,
+      "step": 3355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1891190537355324e-06,
+      "loss": 0.8525,
+      "step": 3356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1852516654589142e-06,
+      "loss": 0.9037,
+      "step": 3357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1813901801908778e-06,
+      "loss": 0.8384,
+      "step": 3358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.177534600517355e-06,
+      "loss": 0.7082,
+      "step": 3359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.17368492902033e-06,
+      "loss": 0.7915,
+      "step": 3360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1698411682778288e-06,
+      "loss": 0.7439,
+      "step": 3361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1660033208639165e-06,
+      "loss": 0.871,
+      "step": 3362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1621713893487007e-06,
+      "loss": 0.7565,
+      "step": 3363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1583453762983289e-06,
+      "loss": 0.8836,
+      "step": 3364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.154525284274981e-06,
+      "loss": 0.91,
+      "step": 3365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1507111158368745e-06,
+      "loss": 0.9319,
+      "step": 3366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.146902873538256e-06,
+      "loss": 0.886,
+      "step": 3367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1431005599294077e-06,
+      "loss": 0.7717,
+      "step": 3368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.139304177556645e-06,
+      "loss": 0.8282,
+      "step": 3369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1355137289623019e-06,
+      "loss": 0.8325,
+      "step": 3370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1317292166847448e-06,
+      "loss": 0.9592,
+      "step": 3371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1279506432583665e-06,
+      "loss": 0.8912,
+      "step": 3372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1241780112135746e-06,
+      "loss": 0.8517,
+      "step": 3373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1204113230768087e-06,
+      "loss": 0.8478,
+      "step": 3374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1166505813705187e-06,
+      "loss": 0.8115,
+      "step": 3375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1128957886131753e-06,
+      "loss": 0.8532,
+      "step": 3376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1091469473192662e-06,
+      "loss": 0.7986,
+      "step": 3377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1054040599992965e-06,
+      "loss": 0.8704,
+      "step": 3378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1016671291597759e-06,
+      "loss": 0.8623,
+      "step": 3379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0979361573032366e-06,
+      "loss": 0.8418,
+      "step": 3380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094211146928207e-06,
+      "loss": 0.8328,
+      "step": 3381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0904921005292324e-06,
+      "loss": 0.8052,
+      "step": 3382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0867790205968654e-06,
+      "loss": 0.8283,
+      "step": 3383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0830719096176557e-06,
+      "loss": 0.9455,
+      "step": 3384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0793707700741595e-06,
+      "loss": 0.908,
+      "step": 3385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0756756044449358e-06,
+      "loss": 0.9183,
+      "step": 3386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.071986415204539e-06,
+      "loss": 0.8201,
+      "step": 3387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0683032048235242e-06,
+      "loss": 0.899,
+      "step": 3388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0646259757684453e-06,
+      "loss": 0.7991,
+      "step": 3389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.060954730501842e-06,
+      "loss": 0.9077,
+      "step": 3390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0572894714822557e-06,
+      "loss": 0.8436,
+      "step": 3391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0536302011642153e-06,
+      "loss": 0.7917,
+      "step": 3392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0499769219982393e-06,
+      "loss": 0.8622,
+      "step": 3393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.046329636430834e-06,
+      "loss": 0.8392,
+      "step": 3394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0426883469044879e-06,
+      "loss": 0.7681,
+      "step": 3395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0390530558576827e-06,
+      "loss": 0.8025,
+      "step": 3396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.035423765724879e-06,
+      "loss": 0.897,
+      "step": 3397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0318004789365155e-06,
+      "loss": 0.9693,
+      "step": 3398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.028183197919016e-06,
+      "loss": 0.8966,
+      "step": 3399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0245719250947794e-06,
+      "loss": 0.8312,
+      "step": 3400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0209666628821802e-06,
+      "loss": 0.9018,
+      "step": 3401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0173674136955702e-06,
+      "loss": 0.9376,
+      "step": 3402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.013774179945275e-06,
+      "loss": 0.8913,
+      "step": 3403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.010186964037585e-06,
+      "loss": 0.8434,
+      "step": 3404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006605768374772e-06,
+      "loss": 0.8852,
+      "step": 3405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0030305953550634e-06,
+      "loss": 0.8832,
+      "step": 3406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.99461447372665e-07,
+      "loss": 0.7983,
+      "step": 3407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.958983268177425e-07,
+      "loss": 0.8599,
+      "step": 3408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.923412360764217e-07,
+      "loss": 0.8547,
+      "step": 3409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.887901775307972e-07,
+      "loss": 0.8021,
+      "step": 3410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.852451535589224e-07,
+      "loss": 0.8468,
+      "step": 3411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.81706166534806e-07,
+      "loss": 0.8799,
+      "step": 3412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.78173218828413e-07,
+      "loss": 0.8808,
+      "step": 3413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.746463128056715e-07,
+      "loss": 0.8932,
+      "step": 3414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.711254508284561e-07,
+      "loss": 0.9287,
+      "step": 3415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.676106352545978e-07,
+      "loss": 0.7751,
+      "step": 3416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.64101868437879e-07,
+      "loss": 0.8752,
+      "step": 3417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.605991527280278e-07,
+      "loss": 0.9693,
+      "step": 3418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.571024904707238e-07,
+      "loss": 0.8508,
+      "step": 3419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.536118840075892e-07,
+      "loss": 0.8249,
+      "step": 3420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.501273356761964e-07,
+      "loss": 0.8758,
+      "step": 3421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.466488478100533e-07,
+      "loss": 0.9705,
+      "step": 3422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.431764227386131e-07,
+      "loss": 0.8875,
+      "step": 3423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.397100627872702e-07,
+      "loss": 0.8079,
+      "step": 3424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.362497702773588e-07,
+      "loss": 0.7671,
+      "step": 3425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.327955475261408e-07,
+      "loss": 0.8116,
+      "step": 3426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.293473968468247e-07,
+      "loss": 0.8937,
+      "step": 3427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.259053205485468e-07,
+      "loss": 0.8605,
+      "step": 3428
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.224693209363755e-07,
+      "loss": 0.8493,
+      "step": 3429
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.190394003113123e-07,
+      "loss": 0.815,
+      "step": 3430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.156155609702844e-07,
