diff --git "a/checkpoint-4347/trainer_state.json" "b/checkpoint-4347/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-4347/trainer_state.json"
@@ -0,0 +1,26107 @@
+{
+  "best_metric": 0.028007030487060547,
+  "best_model_checkpoint": "./phishing-email_sender-detection/checkpoint-4347",
+  "epoch": 1.0,
+  "global_step": 4347,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6686,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.69,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.7391,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7499,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6578,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.6891,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.6242,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7033,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 0.7028,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.7133,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.6853,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6343,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.648,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.6961,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.7181,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6627,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.6816,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6447,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-05,
+      "loss": 0.6826,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.6383,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 0.6439,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.6034,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.6405,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 0.5797,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 0.54,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.5859,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.5354,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.5083,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 0.4484,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 0.3968,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 0.5463,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.3513,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3e-05,
+      "loss": 0.4822,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.2732,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-05,
+      "loss": 0.416,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 0.4949,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-05,
+      "loss": 0.2179,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 0.2175,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.4758,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 0.3614,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1e-05,
+      "loss": 0.3209,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 0.3763,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3e-05,
+      "loss": 0.2795,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.1956,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-05,
+      "loss": 0.0708,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.0829,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7e-05,
+      "loss": 0.3993,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-05,
+      "loss": 0.2465,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9e-05,
+      "loss": 0.2277,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 0.0701,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999421564090699e-05,
+      "loss": 0.0468,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998843128181398e-05,
+      "loss": 0.2811,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998264692272096e-05,
+      "loss": 0.03,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997686256362795e-05,
+      "loss": 0.2636,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9971078204534936e-05,
+      "loss": 0.018,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996529384544193e-05,
+      "loss": 0.1131,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9959509486348916e-05,
+      "loss": 0.367,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99537251272559e-05,
+      "loss": 0.4632,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994794076816289e-05,
+      "loss": 0.3284,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994215640906988e-05,
+      "loss": 0.2216,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9936372049976863e-05,
+      "loss": 0.0559,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993058769088385e-05,
+      "loss": 0.2753,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9924803331790844e-05,
+      "loss": 0.5006,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991901897269783e-05,
+      "loss": 1.164,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991323461360482e-05,
+      "loss": 1.2348,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99074502545118e-05,
+      "loss": 0.3093,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990166589541879e-05,
+      "loss": 0.8251,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989588153632578e-05,
+      "loss": 0.1856,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9890097177232764e-05,
+      "loss": 0.2501,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988431281813975e-05,
+      "loss": 0.0852,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9878528459046745e-05,
+      "loss": 0.1855,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9872744099953725e-05,
+      "loss": 0.3556,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986695974086071e-05,
+      "loss": 0.1568,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9861175381767705e-05,
+      "loss": 0.073,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985539102267469e-05,
+      "loss": 0.0755,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984960666358168e-05,
+      "loss": 0.4994,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9843822304488665e-05,
+      "loss": 0.1714,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.983803794539565e-05,
+      "loss": 0.2064,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.983225358630264e-05,
+      "loss": 0.1902,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9826469227209625e-05,
+      "loss": 0.1104,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.982068486811661e-05,
+      "loss": 0.0493,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9814900509023606e-05,
+      "loss": 0.0396,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.980911614993059e-05,
+      "loss": 0.2298,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.980333179083758e-05,
+      "loss": 0.0309,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9797547431744566e-05,
+      "loss": 0.0096,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.979176307265155e-05,
+      "loss": 0.018,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.978597871355854e-05,
+      "loss": 0.0082,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9780194354465526e-05,
+      "loss": 0.1769,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977440999537252e-05,
+      "loss": 0.5457,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9768625636279507e-05,
+      "loss": 0.051,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9762841277186487e-05,
+      "loss": 0.3768,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975705691809347e-05,
+      "loss": 0.1029,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975127255900047e-05,
+      "loss": 0.0572,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9745488199907454e-05,
+      "loss": 0.9281,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973970384081444e-05,
+      "loss": 0.0235,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973391948172143e-05,
+      "loss": 0.1812,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9728135122628414e-05,
+      "loss": 0.0081,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97223507635354e-05,
+      "loss": 0.1703,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971656640444239e-05,
+      "loss": 1.2149,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971078204534938e-05,
+      "loss": 0.4046,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970499768625637e-05,
+      "loss": 0.3914,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9699213327163354e-05,
+      "loss": 0.0311,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969342896807034e-05,
+      "loss": 0.3601,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968764460897733e-05,
+      "loss": 0.084,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9681860249884315e-05,
+      "loss": 0.3128,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96760758907913e-05,
+      "loss": 0.2018,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967029153169829e-05,
+      "loss": 0.0376,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966450717260528e-05,
+      "loss": 0.2232,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965872281351226e-05,
+      "loss": 0.072,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965293845441925e-05,
+      "loss": 0.1952,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.964715409532624e-05,
+      "loss": 0.294,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.964136973623323e-05,
+      "loss": 0.1601,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9635585377140216e-05,
+      "loss": 0.0696,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.96298010180472e-05,
+      "loss": 0.0179,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.962401665895419e-05,
+      "loss": 0.3688,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9618232299861176e-05,
+      "loss": 0.2558,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.961244794076816e-05,
+      "loss": 0.013,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960666358167515e-05,
+      "loss": 0.0163,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960087922258214e-05,
+      "loss": 0.0802,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959509486348913e-05,
+      "loss": 0.1117,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9589310504396116e-05,
+      "loss": 0.2908,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95835261453031e-05,
+      "loss": 0.0217,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957774178621009e-05,
+      "loss": 0.0267,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957195742711708e-05,
+      "loss": 0.229,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956617306802406e-05,
+      "loss": 0.1311,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956038870893106e-05,
+      "loss": 0.0741,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9554604349838044e-05,
+      "loss": 0.0294,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9548819990745024e-05,
+      "loss": 0.0082,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954303563165201e-05,
+      "loss": 0.2212,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9537251272559004e-05,
+      "loss": 0.0601,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953146691346599e-05,
+      "loss": 0.0035,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952568255437298e-05,
+      "loss": 0.2483,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951989819527997e-05,
+      "loss": 0.033,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951411383618695e-05,
+      "loss": 0.1431,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950832947709394e-05,
+      "loss": 0.1383,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9502545118000924e-05,
+      "loss": 0.0033,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949676075890792e-05,
+      "loss": 0.3997,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9490976399814905e-05,
+      "loss": 0.3878,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948519204072189e-05,
+      "loss": 0.0031,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947940768162888e-05,
+      "loss": 0.6469,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9473623322535865e-05,
+      "loss": 0.0151,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946783896344285e-05,
+      "loss": 0.0047,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946205460434984e-05,
+      "loss": 0.2129,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945627024525683e-05,
+      "loss": 0.0117,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945048588616382e-05,
+      "loss": 0.0831,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9444701527070806e-05,
+      "loss": 0.1456,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9438917167977786e-05,
+      "loss": 0.0919,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.943313280888478e-05,
+      "loss": 0.1233,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9427348449791766e-05,
+      "loss": 0.1347,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.942156409069875e-05,
+      "loss": 0.2499,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.941577973160574e-05,
+      "loss": 0.0425,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.940999537251273e-05,
+      "loss": 0.1608,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940421101341971e-05,
+      "loss": 0.1165,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93984266543267e-05,
+      "loss": 0.0185,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939264229523369e-05,
+      "loss": 0.0156,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938685793614068e-05,
+      "loss": 0.5129,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938107357704767e-05,
+      "loss": 0.0056,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9375289217954653e-05,
+      "loss": 0.6606,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936950485886164e-05,
+      "loss": 0.0059,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936372049976863e-05,
+      "loss": 0.5182,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9357936140675614e-05,
+      "loss": 0.0095,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93521517815826e-05,
+      "loss": 0.1561,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9346367422489594e-05,
+      "loss": 0.243,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934058306339658e-05,
+      "loss": 0.2158,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933479870430357e-05,
+      "loss": 0.0077,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9329014345210554e-05,
+      "loss": 0.2064,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932322998611754e-05,
+      "loss": 0.0081,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931744562702453e-05,
+      "loss": 0.0458,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9311661267931515e-05,
+      "loss": 0.0765,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930587690883851e-05,
+      "loss": 0.0058,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930009254974549e-05,
+      "loss": 0.355,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9294308190652475e-05,
+      "loss": 0.0316,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928852383155946e-05,
+      "loss": 0.0078,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9282739472466455e-05,
+      "loss": 0.0105,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.927695511337344e-05,
+      "loss": 0.028,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.927117075428043e-05,
+      "loss": 0.1344,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9265386395187415e-05,
+      "loss": 0.0057,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.92596020360944e-05,
+      "loss": 0.1584,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.925381767700139e-05,
+      "loss": 0.1124,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9248033317908376e-05,
+      "loss": 0.6822,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.924224895881537e-05,
+      "loss": 0.0383,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9236464599722356e-05,
+      "loss": 0.0048,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.923068024062934e-05,
+      "loss": 0.2834,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.922489588153632e-05,
+      "loss": 0.0186,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9219111522443316e-05,
+      "loss": 0.6017,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.92133271633503e-05,
+      "loss": 0.003,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.920754280425729e-05,
+      "loss": 0.126,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9201758445164276e-05,
+      "loss": 0.2856,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.919597408607127e-05,
+      "loss": 0.0439,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.919018972697825e-05,
+      "loss": 0.1378,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.918440536788524e-05,
+      "loss": 0.0072,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.917862100879223e-05,
+      "loss": 0.004,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.917283664969922e-05,
+      "loss": 0.1644,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9167052290606204e-05,
+      "loss": 0.0053,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.916126793151319e-05,
+      "loss": 0.0085,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915548357242018e-05,
+      "loss": 0.0044,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9149699213327164e-05,
+      "loss": 0.0069,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914391485423415e-05,
+      "loss": 0.0085,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913813049514114e-05,
+      "loss": 0.006,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913234613604813e-05,
+      "loss": 0.0037,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912656177695512e-05,
+      "loss": 0.0062,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9120777417862105e-05,
+      "loss": 0.2285,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911499305876909e-05,
+      "loss": 0.0036,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.910920869967608e-05,
+      "loss": 0.0027,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9103424340583065e-05,
+      "loss": 0.0719,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.909763998149005e-05,
+      "loss": 0.009,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9091855622397045e-05,
+      "loss": 0.0031,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.908607126330403e-05,
+      "loss": 0.0022,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.908028690421101e-05,
+      "loss": 0.0017,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9074502545118e-05,
+      "loss": 0.0272,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.906871818602499e-05,
+      "loss": 0.2707,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.906293382693198e-05,
+      "loss": 0.0018,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9057149467838966e-05,
+      "loss": 0.2378,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.905136510874595e-05,
+      "loss": 0.6234,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.904558074965294e-05,
+      "loss": 0.0083,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9039796390559926e-05,
+      "loss": 0.0016,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.903401203146691e-05,
+      "loss": 0.0011,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9028227672373906e-05,
+      "loss": 0.0012,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.902244331328089e-05,
+      "loss": 0.0078,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.901665895418788e-05,
+      "loss": 0.0333,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9010874595094867e-05,
+      "loss": 0.0017,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.900509023600185e-05,
+      "loss": 0.0011,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.899930587690884e-05,
+      "loss": 0.0013,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.899352151781583e-05,
+      "loss": 0.001,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.898773715872282e-05,
+      "loss": 0.3453,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.898195279962981e-05,
+      "loss": 0.0076,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.897616844053679e-05,
+      "loss": 0.0019,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8970384081443774e-05,
+      "loss": 0.381,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.896459972235077e-05,
+      "loss": 0.0023,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8958815363257754e-05,
+      "loss": 0.0015,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.895303100416474e-05,
+      "loss": 0.0291,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.894724664507173e-05,
+      "loss": 0.0038,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8941462285978714e-05,
+      "loss": 0.7312,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.89356779268857e-05,
+      "loss": 0.3954,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.892989356779269e-05,
+      "loss": 0.0025,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.892410920869968e-05,
+      "loss": 0.0097,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.891832484960667e-05,
+      "loss": 0.004,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8912540490513655e-05,
+      "loss": 0.0225,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.890675613142064e-05,
+      "loss": 0.3332,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.890097177232763e-05,
+      "loss": 0.4211,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8895187413234615e-05,
+      "loss": 0.0027,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.88894030541416e-05,
+      "loss": 0.0046,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.888361869504859e-05,
+      "loss": 0.0696,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.887783433595558e-05,
+      "loss": 0.0505,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.887204997686257e-05,
+      "loss": 0.2421,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.886626561776955e-05,
+      "loss": 0.0122,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.886048125867654e-05,
+      "loss": 0.3002,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.885469689958353e-05,
+      "loss": 0.2557,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8848912540490516e-05,
+      "loss": 0.0042,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.88431281813975e-05,
+      "loss": 0.0147,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8837343822304496e-05,
+      "loss": 0.3045,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8831559463211476e-05,
+      "loss": 0.1605,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.882577510411846e-05,
+      "loss": 0.1505,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.881999074502545e-05,
+      "loss": 0.0086,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8814206385932443e-05,
+      "loss": 0.0173,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.880842202683943e-05,
+      "loss": 0.3636,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.880263766774642e-05,
+      "loss": 0.0087,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8796853308653404e-05,
+      "loss": 0.2814,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.879106894956039e-05,
+      "loss": 0.0136,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.878528459046738e-05,
+      "loss": 0.0083,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8779500231374364e-05,
+      "loss": 0.0055,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.877371587228136e-05,
+      "loss": 0.0126,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8767931513188344e-05,
+      "loss": 0.0087,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.876214715409533e-05,
+      "loss": 0.0039,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.875636279500231e-05,
+      "loss": 0.1079,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8750578435909305e-05,
+      "loss": 0.0138,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.874479407681629e-05,
+      "loss": 0.3801,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.873900971772328e-05,
+      "loss": 0.3141,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8733225358630265e-05,
+      "loss": 0.3068,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.872744099953726e-05,
+      "loss": 0.0033,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.872165664044424e-05,
+      "loss": 0.0029,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8715872281351225e-05,
+      "loss": 0.4724,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.871008792225822e-05,
+      "loss": 0.0031,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8704303563165205e-05,
+      "loss": 0.0031,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.869851920407219e-05,
+      "loss": 0.2407,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.869273484497918e-05,
+      "loss": 0.0351,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8686950485886166e-05,
+      "loss": 0.0061,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.868116612679315e-05,
+      "loss": 0.0052,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.867538176770014e-05,
+      "loss": 0.0158,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8669597408607126e-05,
+      "loss": 0.0422,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.866381304951412e-05,
+      "loss": 0.0707,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8658028690421106e-05,
+      "loss": 0.0077,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.865224433132809e-05,
+      "loss": 0.0929,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.864645997223508e-05,
+      "loss": 0.7698,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8640675613142066e-05,
+      "loss": 0.0034,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.863489125404905e-05,
+      "loss": 0.005,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.862910689495604e-05,
+      "loss": 0.002,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8623322535863033e-05,
+      "loss": 0.0021,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8617538176770013e-05,
+      "loss": 0.0592,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8611753817677e-05,
+      "loss": 0.0018,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.860596945858399e-05,
+      "loss": 0.39,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.860018509949098e-05,
+      "loss": 0.1523,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.859440074039797e-05,
+      "loss": 0.003,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8588616381304954e-05,
+      "loss": 0.0015,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.858283202221194e-05,
+      "loss": 0.0035,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.857704766311893e-05,
+      "loss": 0.7524,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8571263304025914e-05,
+      "loss": 0.0012,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.85654789449329e-05,
+      "loss": 0.0022,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8559694585839895e-05,
+      "loss": 0.3611,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.855391022674688e-05,
+      "loss": 0.331,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.854812586765387e-05,
+      "loss": 0.0017,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.854234150856085e-05,
+      "loss": 0.002,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.853655714946784e-05,
+      "loss": 0.2507,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.853077279037483e-05,
+      "loss": 0.1567,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8524988431281815e-05,
+      "loss": 0.0535,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.851920407218881e-05,
+      "loss": 0.0041,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8513419713095795e-05,
+      "loss": 0.4619,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8507635354002775e-05,
+      "loss": 0.0431,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.850185099490976e-05,
+      "loss": 0.3499,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8496066635816756e-05,
+      "loss": 0.0424,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.849028227672374e-05,
+      "loss": 0.0045,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.848449791763073e-05,
+      "loss": 0.0034,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8478713558537716e-05,
+      "loss": 0.3277,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.84729291994447e-05,
+      "loss": 0.0044,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.846714484035169e-05,
+      "loss": 0.3726,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8461360481258676e-05,
+      "loss": 0.0059,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.845557612216567e-05,
+      "loss": 0.1017,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8449791763072657e-05,
+      "loss": 0.3543,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.844400740397964e-05,
+      "loss": 0.0166,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.843822304488663e-05,
+      "loss": 0.3458,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.843243868579362e-05,
+      "loss": 0.1463,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8426654326700604e-05,
+      "loss": 0.0381,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.842086996760759e-05,
+      "loss": 0.0448,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.841508560851458e-05,
+      "loss": 0.0677,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.840930124942157e-05,
+      "loss": 0.0432,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.840351689032856e-05,
+      "loss": 0.1435,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.839773253123554e-05,
+      "loss": 0.1617,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.839194817214253e-05,
+      "loss": 0.0717,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.838616381304952e-05,
+      "loss": 0.024,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8380379453956504e-05,
+      "loss": 0.2288,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.837459509486349e-05,
+      "loss": 0.0316,
+      "step": 331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.836881073577048e-05,
+      "loss": 0.0081,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8363026376677465e-05,
+      "loss": 0.0095,
+      "step": 333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.835724201758445e-05,
+      "loss": 0.0062,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.835145765849144e-05,
+      "loss": 0.2147,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.834567329939843e-05,
+      "loss": 0.1067,
+      "step": 336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.833988894030542e-05,
+      "loss": 0.0032,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8334104581212405e-05,
+      "loss": 0.0067,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.832832022211939e-05,
+      "loss": 0.0037,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.832253586302638e-05,
+      "loss": 0.6441,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8316751503933365e-05,
+      "loss": 0.0026,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.831096714484035e-05,
+      "loss": 0.0353,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8305182785747346e-05,
+      "loss": 0.0063,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.829939842665433e-05,
+      "loss": 0.3508,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.829361406756131e-05,
+      "loss": 0.0522,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.82878297084683e-05,
+      "loss": 0.0074,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.828204534937529e-05,
+      "loss": 0.0122,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.827626099028228e-05,
+      "loss": 0.4048,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8270476631189266e-05,
+      "loss": 0.0056,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.826469227209625e-05,
+      "loss": 0.0931,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.825890791300324e-05,
+      "loss": 0.0411,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8253123553910227e-05,
+      "loss": 0.004,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.824733919481721e-05,
+      "loss": 0.3982,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.824155483572421e-05,
+      "loss": 0.0192,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8235770476631194e-05,
+      "loss": 0.3102,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.822998611753818e-05,
+      "loss": 0.0466,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.822420175844517e-05,
+      "loss": 0.0148,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8218417399352154e-05,
+      "loss": 0.2458,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.821263304025914e-05,
+      "loss": 0.0837,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.820684868116613e-05,
+      "loss": 0.2583,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8201064322073114e-05,
+      "loss": 0.0073,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.819527996298011e-05,
+      "loss": 0.2444,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8189495603887094e-05,
+      "loss": 0.0121,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8183711244794074e-05,
+      "loss": 0.0177,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.817792688570107e-05,