+      "loss": 0.836,
+      "step": 3431
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.121978052061475e-07,
+      "loss": 0.8448,
+      "step": 3432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.087861353076877e-07,
+      "loss": 0.8553,
+      "step": 3433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.053805535596072e-07,
+      "loss": 0.883,
+      "step": 3434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.019810622425385e-07,
+      "loss": 0.8227,
+      "step": 3435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.985876636330348e-07,
+      "loss": 0.8632,
+      "step": 3436
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.95200360003563e-07,
+      "loss": 0.813,
+      "step": 3437
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.918191536225151e-07,
+      "loss": 0.8024,
+      "step": 3438
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.884440467541988e-07,
+      "loss": 0.8377,
+      "step": 3439
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.850750416588339e-07,
+      "loss": 0.8922,
+      "step": 3440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.817121405925543e-07,
+      "loss": 0.7592,
+      "step": 3441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.783553458074101e-07,
+      "loss": 0.8708,
+      "step": 3442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.750046595513584e-07,
+      "loss": 0.802,
+      "step": 3443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.716600840682687e-07,
+      "loss": 0.8692,
+      "step": 3444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.68321621597914e-07,
+      "loss": 0.8891,
+      "step": 3445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.649892743759769e-07,
+      "loss": 0.932,
+      "step": 3446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.616630446340469e-07,
+      "loss": 0.965,
+      "step": 3447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.583429345996098e-07,
+      "loss": 0.8099,
+      "step": 3448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.550289464960603e-07,
+      "loss": 0.8003,
+      "step": 3449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.517210825426891e-07,
+      "loss": 0.8041,
+      "step": 3450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.484193449546863e-07,
+      "loss": 0.8652,
+      "step": 3451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.451237359431397e-07,
+      "loss": 0.7976,
+      "step": 3452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.418342577150374e-07,
+      "loss": 0.8487,
+      "step": 3453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.385509124732516e-07,
+      "loss": 0.8779,
+      "step": 3454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.352737024165569e-07,
+      "loss": 0.7967,
+      "step": 3455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.320026297396178e-07,
+      "loss": 0.9456,
+      "step": 3456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.28737696632983e-07,
+      "loss": 0.8326,
+      "step": 3457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.254789052830969e-07,
+      "loss": 0.8145,
+      "step": 3458
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.222262578722862e-07,
+      "loss": 0.8508,
+      "step": 3459
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.189797565787627e-07,
+      "loss": 0.933,
+      "step": 3460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.157394035766275e-07,
+      "loss": 0.7928,
+      "step": 3461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.12505201035858e-07,
+      "loss": 0.748,
+      "step": 3462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.092771511223185e-07,
+      "loss": 0.8206,
+      "step": 3463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.060552559977496e-07,
+      "loss": 0.8611,
+      "step": 3464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.028395178197701e-07,
+      "loss": 0.7938,
+      "step": 3465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.99629938741876e-07,
+      "loss": 0.8738,
+      "step": 3466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.964265209134437e-07,
+      "loss": 0.9049,
+      "step": 3467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.932292664797147e-07,
+      "loss": 0.9415,
+      "step": 3468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.900381775818067e-07,
+      "loss": 0.9147,
+      "step": 3469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.868532563567133e-07,
+      "loss": 0.8233,
+      "step": 3470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.836745049372896e-07,
+      "loss": 0.7962,
+      "step": 3471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.805019254522672e-07,
+      "loss": 0.8149,
+      "step": 3472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.773355200262367e-07,
+      "loss": 0.7576,
+      "step": 3473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.741752907796584e-07,
+      "loss": 0.8589,
+      "step": 3474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.710212398288597e-07,
+      "loss": 0.8211,
+      "step": 3475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.678733692860219e-07,
+      "loss": 0.808,
+      "step": 3476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.647316812591954e-07,
+      "loss": 0.8925,
+      "step": 3477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.615961778522862e-07,
+      "loss": 0.9667,
+      "step": 3478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.584668611650592e-07,
+      "loss": 0.8584,
+      "step": 3479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.553437332931368e-07,
+      "loss": 0.8596,
+      "step": 3480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.522267963279983e-07,
+      "loss": 0.8543,
+      "step": 3481
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.49116052356973e-07,
+      "loss": 0.7847,
+      "step": 3482
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.460115034632476e-07,
+      "loss": 0.8552,
+      "step": 3483
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.429131517258592e-07,
+      "loss": 0.9413,
+      "step": 3484
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.398209992196914e-07,
+      "loss": 0.8197,
+      "step": 3485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3673504801548e-07,
+      "loss": 0.8324,
+      "step": 3486
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.336553001798064e-07,
+      "loss": 0.825,
+      "step": 3487
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.305817577750951e-07,
+      "loss": 0.9581,
+      "step": 3488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.27514422859622e-07,
+      "loss": 0.8139,
+      "step": 3489
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.244532974874985e-07,
+      "loss": 0.8735,
+      "step": 3490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.213983837086824e-07,
+      "loss": 0.7981,
+      "step": 3491
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.183496835689718e-07,
+      "loss": 0.8451,
+      "step": 3492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.153071991099991e-07,
+      "loss": 0.8665,
+      "step": 3493
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.12270932369239e-07,
+      "loss": 0.7873,
+      "step": 3494
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.092408853800014e-07,
+      "loss": 0.7889,
+      "step": 3495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.062170601714302e-07,
+      "loss": 0.8012,
+      "step": 3496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.031994587685032e-07,
+      "loss": 0.7871,
+      "step": 3497
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.00188083192026e-07,
+      "loss": 0.8013,
+      "step": 3498
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.971829354586412e-07,
+      "loss": 0.8815,
+      "step": 3499
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.941840175808212e-07,
+      "loss": 0.8055,
+      "step": 3500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.911913315668583e-07,
+      "loss": 0.9073,
+      "step": 3501
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.882048794208807e-07,
+      "loss": 0.8107,
+      "step": 3502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.852246631428384e-07,
+      "loss": 0.7805,
+      "step": 3503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.822506847285015e-07,
+      "loss": 0.8967,
+      "step": 3504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.792829461694705e-07,
+      "loss": 0.8239,
+      "step": 3505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.763214494531601e-07,
+      "loss": 0.8555,
+      "step": 3506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.73366196562808e-07,