+      "loss": 0.1893,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8172142526608055e-05,
+      "loss": 0.1967,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.816635816751504e-05,
+      "loss": 0.0181,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.816057380842203e-05,
+      "loss": 0.0189,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.815478944932902e-05,
+      "loss": 0.0167,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8149005090236e-05,
+      "loss": 0.2791,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.814322073114299e-05,
+      "loss": 0.012,
+      "step": 371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8137436372049975e-05,
+      "loss": 0.009,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.813165201295697e-05,
+      "loss": 0.5513,
+      "step": 373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8125867653863956e-05,
+      "loss": 0.1022,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.812008329477094e-05,
+      "loss": 0.0288,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.811429893567793e-05,
+      "loss": 0.0332,
+      "step": 376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8108514576584916e-05,
+      "loss": 0.0066,
+      "step": 377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.81027302174919e-05,
+      "loss": 0.1278,
+      "step": 378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.809694585839889e-05,
+      "loss": 0.0034,
+      "step": 379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.809116149930588e-05,
+      "loss": 0.0543,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.808537714021287e-05,
+      "loss": 0.2057,
+      "step": 381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8079592781119856e-05,
+      "loss": 0.0031,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8073808422026836e-05,
+      "loss": 0.0121,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.806802406293383e-05,
+      "loss": 0.341,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.806223970384082e-05,
+      "loss": 0.012,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8056455344747803e-05,
+      "loss": 0.0028,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.80506709856548e-05,
+      "loss": 0.1695,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.804488662656178e-05,
+      "loss": 0.2496,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8039102267468764e-05,
+      "loss": 0.0047,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.803331790837575e-05,
+      "loss": 0.1239,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8027533549282744e-05,
+      "loss": 0.2176,
+      "step": 391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.802174919018973e-05,
+      "loss": 0.1195,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.801596483109672e-05,
+      "loss": 0.0036,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8010180472003704e-05,
+      "loss": 0.0073,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.800439611291069e-05,
+      "loss": 0.0033,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.799861175381768e-05,
+      "loss": 0.1354,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7992827394724665e-05,
+      "loss": 0.0141,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.798704303563166e-05,
+      "loss": 0.5054,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7981258676538645e-05,
+      "loss": 0.0255,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.797547431744563e-05,
+      "loss": 0.0055,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.796968995835261e-05,
+      "loss": 0.0106,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7963905599259605e-05,
+      "loss": 0.0491,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.795812124016659e-05,
+      "loss": 0.0218,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.795233688107358e-05,
+      "loss": 0.0033,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7946552521980565e-05,
+      "loss": 0.0884,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.794076816288756e-05,
+      "loss": 0.0016,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.793498380379454e-05,
+      "loss": 0.0889,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7929199444701526e-05,
+      "loss": 0.0826,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.792341508560852e-05,
+      "loss": 0.0018,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7917630726515506e-05,
+      "loss": 0.4704,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.791184636742249e-05,
+      "loss": 0.0012,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.790606200832948e-05,
+      "loss": 0.045,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7900277649236466e-05,
+      "loss": 0.0034,
+      "step": 413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.789449329014345e-05,
+      "loss": 0.0011,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.788870893105044e-05,
+      "loss": 0.3718,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7882924571957426e-05,
+      "loss": 0.0013,
+      "step": 416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.787714021286442e-05,
+      "loss": 0.1332,
+      "step": 417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.787135585377141e-05,
+      "loss": 0.1808,
+      "step": 418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7865571494678394e-05,
+      "loss": 0.1713,
+      "step": 419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.785978713558538e-05,
+      "loss": 0.0019,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.785400277649237e-05,
+      "loss": 0.0029,
+      "step": 421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7848218417399354e-05,
+      "loss": 0.003,
+      "step": 422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.784243405830634e-05,
+      "loss": 0.0282,
+      "step": 423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7836649699213334e-05,
+      "loss": 0.016,
+      "step": 424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.783086534012032e-05,
+      "loss": 0.0479,
+      "step": 425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.78250809810273e-05,
+      "loss": 0.6453,
+      "step": 426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.781929662193429e-05,
+      "loss": 0.0026,
+      "step": 427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.781351226284128e-05,
+      "loss": 0.2436,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.780772790374827e-05,
+      "loss": 0.0054,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7801943544655255e-05,
+      "loss": 0.0036,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.779615918556224e-05,
+      "loss": 0.0074,
+      "step": 431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.779037482646923e-05,
+      "loss": 0.0271,
+      "step": 432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7784590467376215e-05,
+      "loss": 0.0032,
+      "step": 433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.77788061082832e-05,
+      "loss": 0.0024,
+      "step": 434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7773021749190195e-05,
+      "loss": 0.004,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.776723739009718e-05,
+      "loss": 0.0026,
+      "step": 436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.776145303100417e-05,
+      "loss": 0.0039,
+      "step": 437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7755668671911155e-05,
+      "loss": 0.14,
+      "step": 438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.774988431281814e-05,
+      "loss": 0.0058,
+      "step": 439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.774409995372513e-05,
+      "loss": 0.0017,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7738315594632116e-05,
+      "loss": 0.3317,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.77325312355391e-05,
+      "loss": 0.0058,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7726746876446096e-05,
+      "loss": 0.0018,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.772096251735308e-05,
+      "loss": 0.0034,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.771517815826006e-05,
+      "loss": 0.0012,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7709393799167056e-05,
+      "loss": 0.3266,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.770360944007404e-05,
+      "loss": 0.0016,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.769782508098103e-05,
+      "loss": 0.1663,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7692040721888017e-05,
+      "loss": 0.0232,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7686256362795e-05,
+      "loss": 0.1489,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.768047200370199e-05,
+      "loss": 0.0522,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.767468764460898e-05,
+      "loss": 0.0012,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7668903285515964e-05,
+      "loss": 0.0298,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.766311892642296e-05,
+      "loss": 0.0008,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7657334567329944e-05,
+      "loss": 0.0012,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.765155020823693e-05,
+      "loss": 0.0039,
+      "step": 456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.764576584914392e-05,
+      "loss": 0.0009,
+      "step": 457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7639981490050904e-05,
+      "loss": 0.0012,
+      "step": 458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.763419713095789e-05,
+      "loss": 0.0277,
+      "step": 459
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.762841277186488e-05,
+      "loss": 0.044,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.762262841277187e-05,
+      "loss": 0.0874,
+      "step": 461
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.761684405367886e-05,
+      "loss": 0.1852,
+      "step": 462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.761105969458584e-05,
+      "loss": 0.0082,
+      "step": 463
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7605275335492825e-05,
+      "loss": 0.0008,
+      "step": 464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.759949097639982e-05,
+      "loss": 0.4528,
+      "step": 465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7593706617306805e-05,
+      "loss": 0.0006,
+      "step": 466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.758792225821379e-05,
+      "loss": 0.0007,
+      "step": 467
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7582137899120785e-05,
+      "loss": 0.0009,
+      "step": 468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7576353540027765e-05,
+      "loss": 0.0009,
+      "step": 469
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.757056918093475e-05,
+      "loss": 0.0011,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.756478482184174e-05,
+      "loss": 0.0462,
+      "step": 471
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.755900046274873e-05,
+      "loss": 0.0537,
+      "step": 472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.755321610365572e-05,
+      "loss": 0.0127,
+      "step": 473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7547431744562706e-05,
+      "loss": 0.0019,
+      "step": 474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.754164738546969e-05,
+      "loss": 0.3935,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.753586302637668e-05,
+      "loss": 0.0008,
+      "step": 476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7530078667283666e-05,
+      "loss": 0.4934,
+      "step": 477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.752429430819065e-05,
+      "loss": 0.0013,
+      "step": 478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7518509949097646e-05,
+      "loss": 0.0151,
+      "step": 479
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.751272559000463e-05,
+      "loss": 0.0521,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.750694123091162e-05,
+      "loss": 0.0125,
+      "step": 481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.75011568718186e-05,
+      "loss": 0.1556,
+      "step": 482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.749537251272559e-05,
+      "loss": 0.2411,
+      "step": 483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.748958815363258e-05,
+      "loss": 0.1864,
+      "step": 484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.748380379453957e-05,
+      "loss": 0.2453,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7478019435446554e-05,
+      "loss": 0.3357,
+      "step": 486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.747223507635355e-05,
+      "loss": 0.4865,
+      "step": 487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.746645071726053e-05,
+      "loss": 0.3835,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7460666358167514e-05,
+      "loss": 0.4008,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.745488199907451e-05,
+      "loss": 0.0063,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7449097639981494e-05,
+      "loss": 0.0031,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.744331328088848e-05,
+      "loss": 0.0065,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.743752892179547e-05,
+      "loss": 0.0053,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7431744562702454e-05,
+      "loss": 0.3271,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.742596020360944e-05,
+      "loss": 0.0045,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.742017584451643e-05,
+      "loss": 0.0587,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7414391485423415e-05,
+      "loss": 0.1503,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.740860712633041e-05,
+      "loss": 0.0385,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7402822767237395e-05,
+      "loss": 0.0623,
+      "step": 499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.739703840814438e-05,
+      "loss": 0.1824,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.739125404905137e-05,
+      "loss": 0.065,
+      "step": 501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7385469689958355e-05,
+      "loss": 0.2391,
+      "step": 502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.737968533086534e-05,
+      "loss": 0.2553,
+      "step": 503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.737390097177233e-05,
+      "loss": 0.0728,
+      "step": 504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.736811661267932e-05,
+      "loss": 0.0204,
+      "step": 505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.73623322535863e-05,
+      "loss": 0.0203,
+      "step": 506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.735654789449329e-05,
+      "loss": 0.0137,
+      "step": 507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7350763535400276e-05,
+      "loss": 0.1767,
+      "step": 508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.734497917630727e-05,
+      "loss": 0.037,
+      "step": 509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7339194817214256e-05,
+      "loss": 0.1535,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.733341045812124e-05,
+      "loss": 0.0168,
+      "step": 511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.732762609902823e-05,
+      "loss": 0.0463,
+      "step": 512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7321841739935216e-05,
+      "loss": 0.003,
+      "step": 513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.73160573808422e-05,
+      "loss": 0.0031,
+      "step": 514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.731027302174919e-05,
+      "loss": 0.0048,
+      "step": 515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7304488662656183e-05,
+      "loss": 0.0104,
+      "step": 516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.729870430356317e-05,
+      "loss": 0.0018,
+      "step": 517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.729291994447016e-05,
+      "loss": 0.0223,
+      "step": 518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.728713558537714e-05,
+      "loss": 0.0085,
+      "step": 519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.728135122628413e-05,
+      "loss": 0.0796,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.727556686719112e-05,
+      "loss": 0.004,
+      "step": 521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7269782508098104e-05,
+      "loss": 0.0723,
+      "step": 522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.726399814900509e-05,
+      "loss": 0.0038,
+      "step": 523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7258213789912084e-05,
+      "loss": 0.0016,
+      "step": 524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7252429430819064e-05,
+      "loss": 0.0012,