+      "loss": 0.8303,
+      "step": 3507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.704171894774702e-07,
+      "loss": 0.9468,
+      "step": 3508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.674744301720238e-07,
+      "loss": 0.8583,
+      "step": 3509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.645379206171543e-07,
+      "loss": 0.769,
+      "step": 3510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.616076627793689e-07,
+      "loss": 0.8968,
+      "step": 3511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.586836586209833e-07,
+      "loss": 0.8791,
+      "step": 3512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.557659101001279e-07,
+      "loss": 0.8333,
+      "step": 3513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.528544191707453e-07,
+      "loss": 0.9205,
+      "step": 3514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.499491877825836e-07,
+      "loss": 0.8312,
+      "step": 3515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.470502178811999e-07,
+      "loss": 0.8665,
+      "step": 3516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.441575114079613e-07,
+      "loss": 0.8262,
+      "step": 3517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.412710703000368e-07,
+      "loss": 0.8177,
+      "step": 3518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.383908964904029e-07,
+      "loss": 0.8424,
+      "step": 3519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.355169919078374e-07,
+      "loss": 0.8231,
+      "step": 3520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.326493584769167e-07,
+      "loss": 0.8558,
+      "step": 3521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.297879981180244e-07,
+      "loss": 0.9242,
+      "step": 3522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.269329127473378e-07,
+      "loss": 0.9489,
+      "step": 3523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.240841042768342e-07,
+      "loss": 0.9005,
+      "step": 3524
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.212415746142863e-07,
+      "loss": 0.9009,
+      "step": 3525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.184053256632617e-07,
+      "loss": 0.8466,
+      "step": 3526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.155753593231228e-07,
+      "loss": 0.8324,
+      "step": 3527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.127516774890274e-07,
+      "loss": 0.8678,
+      "step": 3528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.099342820519183e-07,
+      "loss": 0.9243,
+      "step": 3529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.071231748985352e-07,
+      "loss": 0.8157,
+      "step": 3530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.043183579114032e-07,
+      "loss": 0.8693,
+      "step": 3531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.015198329688343e-07,
+      "loss": 0.8305,
+      "step": 3532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.987276019449295e-07,
+      "loss": 0.8174,
+      "step": 3533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.959416667095741e-07,
+      "loss": 0.8737,
+      "step": 3534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.931620291284335e-07,
+      "loss": 0.8388,
+      "step": 3535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.903886910629631e-07,
+      "loss": 0.8258,
+      "step": 3536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.876216543703916e-07,
+      "loss": 0.9404,
+      "step": 3537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.848609209037337e-07,
+      "loss": 0.9468,
+      "step": 3538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.821064925117836e-07,
+      "loss": 0.8023,
+      "step": 3539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.79358371039106e-07,
+      "loss": 0.8898,
+      "step": 3540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.766165583260485e-07,
+      "loss": 0.9175,
+      "step": 3541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.738810562087338e-07,
+      "loss": 0.7712,
+      "step": 3542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.711518665190551e-07,
+      "loss": 0.7554,
+      "step": 3543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.684289910846796e-07,
+      "loss": 0.8701,
+      "step": 3544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.657124317290486e-07,
+      "loss": 0.8749,
+      "step": 3545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.630021902713678e-07,
+      "loss": 0.778,
+      "step": 3546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.602982685266179e-07,
+      "loss": 0.9032,
+      "step": 3547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.576006683055469e-07,
+      "loss": 0.8255,
+      "step": 3548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.549093914146652e-07,
+      "loss": 0.8326,
+      "step": 3549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.522244396562537e-07,
+      "loss": 0.8538,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.495458148283505e-07,
+      "loss": 0.8853,
+      "step": 3551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.468735187247675e-07,
+      "loss": 0.8832,
+      "step": 3552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.442075531350677e-07,
+      "loss": 0.8329,
+      "step": 3553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.415479198445794e-07,
+      "loss": 0.868,
+      "step": 3554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.388946206343903e-07,
+      "loss": 0.771,
+      "step": 3555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.362476572813491e-07,
+      "loss": 0.83,
+      "step": 3556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.336070315580544e-07,
+      "loss": 0.8384,
+      "step": 3557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.309727452328673e-07,
+      "loss": 0.8123,
+      "step": 3558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.283448000699009e-07,
+      "loss": 0.9082,
+      "step": 3559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.257231978290212e-07,
+      "loss": 0.9144,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.23107940265849e-07,
+      "loss": 0.9375,
+      "step": 3561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.204990291317535e-07,
+      "loss": 0.8072,
+      "step": 3562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.178964661738539e-07,
+      "loss": 0.8726,
+      "step": 3563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.153002531350204e-07,
+      "loss": 0.8642,
+      "step": 3564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.127103917538679e-07,
+      "loss": 0.7672,
+      "step": 3565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.101268837647611e-07,
+      "loss": 0.8813,
+      "step": 3566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.075497308978095e-07,
+      "loss": 0.8466,
+      "step": 3567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.049789348788625e-07,
+      "loss": 0.7953,
+      "step": 3568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.024144974295164e-07,
+      "loss": 0.8167,
+      "step": 3569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.998564202671119e-07,
+      "loss": 0.8135,
+      "step": 3570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.973047051047242e-07,
+      "loss": 0.9286,
+      "step": 3571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.947593536511697e-07,
+      "loss": 0.8429,
+      "step": 3572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.92220367611006e-07,
+      "loss": 0.951,
+      "step": 3573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896877486845264e-07,
+      "loss": 0.8914,
+      "step": 3574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.871614985677598e-07,
+      "loss": 0.8553,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.846416189524695e-07,
+      "loss": 0.7866,
+      "step": 3576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.821281115261544e-07,
+      "loss": 0.8386,
+      "step": 3577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.796209779720462e-07,
+      "loss": 0.8082,
+      "step": 3578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.771202199691061e-07,
+      "loss": 0.7602,
+      "step": 3579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7462583919202777e-07,
+      "loss": 0.8253,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7213783731123287e-07,
+      "loss": 0.8745,
+      "step": 3581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.696562159928708e-07,
+      "loss": 0.9832,
+      "step": 3582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6718097689882004e-07,