+      "step": 525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.724664507172605e-05,
+      "loss": 0.0009,
+      "step": 526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7240860712633045e-05,
+      "loss": 0.0013,
+      "step": 527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.723507635354003e-05,
+      "loss": 0.0014,
+      "step": 528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.722929199444702e-05,
+      "loss": 0.7049,
+      "step": 529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7223507635354005e-05,
+      "loss": 0.001,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.721772327626099e-05,
+      "loss": 0.0084,
+      "step": 531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.721193891716798e-05,
+      "loss": 0.0042,
+      "step": 532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7206154558074965e-05,
+      "loss": 0.0012,
+      "step": 533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.720037019898195e-05,
+      "loss": 0.0009,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7194585839888945e-05,
+      "loss": 0.0016,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.718880148079593e-05,
+      "loss": 0.0007,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.718301712170292e-05,
+      "loss": 0.0007,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7177232762609906e-05,
+      "loss": 0.0023,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.717144840351689e-05,
+      "loss": 0.0018,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.716566404442388e-05,
+      "loss": 0.0006,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7159879685330866e-05,
+      "loss": 0.0011,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.715409532623786e-05,
+      "loss": 0.0007,
+      "step": 542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7148310967144846e-05,
+      "loss": 0.0008,
+      "step": 543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7142526608051826e-05,
+      "loss": 0.0077,
+      "step": 544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.713674224895881e-05,
+      "loss": 0.0017,
+      "step": 545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7130957889865806e-05,
+      "loss": 0.0006,
+      "step": 546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.712517353077279e-05,
+      "loss": 0.0006,
+      "step": 547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.711938917167978e-05,
+      "loss": 0.0006,
+      "step": 548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7113604812586774e-05,
+      "loss": 0.0006,
+      "step": 549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7107820453493754e-05,
+      "loss": 0.3294,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.710203609440074e-05,
+      "loss": 0.0007,
+      "step": 551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.709625173530773e-05,
+      "loss": 0.0007,
+      "step": 552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.709046737621472e-05,
+      "loss": 0.0007,
+      "step": 553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.708468301712171e-05,
+      "loss": 0.0005,
+      "step": 554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7078898658028694e-05,
+      "loss": 0.2429,
+      "step": 555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.707311429893568e-05,
+      "loss": 0.0005,
+      "step": 556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.706732993984267e-05,
+      "loss": 0.0006,
+      "step": 557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7061545580749654e-05,
+      "loss": 0.0005,
+      "step": 558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.705576122165664e-05,
+      "loss": 0.0004,
+      "step": 559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7049976862563635e-05,
+      "loss": 0.0005,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.704419250347062e-05,
+      "loss": 0.0005,
+      "step": 561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.703840814437761e-05,
+      "loss": 0.0013,
+      "step": 562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.703262378528459e-05,
+      "loss": 0.0006,
+      "step": 563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.702683942619158e-05,
+      "loss": 0.2072,
+      "step": 564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.702105506709857e-05,
+      "loss": 0.0005,
+      "step": 565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7015270708005555e-05,
+      "loss": 0.0006,
+      "step": 566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.700948634891254e-05,
+      "loss": 0.0029,
+      "step": 567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.700370198981953e-05,
+      "loss": 0.5089,
+      "step": 568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6997917630726515e-05,
+      "loss": 0.001,
+      "step": 569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.69921332716335e-05,
+      "loss": 0.0005,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6986348912540496e-05,
+      "loss": 0.0042,
+      "step": 571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.698056455344748e-05,
+      "loss": 0.0006,
+      "step": 572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.697478019435447e-05,
+      "loss": 0.0006,
+      "step": 573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6968995835261456e-05,
+      "loss": 0.0009,
+      "step": 574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.696321147616844e-05,
+      "loss": 0.4147,
+      "step": 575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.695742711707543e-05,
+      "loss": 0.1337,
+      "step": 576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6951642757982416e-05,
+      "loss": 0.433,
+      "step": 577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.69458583988894e-05,
+      "loss": 0.4479,
+      "step": 578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6940074039796397e-05,
+      "loss": 0.2411,
+      "step": 579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.693428968070338e-05,
+      "loss": 0.2091,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.692850532161036e-05,
+      "loss": 0.0013,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.692272096251736e-05,
+      "loss": 0.0038,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6916936603424344e-05,
+      "loss": 0.0016,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.691115224433133e-05,
+      "loss": 0.001,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.690536788523832e-05,
+      "loss": 0.0095,
+      "step": 585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.689958352614531e-05,
+      "loss": 0.2165,
+      "step": 586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.689379916705229e-05,
+      "loss": 0.0013,
+      "step": 587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.688801480795928e-05,
+      "loss": 0.1921,
+      "step": 588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6882230448866264e-05,
+      "loss": 0.5305,
+      "step": 589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.687644608977326e-05,
+      "loss": 0.001,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6870661730680244e-05,
+      "loss": 0.0034,
+      "step": 591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.686487737158723e-05,
+      "loss": 0.0212,
+      "step": 592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.685909301249422e-05,
+      "loss": 0.0013,
+      "step": 593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6853308653401205e-05,
+      "loss": 0.0022,
+      "step": 594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.684752429430819e-05,
+      "loss": 0.0024,
+      "step": 595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.684173993521518e-05,
+      "loss": 0.0055,
+      "step": 596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.683595557612217e-05,
+      "loss": 0.002,
+      "step": 597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.683017121702916e-05,
+      "loss": 0.0549,
+      "step": 598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6824386857936145e-05,
+      "loss": 0.0068,
+      "step": 599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6818602498843125e-05,
+      "loss": 0.0029,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.681281813975012e-05,
+      "loss": 0.0022,
+      "step": 601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6807033780657106e-05,
+      "loss": 0.0418,
+      "step": 602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.680124942156409e-05,
+      "loss": 0.0018,
+      "step": 603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.679546506247108e-05,
+      "loss": 0.001,
+      "step": 604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.678968070337807e-05,
+      "loss": 0.0012,
+      "step": 605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.678389634428505e-05,
+      "loss": 0.0006,
+      "step": 606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.677811198519204e-05,
+      "loss": 0.0299,
+      "step": 607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.677232762609903e-05,
+      "loss": 0.0007,
+      "step": 608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.676654326700602e-05,
+      "loss": 0.2463,
+      "step": 609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6760758907913006e-05,
+      "loss": 0.0309,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.675497454881999e-05,
+      "loss": 0.0009,
+      "step": 611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.674919018972698e-05,
+      "loss": 0.0008,
+      "step": 612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.674340583063397e-05,
+      "loss": 0.0008,
+      "step": 613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6737621471540953e-05,
+      "loss": 0.1538,
+      "step": 614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.673183711244794e-05,
+      "loss": 0.2878,
+      "step": 615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6726052753354934e-05,
+      "loss": 0.0004,
+      "step": 616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.672026839426192e-05,
+      "loss": 0.0009,
+      "step": 617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.671448403516891e-05,
+      "loss": 0.0022,
+      "step": 618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6708699676075894e-05,
+      "loss": 0.002,
+      "step": 619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.670291531698288e-05,
+      "loss": 0.0008,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.669713095788987e-05,
+      "loss": 0.002,
+      "step": 621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6691346598796854e-05,
+      "loss": 0.4646,
+      "step": 622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.668556223970385e-05,
+      "loss": 0.001,
+      "step": 623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.667977788061083e-05,
+      "loss": 0.0005,
+      "step": 624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6673993521517814e-05,
+      "loss": 0.3546,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.66682091624248e-05,
+      "loss": 0.0013,
+      "step": 626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6662424803331795e-05,
+      "loss": 0.0005,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.665664044423878e-05,
+      "loss": 0.001,
+      "step": 628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.665085608514577e-05,
+      "loss": 0.0007,
+      "step": 629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6645071726052755e-05,
+      "loss": 0.0013,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.663928736695974e-05,
+      "loss": 0.0011,
+      "step": 631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.663350300786673e-05,
+      "loss": 0.0011,
+      "step": 632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6627718648773715e-05,
+      "loss": 0.0025,
+      "step": 633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.662193428968071e-05,
+      "loss": 0.001,
+      "step": 634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6616149930587696e-05,
+      "loss": 0.0008,
+      "step": 635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.661036557149468e-05,
+      "loss": 0.0015,
+      "step": 636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.660458121240166e-05,
+      "loss": 0.0018,
+      "step": 637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6598796853308656e-05,
+      "loss": 0.0108,
+      "step": 638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.659301249421564e-05,
+      "loss": 0.0013,
+      "step": 639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.658722813512263e-05,
+      "loss": 0.0016,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.658144377602962e-05,
+      "loss": 0.0119,
+      "step": 641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.657565941693661e-05,
+      "loss": 0.0005,
+      "step": 642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.656987505784359e-05,
+      "loss": 0.001,
+      "step": 643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6564090698750576e-05,
+      "loss": 0.0226,
+      "step": 644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.655830633965757e-05,
+      "loss": 0.0145,
+      "step": 645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.655252198056456e-05,
+      "loss": 0.5024,
+      "step": 646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6546737621471543e-05,
+      "loss": 0.0008,
+      "step": 647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.654095326237853e-05,
+      "loss": 0.0007,
+      "step": 648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.653516890328552e-05,
+      "loss": 0.002,
+      "step": 649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6529384544192504e-05,
+      "loss": 0.0309,
+      "step": 650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.652360018509949e-05,
+      "loss": 0.0018,
+      "step": 651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6517815826006484e-05,
+      "loss": 0.0016,
+      "step": 652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.651203146691347e-05,
+      "loss": 0.0011,
+      "step": 653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.650624710782046e-05,
+      "loss": 0.0008,
+      "step": 654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6500462748727444e-05,
+      "loss": 0.001,
+      "step": 655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.649467838963443e-05,
+      "loss": 0.0016,
+      "step": 656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.648889403054142e-05,
+      "loss": 0.0015,
+      "step": 657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6483109671448405e-05,
+      "loss": 0.0011,
+      "step": 658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.647732531235539e-05,
+      "loss": 0.0013,
+      "step": 659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6471540953262385e-05,
+      "loss": 0.4133,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.646575659416937e-05,
+      "loss": 0.0014,
+      "step": 661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.645997223507635e-05,
+      "loss": 0.0012,
+      "step": 662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6454187875983345e-05,
+      "loss": 0.004,
+      "step": 663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.644840351689033e-05,
+      "loss": 0.0189,
+      "step": 664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.644261915779732e-05,
+      "loss": 0.0013,
+      "step": 665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6436834798704305e-05,
+      "loss": 0.0162,
+      "step": 666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.643105043961129e-05,
+      "loss": 0.0034,
+      "step": 667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.642526608051828e-05,
+      "loss": 0.0011,
+      "step": 668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6419481721425266e-05,
+      "loss": 0.0017,
+      "step": 669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.641369736233225e-05,
+      "loss": 0.0014,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6407913003239246e-05,
+      "loss": 0.0016,
+      "step": 671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.640212864414623e-05,
+      "loss": 0.0009,
+      "step": 672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.639634428505322e-05,
+      "loss": 0.0012,
+      "step": 673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6390559925960206e-05,
+      "loss": 0.0008,
+      "step": 674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.638477556686719e-05,
+      "loss": 0.0009,
+      "step": 675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.637899120777418e-05,
+      "loss": 0.001,