+      "loss": 0.8092,
+      "step": 3583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.647121216866857e-07,
+      "loss": 0.7967,
+      "step": 3584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.622496520097941e-07,
+      "loss": 0.8986,
+      "step": 3585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5979356951719625e-07,
+      "loss": 0.7804,
+      "step": 3586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.573438758536708e-07,
+      "loss": 0.7783,
+      "step": 3587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549005726597111e-07,
+      "loss": 0.938,
+      "step": 3588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5246366157153496e-07,
+      "loss": 0.9719,
+      "step": 3589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5003314422107924e-07,
+      "loss": 0.8619,
+      "step": 3590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.476090222359963e-07,
+      "loss": 0.82,
+      "step": 3591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4519129723965995e-07,
+      "loss": 0.8208,
+      "step": 3592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.42779970851156e-07,
+      "loss": 0.8504,
+      "step": 3593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4037504468528835e-07,
+      "loss": 0.8819,
+      "step": 3594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3797652035257544e-07,
+      "loss": 0.8065,
+      "step": 3595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.355843994592446e-07,
+      "loss": 0.8544,
+      "step": 3596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3319868360723774e-07,
+      "loss": 0.8963,
+      "step": 3597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3081937439420796e-07,
+      "loss": 0.8302,
+      "step": 3598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.284464734135163e-07,
+      "loss": 0.8453,
+      "step": 3599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2607998225423165e-07,
+      "loss": 0.8271,
+      "step": 3600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.23719902501134e-07,
+      "loss": 0.7901,
+      "step": 3601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.213662357347059e-07,
+      "loss": 0.7982,
+      "step": 3602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.190189835311387e-07,
+      "loss": 0.7579,
+      "step": 3603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.166781474623238e-07,
+      "loss": 0.7757,
+      "step": 3604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1434372909586075e-07,
+      "loss": 0.8779,
+      "step": 3605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1201572999505e-07,
+      "loss": 0.874,
+      "step": 3606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.09694151718889e-07,
+      "loss": 0.9026,
+      "step": 3607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0737899582208194e-07,
+      "loss": 0.9573,
+      "step": 3608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.8306,
+      "step": 3609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0276795736382103e-07,
+      "loss": 0.7308,
+      "step": 3610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0047207789026043e-07,
+      "loss": 0.9595,
+      "step": 3611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.98182626971837e-07,
+      "loss": 0.8217,
+      "step": 3612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9589960614173417e-07,
+      "loss": 0.9427,
+      "step": 3613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9362301692883466e-07,
+      "loss": 0.9739,
+      "step": 3614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9135286085770886e-07,
+      "loss": 0.9322,
+      "step": 3615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.890891394486229e-07,
+      "loss": 0.8667,
+      "step": 3616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8683185421753313e-07,
+      "loss": 0.8289,
+      "step": 3617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8458100667608356e-07,
+      "loss": 0.8445,
+      "step": 3618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8233659833160854e-07,
+      "loss": 0.7644,
+      "step": 3619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8009863068713126e-07,
+      "loss": 0.8785,
+      "step": 3620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.778671052413585e-07,
+      "loss": 0.9158,
+      "step": 3621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7564202348868594e-07,
+      "loss": 0.9418,
+      "step": 3622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.73423386919195e-07,
+      "loss": 0.8068,
+      "step": 3623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7121119701864717e-07,
+      "loss": 0.8871,
+      "step": 3624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.690054552684896e-07,
+      "loss": 0.8584,
+      "step": 3625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.668061631458508e-07,
+      "loss": 0.8979,
+      "step": 3626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.646133221235393e-07,
+      "loss": 0.8067,
+      "step": 3627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6242693367004365e-07,
+      "loss": 0.8754,
+      "step": 3628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6024699924952943e-07,
+      "loss": 0.8755,
+      "step": 3629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.580735203218455e-07,
+      "loss": 0.7464,
+      "step": 3630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5590649834251425e-07,
+      "loss": 0.8844,
+      "step": 3631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5374593476273143e-07,
+      "loss": 0.8446,
+      "step": 3632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.515918310293731e-07,
+      "loss": 1.0081,
+      "step": 3633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.494441885849864e-07,
+      "loss": 0.7886,
+      "step": 3634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473030088677898e-07,
+      "loss": 0.7762,
+      "step": 3635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4516829331167844e-07,
+      "loss": 0.8311,
+      "step": 3636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4304004334621554e-07,
+      "loss": 0.7158,
+      "step": 3637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4091826039663324e-07,
+      "loss": 0.8594,
+      "step": 3638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.38802945883836e-07,
+      "loss": 0.8453,
+      "step": 3639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3669410122439515e-07,
+      "loss": 0.8274,
+      "step": 3640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3459172783054884e-07,
+      "loss": 0.8855,
+      "step": 3641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.324958271102019e-07,
+      "loss": 0.8147,
+      "step": 3642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.30406400466925e-07,
+      "loss": 0.8423,
+      "step": 3643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.283234492999532e-07,
+      "loss": 0.8522,
+      "step": 3644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2624697500418634e-07,
+      "loss": 0.7807,
+      "step": 3645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.241769789701832e-07,
+      "loss": 0.805,
+      "step": 3646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.22113462584166e-07,
+      "loss": 0.8643,
+      "step": 3647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2005642722802157e-07,
+      "loss": 0.792,
+      "step": 3648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.180058742792891e-07,
+      "loss": 0.8818,
+      "step": 3649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1596180511117235e-07,
+      "loss": 0.8151,
+      "step": 3650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.139242210925342e-07,
+      "loss": 0.856,
+      "step": 3651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.118931235878875e-07,
+      "loss": 0.7704,
+      "step": 3652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.098685139574098e-07,
+      "loss": 0.8844,
+      "step": 3653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.078503935569255e-07,
+      "loss": 0.8092,
+      "step": 3654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.058387637379212e-07,
+      "loss": 0.7821,
+      "step": 3655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0383362584753165e-07,
+      "loss": 0.818,
+      "step": 3656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.018349812285448e-07,
+      "loss": 0.8292,
+      "step": 3657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.998428312194024e-07,
+      "loss": 0.9358,
+      "step": 3658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.97857177154196e-07,
+      "loss": 0.893,