+      "step": 676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6373206848681167e-05,
+      "loss": 0.0009,
+      "step": 677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.636742248958816e-05,
+      "loss": 0.0011,
+      "step": 678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.636163813049515e-05,
+      "loss": 0.0009,
+      "step": 679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.635585377140213e-05,
+      "loss": 0.0007,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6350069412309114e-05,
+      "loss": 0.0008,
+      "step": 681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.634428505321611e-05,
+      "loss": 0.0021,
+      "step": 682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6338500694123094e-05,
+      "loss": 0.0008,
+      "step": 683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.633271633503008e-05,
+      "loss": 0.1027,
+      "step": 684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.632693197593707e-05,
+      "loss": 0.4195,
+      "step": 685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6321147616844054e-05,
+      "loss": 0.1575,
+      "step": 686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.631536325775104e-05,
+      "loss": 0.0024,
+      "step": 687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.630957889865803e-05,
+      "loss": 0.0005,
+      "step": 688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.630379453956502e-05,
+      "loss": 0.0007,
+      "step": 689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.629801018047201e-05,
+      "loss": 0.0014,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6292225821378995e-05,
+      "loss": 0.0689,
+      "step": 691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.628644146228598e-05,
+      "loss": 0.2848,
+      "step": 692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.628065710319297e-05,
+      "loss": 0.0011,
+      "step": 693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6274872744099955e-05,
+      "loss": 0.0013,
+      "step": 694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.626908838500694e-05,
+      "loss": 0.4072,
+      "step": 695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.626330402591393e-05,
+      "loss": 0.0007,
+      "step": 696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.625751966682092e-05,
+      "loss": 0.0011,
+      "step": 697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.625173530772791e-05,
+      "loss": 0.001,
+      "step": 698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.624595094863489e-05,
+      "loss": 0.001,
+      "step": 699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.624016658954188e-05,
+      "loss": 0.0013,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.623438223044887e-05,
+      "loss": 0.0772,
+      "step": 701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6228597871355856e-05,
+      "loss": 0.001,
+      "step": 702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.622281351226284e-05,
+      "loss": 0.0948,
+      "step": 703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6217029153169836e-05,
+      "loss": 0.0014,
+      "step": 704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6211244794076816e-05,
+      "loss": 0.0011,
+      "step": 705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.62054604349838e-05,
+      "loss": 0.0014,
+      "step": 706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.619967607589079e-05,
+      "loss": 0.3036,
+      "step": 707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.619389171679778e-05,
+      "loss": 0.0011,
+      "step": 708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.618810735770477e-05,
+      "loss": 0.0187,
+      "step": 709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6182322998611757e-05,
+      "loss": 0.3572,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.617653863951874e-05,
+      "loss": 0.1583,
+      "step": 711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.617075428042573e-05,
+      "loss": 0.0285,
+      "step": 712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.616496992133272e-05,
+      "loss": 0.002,
+      "step": 713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6159185562239704e-05,
+      "loss": 0.1407,
+      "step": 714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.61534012031467e-05,
+      "loss": 0.0025,
+      "step": 715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6147616844053684e-05,
+      "loss": 0.0033,
+      "step": 716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.614183248496067e-05,
+      "loss": 0.027,
+      "step": 717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.613604812586765e-05,
+      "loss": 0.0021,
+      "step": 718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6130263766774644e-05,
+      "loss": 0.0057,
+      "step": 719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.612447940768163e-05,
+      "loss": 0.3394,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.611869504858862e-05,
+      "loss": 0.0026,
+      "step": 721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.611291068949561e-05,
+      "loss": 0.2561,
+      "step": 722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.61071263304026e-05,
+      "loss": 0.2555,
+      "step": 723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.610134197130958e-05,
+      "loss": 0.0048,
+      "step": 724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6095557612216565e-05,
+      "loss": 0.0092,
+      "step": 725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.608977325312356e-05,
+      "loss": 0.0088,
+      "step": 726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6083988894030545e-05,
+      "loss": 0.3234,
+      "step": 727
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.607820453493753e-05,
+      "loss": 0.0052,
+      "step": 728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.607242017584452e-05,
+      "loss": 0.0132,
+      "step": 729
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6066635816751505e-05,
+      "loss": 0.0252,
+      "step": 730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.606085145765849e-05,
+      "loss": 0.4563,
+      "step": 731
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.605506709856548e-05,
+      "loss": 0.0029,
+      "step": 732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.604928273947247e-05,
+      "loss": 0.0151,
+      "step": 733
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.604349838037946e-05,
+      "loss": 0.5312,
+      "step": 734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6037714021286446e-05,
+      "loss": 0.0562,
+      "step": 735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.603192966219343e-05,
+      "loss": 0.0742,
+      "step": 736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.602614530310042e-05,
+      "loss": 0.1167,
+      "step": 737
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6020360944007406e-05,
+      "loss": 0.2403,
+      "step": 738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.601457658491439e-05,
+      "loss": 0.0323,
+      "step": 739
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.600879222582138e-05,
+      "loss": 0.0317,
+      "step": 740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.600300786672837e-05,
+      "loss": 0.0074,
+      "step": 741
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.599722350763535e-05,
+      "loss": 0.0034,
+      "step": 742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.599143914854234e-05,
+      "loss": 0.0041,
+      "step": 743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5985654789449333e-05,
+      "loss": 0.0089,
+      "step": 744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.597987043035632e-05,
+      "loss": 0.0026,
+      "step": 745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.597408607126331e-05,
+      "loss": 0.0043,
+      "step": 746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5968301712170294e-05,
+      "loss": 0.0046,
+      "step": 747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.596251735307728e-05,
+      "loss": 0.0562,
+      "step": 748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.595673299398427e-05,
+      "loss": 0.0047,
+      "step": 749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5950948634891254e-05,
+      "loss": 0.0016,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.594516427579824e-05,
+      "loss": 0.0256,
+      "step": 751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5939379916705234e-05,
+      "loss": 0.0048,
+      "step": 752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.593359555761222e-05,
+      "loss": 0.0022,
+      "step": 753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.592781119851921e-05,
+      "loss": 0.0012,
+      "step": 754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5922026839426195e-05,
+      "loss": 0.0017,
+      "step": 755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.591624248033318e-05,
+      "loss": 0.3095,
+      "step": 756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.591045812124017e-05,
+      "loss": 0.1266,
+      "step": 757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5904673762147155e-05,
+      "loss": 0.0008,
+      "step": 758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.589888940305415e-05,
+      "loss": 0.0041,
+      "step": 759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5893105043961135e-05,
+      "loss": 0.001,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5887320684868115e-05,
+      "loss": 0.0081,
+      "step": 761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.58815363257751e-05,
+      "loss": 0.0011,
+      "step": 762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5875751966682095e-05,
+      "loss": 0.0477,
+      "step": 763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.586996760758908e-05,
+      "loss": 0.077,
+      "step": 764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.586418324849607e-05,
+      "loss": 0.0005,
+      "step": 765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5858398889403056e-05,
+      "loss": 0.0004,
+      "step": 766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.585261453031004e-05,
+      "loss": 0.0557,
+      "step": 767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.584683017121703e-05,
+      "loss": 0.8505,
+      "step": 768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5841045812124016e-05,
+      "loss": 0.0013,
+      "step": 769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.583526145303101e-05,
+      "loss": 0.0007,
+      "step": 770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5829477093937996e-05,
+      "loss": 0.371,
+      "step": 771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.582369273484498e-05,
+      "loss": 0.0021,
+      "step": 772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.581790837575197e-05,
+      "loss": 0.4475,
+      "step": 773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5812124016658956e-05,
+      "loss": 0.0774,
+      "step": 774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.580633965756594e-05,
+      "loss": 0.0056,
+      "step": 775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.580055529847293e-05,
+      "loss": 0.0014,
+      "step": 776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.579477093937992e-05,
+      "loss": 0.0225,
+      "step": 777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.578898658028691e-05,
+      "loss": 0.0307,
+      "step": 778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.57832022211939e-05,
+      "loss": 0.0019,
+      "step": 779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.577741786210088e-05,
+      "loss": 0.0044,
+      "step": 780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.577163350300787e-05,
+      "loss": 0.4422,
+      "step": 781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.576584914391486e-05,
+      "loss": 0.0044,
+      "step": 782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5760064784821844e-05,
+      "loss": 0.2247,
+      "step": 783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.575428042572883e-05,
+      "loss": 0.0063,
+      "step": 784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.574849606663582e-05,
+      "loss": 0.0146,
+      "step": 785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5742711707542804e-05,
+      "loss": 0.0038,
+      "step": 786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.573692734844979e-05,
+      "loss": 0.0152,
+      "step": 787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.573114298935678e-05,
+      "loss": 0.0089,
+      "step": 788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.572535863026377e-05,
+      "loss": 0.0041,
+      "step": 789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.571957427117076e-05,
+      "loss": 0.0079,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5713789912077745e-05,
+      "loss": 0.0071,
+      "step": 791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.570800555298473e-05,
+      "loss": 0.0082,
+      "step": 792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.570222119389172e-05,
+      "loss": 0.0028,
+      "step": 793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5696436834798705e-05,
+      "loss": 0.0056,
+      "step": 794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.569065247570569e-05,
+      "loss": 0.0118,
+      "step": 795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5684868116612685e-05,
+      "loss": 0.5844,
+      "step": 796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.567908375751967e-05,
+      "loss": 0.0017,
+      "step": 797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.567329939842665e-05,
+      "loss": 0.3816,
+      "step": 798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.566751503933364e-05,
+      "loss": 0.3859,
+      "step": 799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.566173068024063e-05,
+      "loss": 0.0024,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.565594632114762e-05,
+      "loss": 0.1228,
+      "step": 801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5650161962054606e-05,
+      "loss": 0.101,
+      "step": 802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.56443776029616e-05,
+      "loss": 0.0656,
+      "step": 803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.563859324386858e-05,
+      "loss": 0.0049,
+      "step": 804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5632808884775566e-05,
+      "loss": 0.0047,
+      "step": 805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.562702452568255e-05,
+      "loss": 0.0038,
+      "step": 806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5621240166589547e-05,
+      "loss": 0.086,
+      "step": 807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.561545580749653e-05,
+      "loss": 0.3525,
+      "step": 808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.560967144840352e-05,
+      "loss": 0.0046,
+      "step": 809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.560388708931051e-05,
+      "loss": 0.003,
+      "step": 810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5598102730217494e-05,
+      "loss": 0.0033,
+      "step": 811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.559231837112448e-05,
+      "loss": 0.0046,
+      "step": 812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.558653401203147e-05,
+      "loss": 0.4666,
+      "step": 813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.558074965293846e-05,
+      "loss": 0.0521,
+      "step": 814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.557496529384545e-05,
+      "loss": 0.0058,
+      "step": 815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5569180934752434e-05,
+      "loss": 0.0019,
+      "step": 816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5563396575659414e-05,
+      "loss": 0.0052,
+      "step": 817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.555761221656641e-05,
+      "loss": 0.0039,
+      "step": 818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5551827857473394e-05,
+      "loss": 0.1693,
+      "step": 819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.554604349838038e-05,
+      "loss": 0.154,
+      "step": 820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.554025913928737e-05,
+      "loss": 0.1258,
+      "step": 821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.553447478019436e-05,
+      "loss": 0.0047,
+      "step": 822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.552869042110134e-05,
+      "loss": 0.0067,
+      "step": 823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.552290606200833e-05,
+      "loss": 0.2482,
+      "step": 824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.551712170291532e-05,
+      "loss": 0.0042,
+      "step": 825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.551133734382231e-05,
+      "loss": 0.0029,
+      "step": 826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5505552984729295e-05,
+      "loss": 0.0419,
+      "step": 827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.549976862563628e-05,
+      "loss": 0.017,
+      "step": 828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.549398426654327e-05,
+      "loss": 0.0429,
+      "step": 829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5488199907450256e-05,
+      "loss": 0.2032,
+      "step": 830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.548241554835724e-05,
+      "loss": 0.0084,
+      "step": 831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.547663118926423e-05,
+      "loss": 0.0873,
+      "step": 832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.547084683017122e-05,
+      "loss": 0.3456,
+      "step": 833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.546506247107821e-05,
+      "loss": 0.0013,
+      "step": 834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5459278111985196e-05,
+      "loss": 0.4722,
+      "step": 835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.545349375289218e-05,
+      "loss": 0.0024,
+      "step": 836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.544770939379917e-05,
+      "loss": 0.042,
+      "step": 837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5441925034706156e-05,
+      "loss": 0.0028,
+      "step": 838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.543614067561314e-05,
+      "loss": 0.3532,
+      "step": 839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.543035631652014e-05,
+      "loss": 0.0037,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.542457195742712e-05,
+      "loss": 0.0731,
+      "step": 841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.54187875983341e-05,
+      "loss": 0.0038,
+      "step": 842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.541300323924109e-05,
+      "loss": 0.0038,
+      "step": 843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5407218880148084e-05,
+      "loss": 0.0053,
+      "step": 844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.540143452105507e-05,
+      "loss": 0.2069,
+      "step": 845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.539565016196206e-05,
+      "loss": 0.003,
+      "step": 846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5389865802869044e-05,
+      "loss": 0.0061,
+      "step": 847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.538408144377603e-05,
+      "loss": 0.0067,
+      "step": 848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.537829708468302e-05,
+      "loss": 0.0071,
+      "step": 849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5372512725590004e-05,
+      "loss": 0.0046,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5366728366497e-05,
+      "loss": 0.0025,
+      "step": 851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5360944007403984e-05,
+      "loss": 0.0023,
+      "step": 852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.535515964831097e-05,
+      "loss": 0.0013,
+      "step": 853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.534937528921796e-05,
+      "loss": 0.0013,
+      "step": 854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5343590930124945e-05,
+      "loss": 0.4117,
+      "step": 855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.533780657103193e-05,
+      "loss": 0.0036,
+      "step": 856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.533202221193892e-05,
+      "loss": 0.0012,
+      "step": 857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5326237852845905e-05,
+      "loss": 0.0015,
+      "step": 858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.53204534937529e-05,
+      "loss": 0.0022,
+      "step": 859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.531466913465988e-05,
+      "loss": 0.0758,
+      "step": 860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5308884775566865e-05,
+      "loss": 0.0014,
+      "step": 861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.530310041647386e-05,
+      "loss": 0.0012,
+      "step": 862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5297316057380846e-05,
+      "loss": 0.004,
+      "step": 863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.529153169828783e-05,
+      "loss": 0.2118,
+      "step": 864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.528574733919482e-05,
+      "loss": 0.0032,
+      "step": 865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5279962980101806e-05,
+      "loss": 0.0025,
+      "step": 866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.527417862100879e-05,
+      "loss": 0.0021,
+      "step": 867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526839426191578e-05,
+      "loss": 0.0193,
+      "step": 868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5262609902822766e-05,
+      "loss": 0.0057,
+      "step": 869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.525682554372976e-05,
+      "loss": 0.0025,
+      "step": 870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5251041184636746e-05,
+      "loss": 0.0015,
+      "step": 871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.524525682554373e-05,
+      "loss": 0.0011,
+      "step": 872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.523947246645072e-05,
+      "loss": 0.2657,
+      "step": 873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.523368810735771e-05,
+      "loss": 0.0013,
+      "step": 874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5227903748264693e-05,
+      "loss": 0.0736,
+      "step": 875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.522211938917168e-05,
+      "loss": 0.0667,
+      "step": 876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5216335030078674e-05,
+      "loss": 0.3045,
+      "step": 877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.521055067098566e-05,
+      "loss": 0.0007,
+      "step": 878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.520476631189264e-05,
+      "loss": 0.0675,
+      "step": 879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.519898195279963e-05,
+      "loss": 0.0016,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.519319759370662e-05,
+      "loss": 0.0031,
+      "step": 881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.518741323461361e-05,
+      "loss": 0.0163,
+      "step": 882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5181628875520594e-05,
+      "loss": 0.6624,
+      "step": 883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.517584451642759e-05,
+      "loss": 0.0056,
+      "step": 884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.517006015733457e-05,
+      "loss": 0.02,
+      "step": 885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5164275798241555e-05,
+      "loss": 0.0122,
+      "step": 886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.515849143914854e-05,
+      "loss": 0.0015,
+      "step": 887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5152707080055535e-05,
+      "loss": 0.0773,
+      "step": 888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.514692272096252e-05,
+      "loss": 0.0033,
+      "step": 889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.514113836186951e-05,
+      "loss": 0.0089,
+      "step": 890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5135354002776495e-05,
+      "loss": 0.0273,
+      "step": 891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.512956964368348e-05,
+      "loss": 0.0101,
+      "step": 892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.512378528459047e-05,
+      "loss": 0.0046,
+      "step": 893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5118000925497455e-05,
+      "loss": 0.002,
+      "step": 894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.511221656640445e-05,
+      "loss": 0.0018,
+      "step": 895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5106432207311436e-05,
+      "loss": 0.0016,
+      "step": 896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.510064784821842e-05,
+      "loss": 0.0564,
+      "step": 897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.50948634891254e-05,
+      "loss": 0.0013,
+      "step": 898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5089079130032396e-05,
+      "loss": 0.0008,
+      "step": 899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.508329477093938e-05,
+      "loss": 0.0023,
+      "step": 900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.507751041184637e-05,
+      "loss": 0.0008,
+      "step": 901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5071726052753356e-05,
+      "loss": 0.2956,
+      "step": 902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.506594169366034e-05,
+      "loss": 0.2933,
+      "step": 903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.506015733456733e-05,
+      "loss": 0.2885,
+      "step": 904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5054372975474316e-05,
+      "loss": 0.0013,
+      "step": 905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.504858861638131e-05,
+      "loss": 0.0022,
+      "step": 906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.50428042572883e-05,
+      "loss": 0.1286,
+      "step": 907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5037019898195284e-05,
+      "loss": 0.0012,
+      "step": 908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.503123553910227e-05,
+      "loss": 0.4157,
+      "step": 909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.502545118000926e-05,
+      "loss": 0.0008,
+      "step": 910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5019666820916244e-05,
+      "loss": 0.0013,
+      "step": 911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.501388246182323e-05,
+      "loss": 0.1339,
+      "step": 912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.500809810273022e-05,
+      "loss": 0.0013,
+      "step": 913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.500231374363721e-05,
+      "loss": 0.0021,
+      "step": 914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.49965293845442e-05,
+      "loss": 0.0024,
+      "step": 915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.499074502545118e-05,
+      "loss": 0.1731,
+      "step": 916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.498496066635817e-05,
+      "loss": 0.0017,
+      "step": 917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.497917630726516e-05,
+      "loss": 0.0013,
+      "step": 918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4973391948172145e-05,
+      "loss": 0.0029,
+      "step": 919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.496760758907913e-05,
+      "loss": 0.3884,
+      "step": 920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4961823229986125e-05,
+      "loss": 0.0013,
+      "step": 921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4956038870893105e-05,
+      "loss": 0.0111,
+      "step": 922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.495025451180009e-05,
+      "loss": 0.0054,
+      "step": 923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.494447015270708e-05,
+      "loss": 0.0965,
+      "step": 924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.493868579361407e-05,
+      "loss": 0.0032,
+      "step": 925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.493290143452106e-05,
+      "loss": 0.012,
+      "step": 926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4927117075428045e-05,
+      "loss": 0.1627,
+      "step": 927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.492133271633503e-05,
+      "loss": 0.0028,
+      "step": 928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.491554835724202e-05,
+      "loss": 0.1346,
+      "step": 929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4909763998149006e-05,
+      "loss": 0.0041,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.490397963905599e-05,
+      "loss": 0.0457,
+      "step": 931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4898195279962986e-05,
+      "loss": 0.1004,
+      "step": 932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.489241092086997e-05,
+      "loss": 0.001,
+      "step": 933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.488662656177696e-05,
+      "loss": 0.0032,
+      "step": 934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.488084220268394e-05,
+      "loss": 0.0565,
+      "step": 935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.487505784359093e-05,
+      "loss": 0.0146,
+      "step": 936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.486927348449792e-05,
+      "loss": 0.0623,
+      "step": 937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4863489125404907e-05,
+      "loss": 0.1734,
+      "step": 938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.485770476631189e-05,
+      "loss": 0.0015,
+      "step": 939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.485192040721889e-05,
+      "loss": 0.0695,
+      "step": 940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.484613604812587e-05,
+      "loss": 0.001,
+      "step": 941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4840351689032854e-05,
+      "loss": 0.0044,
+      "step": 942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.483456732993985e-05,
+      "loss": 0.0008,
+      "step": 943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4828782970846834e-05,
+      "loss": 0.3411,
+      "step": 944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.482299861175382e-05,
+      "loss": 0.0627,
+      "step": 945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.481721425266081e-05,
+      "loss": 0.1003,
+      "step": 946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4811429893567794e-05,
+      "loss": 0.0009,
+      "step": 947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.480564553447478e-05,
+      "loss": 0.5124,
+      "step": 948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.479986117538177e-05,
+      "loss": 0.0022,
+      "step": 949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4794076816288754e-05,
+      "loss": 0.0012,
+      "step": 950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.478829245719575e-05,
+      "loss": 0.0022,
+      "step": 951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4782508098102735e-05,
+      "loss": 0.1717,
+      "step": 952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.477672373900972e-05,
+      "loss": 0.0203,
+      "step": 953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.477093937991671e-05,
+      "loss": 0.0028,
+      "step": 954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4765155020823695e-05,
+      "loss": 0.0157,
+      "step": 955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.475937066173068e-05,
+      "loss": 0.2828,
+      "step": 956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.475358630263767e-05,
+      "loss": 0.0221,
+      "step": 957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.474780194354466e-05,
+      "loss": 0.0027,
+      "step": 958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.474201758445164e-05,
+      "loss": 0.0021,
+      "step": 959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.473623322535863e-05,
+      "loss": 0.0196,
+      "step": 960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4730448866265616e-05,
+      "loss": 0.0015,
+      "step": 961
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.472466450717261e-05,
+      "loss": 0.4416,
+      "step": 962
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4718880148079596e-05,
+      "loss": 0.0025,
+      "step": 963
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.471309578898658e-05,
+      "loss": 0.1199,
+      "step": 964
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.470731142989357e-05,
+      "loss": 0.0057,
+      "step": 965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4701527070800556e-05,
+      "loss": 0.0017,
+      "step": 966
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.469574271170754e-05,
+      "loss": 0.098,
+      "step": 967
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.468995835261453e-05,
+      "loss": 0.0464,
+      "step": 968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.468417399352152e-05,
+      "loss": 0.2454,
+      "step": 969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.467838963442851e-05,
+      "loss": 0.0222,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.46726052753355e-05,
+      "loss": 0.0419,
+      "step": 971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.466682091624248e-05,