+      "step": 3659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9587802036266546e-07,
+      "loss": 0.8894,
+      "step": 3660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.939053621702015e-07,
+      "loss": 0.8778,
+      "step": 3661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9193920389784304e-07,
+      "loss": 0.7597,
+      "step": 3662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8997954686227456e-07,
+      "loss": 0.8573,
+      "step": 3663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.880263923758275e-07,
+      "loss": 0.8967,
+      "step": 3664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.860797417464789e-07,
+      "loss": 0.8055,
+      "step": 3665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8413959627784927e-07,
+      "loss": 0.9249,
+      "step": 3666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8220595726920595e-07,
+      "loss": 0.8531,
+      "step": 3667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.802788260154543e-07,
+      "loss": 0.8735,
+      "step": 3668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7835820380714527e-07,
+      "loss": 0.8759,
+      "step": 3669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.764440919304723e-07,
+      "loss": 0.7675,
+      "step": 3670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7453649166726217e-07,
+      "loss": 0.8059,
+      "step": 3671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7263540429498747e-07,
+      "loss": 0.8414,
+      "step": 3672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.707408310867587e-07,
+      "loss": 0.832,
+      "step": 3673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.688527733113211e-07,
+      "loss": 0.8334,
+      "step": 3674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6697123223305533e-07,
+      "loss": 0.888,
+      "step": 3675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6509620911198577e-07,
+      "loss": 0.7693,
+      "step": 3676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.632277052037624e-07,
+      "loss": 0.7728,
+      "step": 3677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6136572175967654e-07,
+      "loss": 0.9505,
+      "step": 3678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.595102600266497e-07,
+      "loss": 0.8367,
+      "step": 3679
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5766132124723565e-07,
+      "loss": 0.9015,
+      "step": 3680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5581890665962286e-07,
+      "loss": 0.8541,
+      "step": 3681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5398301749762544e-07,
+      "loss": 0.8814,
+      "step": 3682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5215365499069446e-07,
+      "loss": 0.858,
+      "step": 3683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.503308203639032e-07,
+      "loss": 0.7718,
+      "step": 3684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4851451483795864e-07,
+      "loss": 0.8247,
+      "step": 3685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4670473962919327e-07,
+      "loss": 0.8115,
+      "step": 3686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.449014959495666e-07,
+      "loss": 0.9184,
+      "step": 3687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4310478500666365e-07,
+      "loss": 0.7279,
+      "step": 3688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.413146080036954e-07,
+      "loss": 0.9069,
+      "step": 3689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3953096613949954e-07,
+      "loss": 0.8261,
+      "step": 3690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3775386060853057e-07,
+      "loss": 0.877,
+      "step": 3691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3598329260087316e-07,
+      "loss": 0.8377,
+      "step": 3692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.342192633022289e-07,
+      "loss": 0.8399,
+      "step": 3693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3246177389392388e-07,
+      "loss": 0.8527,
+      "step": 3694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3071082555290225e-07,
+      "loss": 0.8756,
+      "step": 3695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2896641945172937e-07,
+      "loss": 0.825,
+      "step": 3696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2722855675858747e-07,
+      "loss": 0.9193,
+      "step": 3697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2549723863728113e-07,
+      "loss": 0.9331,
+      "step": 3698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2377246624722626e-07,
+      "loss": 0.8755,
+      "step": 3699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2205424074346004e-07,
+      "loss": 0.761,
+      "step": 3700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2034256327663316e-07,
+      "loss": 0.9027,
+      "step": 3701
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1863743499301094e-07,
+      "loss": 0.8277,
+      "step": 3702
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1693885703447325e-07,
+      "loss": 0.8388,
+      "step": 3703
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.152468305385158e-07,
+      "loss": 0.8425,
+      "step": 3704
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1356135663824328e-07,
+      "loss": 0.8062,
+      "step": 3705
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1188243646237505e-07,
+      "loss": 0.8803,
+      "step": 3706
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1021007113523839e-07,
+      "loss": 0.869,
+      "step": 3707
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0854426177677523e-07,
+      "loss": 0.7981,
+      "step": 3708
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0688500950253432e-07,
+      "loss": 0.8729,
+      "step": 3709
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0523231542367238e-07,
+      "loss": 0.8041,
+      "step": 3710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0358618064695858e-07,
+      "loss": 0.8284,
+      "step": 3711
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0194660627476436e-07,
+      "loss": 0.9104,
+      "step": 3712
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.003135934050715e-07,
+      "loss": 0.77,
+      "step": 3713
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.986871431314652e-07,
+      "loss": 0.8496,
+      "step": 3714
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9706725654313862e-07,
+      "loss": 0.8943,
+      "step": 3715
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9545393472488738e-07,
+      "loss": 0.8668,
+      "step": 3716
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9384717875710945e-07,
+      "loss": 0.8979,
+      "step": 3717
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.922469897158108e-07,
+      "loss": 0.9274,
+      "step": 3718
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.906533686725942e-07,
+      "loss": 0.8757,
+      "step": 3719
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8906631669466825e-07,
+      "loss": 0.765,
+      "step": 3720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8748583484483828e-07,
+      "loss": 0.895,
+      "step": 3721
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8591192418151328e-07,
+      "loss": 0.941,
+      "step": 3722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8434458575870118e-07,
+      "loss": 0.8776,
+      "step": 3723
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8278382062600686e-07,
+      "loss": 0.827,
+      "step": 3724
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8122962982863425e-07,
+      "loss": 0.8258,
+      "step": 3725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7968201440738519e-07,
+      "loss": 0.8743,
+      "step": 3726
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7814097539865626e-07,
+      "loss": 0.9307,
+      "step": 3727
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7660651383444304e-07,
+      "loss": 0.9026,
+      "step": 3728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7507863074233577e-07,
+      "loss": 0.818,
+      "step": 3729
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.73557327145516e-07,
+      "loss": 0.8065,
+      "step": 3730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7204260406276095e-07,
+      "loss": 0.9048,
+      "step": 3731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7053446250844264e-07,
+      "loss": 0.8047,
+      "step": 3732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.690329034925231e-07,
+      "loss": 0.748,
+      "step": 3733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6753792802056025e-07,
+      "loss": 0.8253,
+      "step": 3734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6604953709369764e-07,