+      "loss": 0.8205,
+      "step": 972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.466103655714947e-05,
+      "loss": 0.0007,
+      "step": 973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.465525219805646e-05,
+      "loss": 0.0007,
+      "step": 974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4649467838963444e-05,
+      "loss": 0.3238,
+      "step": 975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.464368347987044e-05,
+      "loss": 0.3468,
+      "step": 976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4637899120777424e-05,
+      "loss": 0.3731,
+      "step": 977
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4632114761684404e-05,
+      "loss": 0.227,
+      "step": 978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.462633040259139e-05,
+      "loss": 0.0012,
+      "step": 979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4620546043498384e-05,
+      "loss": 0.0031,
+      "step": 980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.461476168440537e-05,
+      "loss": 0.0021,
+      "step": 981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.460897732531236e-05,
+      "loss": 0.02,
+      "step": 982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4603192966219344e-05,
+      "loss": 0.0043,
+      "step": 983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.459740860712633e-05,
+      "loss": 0.0026,
+      "step": 984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.459162424803332e-05,
+      "loss": 0.0042,
+      "step": 985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4585839888940305e-05,
+      "loss": 0.1742,
+      "step": 986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.45800555298473e-05,
+      "loss": 0.0025,
+      "step": 987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4574271170754285e-05,
+      "loss": 0.0224,
+      "step": 988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.456848681166127e-05,
+      "loss": 0.0192,
+      "step": 989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.456270245256826e-05,
+      "loss": 0.2548,
+      "step": 990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4556918093475245e-05,
+      "loss": 0.0149,
+      "step": 991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.455113373438223e-05,
+      "loss": 0.1369,
+      "step": 992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.454534937528922e-05,
+      "loss": 0.0271,
+      "step": 993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4539565016196206e-05,
+      "loss": 0.0148,
+      "step": 994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.45337806571032e-05,
+      "loss": 0.0019,
+      "step": 995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4527996298010186e-05,
+      "loss": 0.0041,
+      "step": 996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4522211938917166e-05,
+      "loss": 0.0033,
+      "step": 997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.451642757982416e-05,
+      "loss": 0.0617,
+      "step": 998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4510643220731146e-05,
+      "loss": 0.0021,
+      "step": 999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.450485886163813e-05,
+      "loss": 0.0011,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.449907450254512e-05,
+      "loss": 0.0147,
+      "step": 1001
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.449329014345211e-05,
+      "loss": 0.0039,
+      "step": 1002
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.448750578435909e-05,
+      "loss": 0.0013,
+      "step": 1003
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.448172142526608e-05,
+      "loss": 0.0009,
+      "step": 1004
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.447593706617307e-05,
+      "loss": 0.5553,
+      "step": 1005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.447015270708006e-05,
+      "loss": 0.0011,
+      "step": 1006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.446436834798705e-05,
+      "loss": 0.0298,
+      "step": 1007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4458583988894034e-05,
+      "loss": 0.0076,
+      "step": 1008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.445279962980102e-05,
+      "loss": 0.2484,
+      "step": 1009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.444701527070801e-05,
+      "loss": 0.0027,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4441230911614994e-05,
+      "loss": 0.0426,
+      "step": 1011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.443544655252198e-05,
+      "loss": 0.0011,
+      "step": 1012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4429662193428974e-05,
+      "loss": 0.2094,
+      "step": 1013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.442387783433596e-05,
+      "loss": 0.0007,
+      "step": 1014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.441809347524295e-05,
+      "loss": 0.0008,
+      "step": 1015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.441230911614993e-05,
+      "loss": 0.1395,
+      "step": 1016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.440652475705692e-05,
+      "loss": 0.0011,
+      "step": 1017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.440074039796391e-05,
+      "loss": 0.334,
+      "step": 1018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4394956038870895e-05,
+      "loss": 0.0018,
+      "step": 1019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.438917167977788e-05,
+      "loss": 0.0009,
+      "step": 1020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.438338732068487e-05,
+      "loss": 0.4725,
+      "step": 1021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4377602961591855e-05,
+      "loss": 0.0062,
+      "step": 1022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.437181860249884e-05,
+      "loss": 0.0088,
+      "step": 1023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4366034243405835e-05,
+      "loss": 0.0396,
+      "step": 1024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.436024988431282e-05,
+      "loss": 0.3422,
+      "step": 1025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.435446552521981e-05,
+      "loss": 0.023,
+      "step": 1026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4348681166126796e-05,
+      "loss": 0.0054,
+      "step": 1027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.434289680703378e-05,
+      "loss": 0.0144,
+      "step": 1028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.433711244794077e-05,
+      "loss": 0.2013,
+      "step": 1029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4331328088847756e-05,
+      "loss": 0.0105,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.432554372975474e-05,
+      "loss": 0.3493,
+      "step": 1031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4319759370661736e-05,
+      "loss": 0.0031,
+      "step": 1032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.431397501156872e-05,
+      "loss": 0.0017,
+      "step": 1033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.43081906524757e-05,
+      "loss": 0.1046,
+      "step": 1034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4302406293382697e-05,
+      "loss": 0.012,
+      "step": 1035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.429662193428968e-05,
+      "loss": 0.0012,
+      "step": 1036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.429083757519667e-05,
+      "loss": 0.0033,
+      "step": 1037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.428505321610366e-05,
+      "loss": 0.0055,
+      "step": 1038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.427926885701065e-05,
+      "loss": 0.3089,
+      "step": 1039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.427348449791763e-05,
+      "loss": 0.0034,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.426770013882462e-05,
+      "loss": 0.3945,
+      "step": 1041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4261915779731604e-05,
+      "loss": 0.0045,
+      "step": 1042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.42561314206386e-05,
+      "loss": 0.0012,
+      "step": 1043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4250347061545584e-05,
+      "loss": 0.0519,
+      "step": 1044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.424456270245257e-05,
+      "loss": 0.0041,
+      "step": 1045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.423877834335956e-05,
+      "loss": 0.0035,
+      "step": 1046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4232993984266544e-05,
+      "loss": 0.0009,
+      "step": 1047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.422720962517353e-05,
+      "loss": 0.0016,
+      "step": 1048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.422142526608052e-05,
+      "loss": 0.0121,
+      "step": 1049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.421564090698751e-05,
+      "loss": 0.0024,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.42098565478945e-05,
+      "loss": 0.0024,
+      "step": 1051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4204072188801485e-05,
+      "loss": 0.0041,
+      "step": 1052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4198287829708465e-05,
+      "loss": 0.0008,
+      "step": 1053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.419250347061546e-05,
+      "loss": 0.0032,
+      "step": 1054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4186719111522445e-05,
+      "loss": 0.0008,
+      "step": 1055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.418093475242943e-05,
+      "loss": 0.0014,
+      "step": 1056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4175150393336425e-05,
+      "loss": 0.0013,
+      "step": 1057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.416936603424341e-05,
+      "loss": 0.0015,
+      "step": 1058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.416358167515039e-05,
+      "loss": 0.0008,
+      "step": 1059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.415779731605738e-05,
+      "loss": 0.0007,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.415201295696437e-05,
+      "loss": 0.0036,
+      "step": 1061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.414622859787136e-05,
+      "loss": 0.0008,
+      "step": 1062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4140444238778346e-05,
+      "loss": 0.0062,
+      "step": 1063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.413465987968533e-05,
+      "loss": 0.0035,
+      "step": 1064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.412887552059232e-05,
+      "loss": 0.4048,
+      "step": 1065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4123091161499306e-05,
+      "loss": 0.0011,
+      "step": 1066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.411730680240629e-05,
+      "loss": 0.3005,
+      "step": 1067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4111522443313287e-05,
+      "loss": 0.0028,
+      "step": 1068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.410573808422027e-05,
+      "loss": 0.144,
+      "step": 1069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.409995372512726e-05,
+      "loss": 0.0011,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.409416936603425e-05,
+      "loss": 0.2312,
+      "step": 1071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4088385006941234e-05,
+      "loss": 0.0269,
+      "step": 1072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.408260064784822e-05,
+      "loss": 0.1119,
+      "step": 1073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.407681628875521e-05,
+      "loss": 0.001,
+      "step": 1074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4071031929662194e-05,
+      "loss": 0.0013,
+      "step": 1075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.406524757056919e-05,
+      "loss": 0.0009,
+      "step": 1076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.405946321147617e-05,
+      "loss": 0.0436,
+      "step": 1077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4053678852383154e-05,
+      "loss": 0.0014,
+      "step": 1078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.404789449329015e-05,
+      "loss": 0.0025,
+      "step": 1079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4042110134197134e-05,
+      "loss": 0.0012,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.403632577510412e-05,
+      "loss": 0.115,
+      "step": 1081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.403054141601111e-05,
+      "loss": 0.7949,
+      "step": 1082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4024757056918095e-05,
+      "loss": 0.0014,
+      "step": 1083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.401897269782508e-05,
+      "loss": 0.0009,
+      "step": 1084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.401318833873207e-05,
+      "loss": 0.001,
+      "step": 1085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4007403979639055e-05,
+      "loss": 0.0122,
+      "step": 1086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.400161962054605e-05,
+      "loss": 0.0012,
+      "step": 1087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3995835261453035e-05,
+      "loss": 0.0015,
+      "step": 1088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.399005090236002e-05,
+      "loss": 0.0033,
+      "step": 1089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3984266543267e-05,
+      "loss": 0.0008,
+      "step": 1090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3978482184173996e-05,
+      "loss": 0.001,
+      "step": 1091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.397269782508098e-05,
+      "loss": 0.0018,
+      "step": 1092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.396691346598797e-05,
+      "loss": 0.0006,
+      "step": 1093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.396112910689496e-05,
+      "loss": 0.0301,
+      "step": 1094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.395534474780195e-05,
+      "loss": 0.0008,
+      "step": 1095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.394956038870893e-05,
+      "loss": 0.0016,
+      "step": 1096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3943776029615916e-05,
+      "loss": 0.0014,
+      "step": 1097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.393799167052291e-05,
+      "loss": 0.0013,
+      "step": 1098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3932207311429896e-05,
+      "loss": 0.0006,
+      "step": 1099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.392642295233688e-05,
+      "loss": 0.001,
+      "step": 1100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.392063859324387e-05,
+      "loss": 0.1776,
+      "step": 1101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.391485423415086e-05,
+      "loss": 0.0005,
+      "step": 1102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3909069875057843e-05,
+      "loss": 0.0012,
+      "step": 1103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.390328551596483e-05,
+      "loss": 0.4632,
+      "step": 1104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3897501156871824e-05,
+      "loss": 0.0127,
+      "step": 1105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.389171679777881e-05,
+      "loss": 0.0066,
+      "step": 1106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.38859324386858e-05,
+      "loss": 0.0006,
+      "step": 1107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3880148079592784e-05,
+      "loss": 0.0039,
+      "step": 1108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.387436372049977e-05,
+      "loss": 0.0011,
+      "step": 1109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.386857936140676e-05,
+      "loss": 0.0019,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3862795002313744e-05,
+      "loss": 0.4432,
+      "step": 1111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.385701064322073e-05,
+      "loss": 0.0008,
+      "step": 1112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3851226284127725e-05,
+      "loss": 0.0024,
+      "step": 1113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.384544192503471e-05,
+      "loss": 0.0018,
+      "step": 1114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.383965756594169e-05,
+      "loss": 0.0031,
+      "step": 1115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3833873206848685e-05,
+      "loss": 0.0025,
+      "step": 1116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.382808884775567e-05,
+      "loss": 0.007,
+      "step": 1117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.382230448866266e-05,
+      "loss": 0.0022,
+      "step": 1118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3816520129569645e-05,