+      "loss": 0.8471,
+      "step": 3735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6456773170867247e-07,
+      "loss": 0.7641,
+      "step": 3736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6309251285781424e-07,
+      "loss": 0.898,
+      "step": 3737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6162388152903498e-07,
+      "loss": 0.803,
+      "step": 3738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6016183870584123e-07,
+      "loss": 0.8172,
+      "step": 3739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.587063853673254e-07,
+      "loss": 0.8775,
+      "step": 3740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.572575224881645e-07,
+      "loss": 0.8144,
+      "step": 3741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5581525103862683e-07,
+      "loss": 0.8488,
+      "step": 3742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5437957198456422e-07,
+      "loss": 0.7904,
+      "step": 3743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5295048628741204e-07,
+      "loss": 0.8286,
+      "step": 3744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5152799490419368e-07,
+      "loss": 0.9021,
+      "step": 3745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5011209878751266e-07,
+      "loss": 0.8901,
+      "step": 3746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.487027988855605e-07,
+      "loss": 0.8374,
+      "step": 3747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4730009614210784e-07,
+      "loss": 0.9129,
+      "step": 3748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4590399149650769e-07,
+      "loss": 0.9496,
+      "step": 3749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4451448588369444e-07,
+      "loss": 0.8219,
+      "step": 3750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4313158023418706e-07,
+      "loss": 0.9655,
+      "step": 3751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4175527547407807e-07,
+      "loss": 0.9249,
+      "step": 3752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4038557252504582e-07,
+      "loss": 0.8702,
+      "step": 3753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.390224723043443e-07,
+      "loss": 0.7214,
+      "step": 3754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3766597572480445e-07,
+      "loss": 0.9159,
+      "step": 3755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3631608369483852e-07,
+      "loss": 0.8469,
+      "step": 3756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.349727971184356e-07,
+      "loss": 0.7937,
+      "step": 3757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.336361168951583e-07,
+      "loss": 0.8,
+      "step": 3758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3230604392014623e-07,
+      "loss": 0.7856,
+      "step": 3759
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.309825790841146e-07,
+      "loss": 0.8226,
+      "step": 3760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2966572327335447e-07,
+      "loss": 0.939,
+      "step": 3761
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2835547736973152e-07,
+      "loss": 0.8463,
+      "step": 3762
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2705184225068057e-07,
+      "loss": 0.8566,
+      "step": 3763
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2575481878921324e-07,
+      "loss": 0.9511,
+      "step": 3764
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2446440785391368e-07,
+      "loss": 0.9093,
+      "step": 3765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2318061030893503e-07,
+      "loss": 0.7898,
+      "step": 3766
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2190342701400515e-07,
+      "loss": 0.7805,
+      "step": 3767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2063285882441987e-07,
+      "loss": 0.8535,
+      "step": 3768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1936890659104418e-07,
+      "loss": 0.85,
+      "step": 3769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1811157116031647e-07,
+      "loss": 0.8239,
+      "step": 3770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1686085337423991e-07,
+      "loss": 0.891,
+      "step": 3771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1561675407038786e-07,
+      "loss": 0.8784,
+      "step": 3772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1437927408190275e-07,
+      "loss": 0.8093,
+      "step": 3773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1314841423749279e-07,
+      "loss": 0.7786,
+      "step": 3774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1192417536143085e-07,
+      "loss": 0.9454,
+      "step": 3775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1070655827356113e-07,
+      "loss": 0.8794,
+      "step": 3776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0949556378928806e-07,
+      "loss": 0.9703,
+      "step": 3777
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.08291192719584e-07,
+      "loss": 0.8506,
+      "step": 3778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.070934458709838e-07,
+      "loss": 0.9013,
+      "step": 3779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.059023240455892e-07,
+      "loss": 0.7833,
+      "step": 3780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0471782804106323e-07,
+      "loss": 0.8649,
+      "step": 3781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0353995865063138e-07,
+      "loss": 0.881,
+      "step": 3782
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0236871666308268e-07,
+      "loss": 0.91,
+      "step": 3783
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.012041028627686e-07,
+      "loss": 0.9374,
+      "step": 3784
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0004611802959863e-07,
+      "loss": 0.8795,
+      "step": 3785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.889476293904687e-08,
+      "loss": 0.8046,
+      "step": 3786
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.775003836214658e-08,
+      "loss": 0.8528,
+      "step": 3787
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.661194506548788e-08,
+      "loss": 0.787,
+      "step": 3788
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.548048381122333e-08,
+      "loss": 0.8322,
+      "step": 3789
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.435565535706459e-08,
+      "loss": 0.9487,
+      "step": 3790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.323746045627913e-08,
+      "loss": 1.0192,
+      "step": 3791
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.212589985769239e-08,
+      "loss": 0.8756,
+      "step": 3792
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.10209743056889e-08,
+      "loss": 0.8722,
+      "step": 3793
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.992268454020681e-08,
+      "loss": 0.8138,
+      "step": 3794
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.883103129674553e-08,
+      "loss": 0.8144,
+      "step": 3795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.774601530635363e-08,
+      "loss": 0.7932,
+      "step": 3796
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.666763729563987e-08,
+      "loss": 0.8055,
+      "step": 3797
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.55958979867666e-08,
+      "loss": 0.8679,
+      "step": 3798
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.453079809744857e-08,
+      "loss": 0.7819,
+      "step": 3799
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.347233834095525e-08,
+      "loss": 0.8714,
+      "step": 3800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.242051942611184e-08,
+      "loss": 0.8045,
+      "step": 3801
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.13753420572927e-08,
+      "loss": 0.9338,
+      "step": 3802
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.033680693442569e-08,
+      "loss": 0.7962,
+      "step": 3803
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.930491475299229e-08,
+      "loss": 0.8362,
+      "step": 3804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.827966620402306e-08,
+      "loss": 0.8582,
+      "step": 3805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.726106197409988e-08,
+      "loss": 0.8526,
+      "step": 3806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.624910274535713e-08,
+      "loss": 0.8609,
+      "step": 3807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.524378919547493e-08,
+      "loss": 0.8146,
+      "step": 3808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.424512199768918e-08,
+      "loss": 0.8503,
+      "step": 3809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.325310182077938e-08,
+      "loss": 0.8019,
+      "step": 3810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.226772932907522e-08,
+      "loss": 0.9403,
+      "step": 3811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.128900518245774e-08,
+      "loss": 0.8477,
+      "step": 3812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.031693003635042e-08,
+      "loss": 0.8307,
+      "step": 3813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.935150454172813e-08,
+      "loss": 0.7896,
+      "step": 3814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.839272934511143e-08,
+      "loss": 0.839,
+      "step": 3815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.744060508856564e-08,
+      "loss": 0.9038,
+      "step": 3816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.649513240970518e-08,
+      "loss": 0.8631,
+      "step": 3817
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.55563119416891e-08,
+      "loss": 0.8181,
+      "step": 3818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.462414431322005e-08,
+      "loss": 0.8333,
+      "step": 3819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.369863014854649e-08,
+      "loss": 0.8496,
+      "step": 3820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.27797700674615e-08,
+      "loss": 0.8466,
+      "step": 3821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.186756468530286e-08,
+      "loss": 0.9206,
+      "step": 3822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.09620146129497e-08,
+      "loss": 0.847,
+      "step": 3823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.006312045682583e-08,
+      "loss": 0.8476,
+      "step": 3824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.917088281889638e-08,
+      "loss": 0.8979,
+      "step": 3825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.828530229667228e-08,
+      "loss": 0.8615,
+      "step": 3826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.740637948320138e-08,
+      "loss": 0.7577,
+      "step": 3827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.653411496707617e-08,
+      "loss": 0.8181,
+      "step": 3828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.56685093324294e-08,
+      "loss": 0.7833,
+      "step": 3829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4809563158936265e-08,
+      "loss": 0.8584,
+      "step": 3830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3957277021808865e-08,
+      "loss": 0.8934,
+      "step": 3831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3111651491802864e-08,
+      "loss": 0.819,
+      "step": 3832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2272687135210834e-08,
+      "loss": 0.8296,
+      "step": 3833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.144038451386446e-08,
+      "loss": 0.9078,
+      "step": 3834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0614744185137895e-08,
+      "loss": 0.8912,
+      "step": 3835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9795766701937755e-08,
+      "loss": 0.8387,
+      "step": 3836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.898345261271531e-08,
+      "loss": 0.8065,
+      "step": 3837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8177802461453206e-08,
+      "loss": 0.8216,
+      "step": 3838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.737881678767653e-08,
+      "loss": 0.9448,
+      "step": 3839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6586496126445056e-08,
+      "loss": 0.8519,
+      "step": 3840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.580084100835547e-08,
+      "loss": 0.82,
+      "step": 3841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.502185195953912e-08,
+      "loss": 0.8837,
+      "step": 3842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.424952950166761e-08,
+      "loss": 1.02,
+      "step": 3843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3483874151941664e-08,
+      "loss": 0.7615,
+      "step": 3844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.272488642310335e-08,
+      "loss": 0.8901,
+      "step": 3845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1972566823427206e-08,
+      "loss": 0.8521,
+      "step": 3846
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.122691585672134e-08,
+      "loss": 0.9605,
+      "step": 3847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0487934022328533e-08,
+      "loss": 0.9342,
+      "step": 3848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9755621815126265e-08,
+      "loss": 0.9047,
+      "step": 3849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.902997972552558e-08,
+      "loss": 0.95,
+      "step": 3850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8311008239468874e-08,
+      "loss": 0.9039,
+      "step": 3851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.759870783843433e-08,
+      "loss": 0.8397,
+      "step": 3852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6893078999429285e-08,
+      "loss": 0.8329,
+      "step": 3853
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.619412219499685e-08,
+      "loss": 0.8577,
+      "step": 3854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5501837893209266e-08,
+      "loss": 0.8123,
+      "step": 3855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.481622655767125e-08,
+      "loss": 0.8427,
+      "step": 3856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.413728864751997e-08,
+      "loss": 0.8744,
+      "step": 3857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.346502461742174e-08,
+      "loss": 0.7673,
+      "step": 3858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.27994349175742e-08,
+      "loss": 0.8499,
+      "step": 3859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2140519993706375e-08,
+      "loss": 0.791,
+      "step": 3860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.148828028707529e-08,
+      "loss": 0.7977,
+      "step": 3861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0842716234470435e-08,
+      "loss": 0.7162,
+      "step": 3862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0203828268208227e-08,
+      "loss": 0.8045,
+      "step": 3863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9571616816135295e-08,
+      "loss": 0.9054,
+      "step": 3864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8946082301627433e-08,
+      "loss": 0.7821,
+      "step": 3865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8327225143589543e-08,
+      "loss": 0.9099,
+      "step": 3866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.771504575645456e-08,
+      "loss": 0.8315,
+      "step": 3867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7109544550182333e-08,
+      "loss": 0.8086,
+      "step": 3868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.651072193026072e-08,
+      "loss": 0.9446,
+      "step": 3869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.591857829770672e-08,
+      "loss": 0.8418,
+      "step": 3870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5333114049063135e-08,
+      "loss": 0.9331,
+      "step": 3871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4754329576400782e-08,
+      "loss": 0.8131,
+      "step": 3872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4182225267316283e-08,
+      "loss": 0.9217,
+      "step": 3873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3616801504932065e-08,
+      "loss": 0.762,
+      "step": 3874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3058058667899676e-08,
+      "loss": 0.9004,
+      "step": 3875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.250599713039314e-08,
+      "loss": 0.8403,
+      "step": 3876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1960617262114515e-08,
+      "loss": 1.0227,
+      "step": 3877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.142191942829053e-08,
+      "loss": 0.7707,
+      "step": 3878
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0889903989673722e-08,
+      "loss": 0.8855,
+      "step": 3879
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0364571302540213e-08,
+      "loss": 0.8926,
+      "step": 3880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.984592171869082e-08,
+      "loss": 0.8805,
+      "step": 3881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.933395558545437e-08,
+      "loss": 0.9268,
+      "step": 3882
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.882867324567883e-08,
+      "loss": 0.8442,
+      "step": 3883
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8330075037739093e-08,
+      "loss": 0.7945,
+      "step": 3884
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7838161295534728e-08,
+      "loss": 0.9029,
+      "step": 3885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7352932348486674e-08,
+      "loss": 0.7858,
+      "step": 3886
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6874388521540553e-08,
+      "loss": 0.852,
+      "step": 3887
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6402530135164462e-08,
+      "loss": 0.8552,
+      "step": 3888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5937357505348972e-08,
+      "loss": 0.7772,
+      "step": 3889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5478870943610447e-08,
+      "loss": 0.7708,
+      "step": 3890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5027070756983285e-08,
+      "loss": 0.8621,
+      "step": 3891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4581957248026579e-08,
+      "loss": 0.7927,
+      "step": 3892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4143530714822996e-08,
+      "loss": 0.8099,
+      "step": 3893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3711791450974344e-08,
+      "loss": 0.8524,
+      "step": 3894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3286739745603794e-08,
+      "loss": 0.8374,
+      "step": 3895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2868375883360318e-08,
+      "loss": 0.8588,
+      "step": 3896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2456700144408696e-08,
+      "loss": 0.8733,
+      "step": 3897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2051712804439509e-08,
+      "loss": 0.8741,
+      "step": 3898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1653414134661367e-08,
+      "loss": 0.8922,
+      "step": 3899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1261804401804244e-08,
+      "loss": 0.7856,
+      "step": 3900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.087688386811947e-08,
+      "loss": 0.7973,
+      "step": 3901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0498652791378628e-08,
+      "loss": 0.8147,
+      "step": 3902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0127111424872437e-08,
+      "loss": 0.8407,
+      "step": 3903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.762260017414094e-09,
+      "loss": 0.8354,
+      "step": 3904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.404098813332596e-09,
+      "loss": 0.8942,
+      "step": 3905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.052628052481638e-09,
+      "loss": 0.8582,
+      "step": 3906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.707847970231832e-09,
+      "loss": 0.9443,
+      "step": 3907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.369758797471816e-09,
+      "loss": 0.9073,
+      "step": 3908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.038360760612706e-09,
+      "loss": 0.8155,
+      "step": 3909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.713654081582534e-09,
+      "loss": 0.8155,
+      "step": 3910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.395638977828468e-09,
+      "loss": 0.8344,
+      "step": 3911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.084315662317931e-09,
+      "loss": 0.7994,
+      "step": 3912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.779684343535265e-09,
+      "loss": 0.8894,
+      "step": 3913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.481745225485059e-09,
+      "loss": 0.8704,
+      "step": 3914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.1904985076888245e-09,
+      "loss": 0.7799,
+      "step": 3915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9059443851872124e-09,
+      "loss": 0.8655,
+      "step": 3916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.628083048538901e-09,
+      "loss": 0.8948,
+      "step": 3917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.356914683819492e-09,
+      "loss": 0.8779,
+      "step": 3918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.092439472623723e-09,
+      "loss": 0.7446,
+      "step": 3919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834657592065473e-09,
+      "loss": 0.8147,
+      "step": 3920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.58356921477221e-09,
+      "loss": 0.8981,
+      "step": 3921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3391745088927626e-09,
+      "loss": 0.9204,
+      "step": 3922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.101473638090658e-09,
+      "loss": 0.9369,
+      "step": 3923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.870466761549674e-09,
+      "loss": 0.8776,
+      "step": 3924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6461540339682855e-09,
+      "loss": 0.8866,
+      "step": 3925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4285356055629994e-09,
+      "loss": 0.9021,
+      "step": 3926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2176116220672404e-09,
+      "loss": 0.8653,
+      "step": 3927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0133822247313537e-09,
+      "loss": 0.899,
+      "step": 3928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8158475503226033e-09,
+      "loss": 0.8489,
+      "step": 3929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.625007731124063e-09,
+      "loss": 0.7623,
+      "step": 3930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.440862894936835e-09,
+      "loss": 0.7981,
+      "step": 3931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2634131650778324e-09,
+      "loss": 0.9849,
+      "step": 3932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0926586603808864e-09,
+      "loss": 0.8326,
+      "step": 3933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9285994951956378e-09,
+      "loss": 0.8838,
+      "step": 3934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.771235779388647e-09,
+      "loss": 0.8804,
+      "step": 3935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6205676183411733e-09,
+      "loss": 0.7768,
+      "step": 3936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.476595112953616e-09,
+      "loss": 0.8005,
+      "step": 3937
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3393183596377425e-09,
+      "loss": 0.7582,
+      "step": 3938
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2087374503277905e-09,
+      "loss": 0.6928,
+      "step": 3939
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.084852472467146e-09,
+      "loss": 0.7787,
+      "step": 3940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.676635090205556e-10,
+      "loss": 0.9086,
+      "step": 3941
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.571706384663536e-10,
+      "loss": 0.8338,
+      "step": 3942
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.533739347975743e-10,
+      "loss": 0.9123,
+      "step": 3943
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.562734675241711e-10,
+      "loss": 0.7384,
+      "step": 3944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.658693016730165e-10,
+      "loss": 0.8677,
+      "step": 3945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.821614977845724e-10,
+      "loss": 0.7812,
+      "step": 3946
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0515011191621933e-10,
+      "loss": 0.9898,
+      "step": 3947
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.348351956400375e-10,
+      "loss": 0.8252,
+      "step": 3948
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7121679604391606e-10,
+      "loss": 0.8639,
+      "step": 3949
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1429495573266345e-10,
+      "loss": 0.8693,
+      "step": 3950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.640697128246771e-10,
+      "loss": 0.8034,
+      "step": 3951
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2054110095416348e-10,
+      "loss": 0.8537,
+      "step": 3952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.370914927113837e-11,
+      "loss": 0.7883,
+      "step": 3953
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.357388244142669e-11,
+      "loss": 0.8335,
+      "step": 3954
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.013532064555236e-11,
+      "loss": 0.8741,
+      "step": 3955
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.339347957984849e-11,
+      "loss": 0.8735,
+      "step": 3956
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.348370455347194e-12,
+      "loss": 0.8378,
+      "step": 3957
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8221,
+      "step": 3958
+    },
+    {
+      "epoch": 1.0,
+      "step": 3958,
+      "total_flos": 110893502029824.0,
+      "train_loss": 0.9083503452649678,
+      "train_runtime": 33080.4874,
+      "train_samples_per_second": 1.914,
+      "train_steps_per_second": 0.12
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3958,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 110893502029824.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}