+      "loss": 0.0022,
+      "step": 1119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.381073577047664e-05,
+      "loss": 0.0106,
+      "step": 1120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.380495141138362e-05,
+      "loss": 0.0034,
+      "step": 1121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3799167052290605e-05,
+      "loss": 0.0027,
+      "step": 1122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.379338269319759e-05,
+      "loss": 0.0299,
+      "step": 1123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3787598334104586e-05,
+      "loss": 0.0017,
+      "step": 1124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.378181397501157e-05,
+      "loss": 0.0018,
+      "step": 1125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.377602961591856e-05,
+      "loss": 0.0012,
+      "step": 1126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3770245256825546e-05,
+      "loss": 0.2657,
+      "step": 1127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.376446089773253e-05,
+      "loss": 0.0016,
+      "step": 1128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.375867653863952e-05,
+      "loss": 0.0028,
+      "step": 1129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3752892179546506e-05,
+      "loss": 0.0027,
+      "step": 1130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.37471078204535e-05,
+      "loss": 0.1077,
+      "step": 1131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3741323461360486e-05,
+      "loss": 0.0038,
+      "step": 1132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.373553910226747e-05,
+      "loss": 0.0022,
+      "step": 1133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.372975474317445e-05,
+      "loss": 0.1615,
+      "step": 1134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.372397038408145e-05,
+      "loss": 0.0034,
+      "step": 1135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3718186024988433e-05,
+      "loss": 0.0015,
+      "step": 1136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.371240166589542e-05,
+      "loss": 0.0111,
+      "step": 1137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3706617306802414e-05,
+      "loss": 0.0016,
+      "step": 1138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3700832947709394e-05,
+      "loss": 0.0012,
+      "step": 1139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.369504858861638e-05,
+      "loss": 0.0045,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.368926422952337e-05,
+      "loss": 0.0014,
+      "step": 1141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.368347987043036e-05,
+      "loss": 0.0021,
+      "step": 1142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.367769551133735e-05,
+      "loss": 0.0011,
+      "step": 1143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3671911152244334e-05,
+      "loss": 0.2442,
+      "step": 1144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.366612679315132e-05,
+      "loss": 0.0012,
+      "step": 1145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.366034243405831e-05,
+      "loss": 0.0021,
+      "step": 1146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3654558074965295e-05,
+      "loss": 0.0008,
+      "step": 1147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.364877371587228e-05,
+      "loss": 0.0013,
+      "step": 1148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3642989356779275e-05,
+      "loss": 0.0039,
+      "step": 1149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.363720499768626e-05,
+      "loss": 0.0016,
+      "step": 1150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.363142063859325e-05,
+      "loss": 0.0009,
+      "step": 1151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.362563627950023e-05,
+      "loss": 0.0007,
+      "step": 1152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.361985192040722e-05,
+      "loss": 0.0132,
+      "step": 1153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.361406756131421e-05,
+      "loss": 0.0008,
+      "step": 1154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3608283202221195e-05,
+      "loss": 0.0165,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.360249884312818e-05,
+      "loss": 0.0007,
+      "step": 1156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3596714484035176e-05,
+      "loss": 0.0007,
+      "step": 1157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3590930124942156e-05,
+      "loss": 0.0006,
+      "step": 1158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.358514576584914e-05,
+      "loss": 0.004,
+      "step": 1159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3579361406756136e-05,
+      "loss": 0.0004,
+      "step": 1160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.357357704766312e-05,
+      "loss": 0.3487,
+      "step": 1161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.356779268857011e-05,
+      "loss": 0.0005,
+      "step": 1162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3562008329477096e-05,
+      "loss": 0.2542,
+      "step": 1163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.355622397038408e-05,
+      "loss": 0.0009,
+      "step": 1164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.355043961129107e-05,
+      "loss": 0.0005,
+      "step": 1165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3544655252198057e-05,
+      "loss": 0.0009,
+      "step": 1166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.353887089310504e-05,
+      "loss": 0.0005,
+      "step": 1167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.353308653401204e-05,
+      "loss": 0.0005,
+      "step": 1168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3527302174919024e-05,
+      "loss": 0.0014,
+      "step": 1169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.352151781582601e-05,
+      "loss": 0.3557,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3515733456733e-05,
+      "loss": 0.2741,
+      "step": 1171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3509949097639984e-05,
+      "loss": 0.0005,
+      "step": 1172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.350416473854697e-05,
+      "loss": 0.2104,
+      "step": 1173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.349838037945396e-05,
+      "loss": 0.0004,
+      "step": 1174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.349259602036095e-05,
+      "loss": 0.0008,
+      "step": 1175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.348681166126794e-05,
+      "loss": 0.0013,
+      "step": 1176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.348102730217492e-05,
+      "loss": 0.0009,
+      "step": 1177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3475242943081904e-05,
+      "loss": 0.0017,
+      "step": 1178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.34694585839889e-05,
+      "loss": 0.0009,
+      "step": 1179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3463674224895885e-05,
+      "loss": 0.322,
+      "step": 1180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.345788986580287e-05,
+      "loss": 0.1479,
+      "step": 1181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.345210550670986e-05,
+      "loss": 0.0014,
+      "step": 1182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3446321147616845e-05,
+      "loss": 0.0041,
+      "step": 1183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.344053678852383e-05,
+      "loss": 0.0014,
+      "step": 1184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.343475242943082e-05,
+      "loss": 0.0028,
+      "step": 1185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.342896807033781e-05,
+      "loss": 0.0094,
+      "step": 1186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.34231837112448e-05,
+      "loss": 0.0023,
+      "step": 1187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3417399352151786e-05,
+      "loss": 0.0026,
+      "step": 1188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.341161499305877e-05,
+      "loss": 0.0109,
+      "step": 1189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.340583063396576e-05,
+      "loss": 0.0019,
+      "step": 1190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3400046274872746e-05,
+      "loss": 0.0023,
+      "step": 1191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.339426191577973e-05,
+      "loss": 0.3079,
+      "step": 1192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.338847755668672e-05,
+      "loss": 0.0017,
+      "step": 1193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.338269319759371e-05,
+      "loss": 0.0012,
+      "step": 1194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.337690883850069e-05,
+      "loss": 0.001,
+      "step": 1195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.337112447940768e-05,
+      "loss": 0.1129,
+      "step": 1196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.336534012031467e-05,
+      "loss": 0.001,
+      "step": 1197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.335955576122166e-05,
+      "loss": 0.0013,
+      "step": 1198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3353771402128647e-05,
+      "loss": 0.0035,
+      "step": 1199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.334798704303563e-05,
+      "loss": 0.0008,
+      "step": 1200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.334220268394262e-05,
+      "loss": 0.0032,
+      "step": 1201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.333641832484961e-05,
+      "loss": 0.004,
+      "step": 1202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3330633965756594e-05,
+      "loss": 0.0009,
+      "step": 1203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.332484960666358e-05,
+      "loss": 0.4035,
+      "step": 1204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3319065247570574e-05,
+      "loss": 0.0093,
+      "step": 1205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.331328088847756e-05,
+      "loss": 0.0006,
+      "step": 1206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.330749652938455e-05,
+      "loss": 0.0007,
+      "step": 1207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3301712170291534e-05,
+      "loss": 0.0004,
+      "step": 1208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.329592781119852e-05,
+      "loss": 0.0023,
+      "step": 1209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.329014345210551e-05,
+      "loss": 0.001,
+      "step": 1210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3284359093012494e-05,
+      "loss": 0.1365,
+      "step": 1211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.327857473391949e-05,
+      "loss": 0.012,
+      "step": 1212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3272790374826475e-05,
+      "loss": 0.0013,
+      "step": 1213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3267006015733455e-05,
+      "loss": 0.0042,
+      "step": 1214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.326122165664044e-05,
+      "loss": 0.0009,
+      "step": 1215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3255437297547435e-05,
+      "loss": 0.0004,
+      "step": 1216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.324965293845442e-05,
+      "loss": 0.0006,
+      "step": 1217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.324386857936141e-05,
+      "loss": 0.2502,
+      "step": 1218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.32380842202684e-05,
+      "loss": 0.0007,
+      "step": 1219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.323229986117538e-05,
+      "loss": 0.0018,
+      "step": 1220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.322651550208237e-05,
+      "loss": 0.0005,
+      "step": 1221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3220731142989356e-05,
+      "loss": 0.0007,
+      "step": 1222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.321494678389635e-05,
+      "loss": 0.0202,
+      "step": 1223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3209162424803336e-05,
+      "loss": 0.0007,
+      "step": 1224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.320337806571032e-05,
+      "loss": 0.0008,
+      "step": 1225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.319759370661731e-05,
+      "loss": 0.0004,
+      "step": 1226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3191809347524296e-05,
+      "loss": 0.0008,
+      "step": 1227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.318602498843128e-05,
+      "loss": 0.0085,
+      "step": 1228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.318024062933827e-05,
+      "loss": 0.0332,
+      "step": 1229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.317445627024526e-05,
+      "loss": 0.2743,
+      "step": 1230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.316867191115225e-05,
+      "loss": 0.0007,
+      "step": 1231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.316288755205924e-05,
+      "loss": 0.0006,
+      "step": 1232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.315710319296622e-05,
+      "loss": 0.4141,
+      "step": 1233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.315131883387321e-05,
+      "loss": 0.0155,
+      "step": 1234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.31455344747802e-05,
+      "loss": 0.0005,
+      "step": 1235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.3139750115687184e-05,
+      "loss": 0.0071,
+      "step": 1236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.313396575659417e-05,
+      "loss": 0.0005,
+      "step": 1237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.312818139750116e-05,
+      "loss": 0.0006,
+      "step": 1238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.3122397038408144e-05,
+      "loss": 0.3014,
+      "step": 1239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.311661267931513e-05,
+      "loss": 0.0007,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.3110828320222124e-05,
+      "loss": 0.0007,
+      "step": 1241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.310504396112911e-05,
+      "loss": 0.4061,
+      "step": 1242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.30992596020361e-05,
+      "loss": 0.0014,
+      "step": 1243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.3093475242943085e-05,
+      "loss": 0.0006,
+      "step": 1244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.308769088385007e-05,
+      "loss": 0.0074,
+      "step": 1245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.308190652475706e-05,
+      "loss": 0.0504,
+      "step": 1246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.3076122165664045e-05,
+      "loss": 0.0033,
+      "step": 1247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.307033780657103e-05,
+      "loss": 0.0029,
+      "step": 1248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.3064553447478025e-05,
+      "loss": 0.0009,
+      "step": 1249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.305876908838501e-05,
+      "loss": 0.0008,
+      "step": 1250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.305298472929199e-05,
+      "loss": 0.0016,
+      "step": 1251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.3047200370198985e-05,
+      "loss": 0.0589,
+      "step": 1252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.304141601110597e-05,
+      "loss": 0.0011,
+      "step": 1253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.303563165201296e-05,
+      "loss": 0.0012,
+      "step": 1254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.3029847292919946e-05,
+      "loss": 0.1055,
+      "step": 1255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.302406293382694e-05,
+      "loss": 0.4162,
+      "step": 1256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.301827857473392e-05,
+      "loss": 0.0006,
+      "step": 1257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.3012494215640906e-05,
+      "loss": 0.2895,
+      "step": 1258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.300670985654789e-05,
+      "loss": 0.4283,
+      "step": 1259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.3000925497454886e-05,
+      "loss": 0.0008,
+      "step": 1260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.299514113836187e-05,
+      "loss": 0.2384,
+      "step": 1261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.298935677926886e-05,
+      "loss": 0.0015,
+      "step": 1262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.2983572420175846e-05,
+      "loss": 0.0011,
+      "step": 1263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.297778806108283e-05,
+      "loss": 0.0015,
+      "step": 1264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.297200370198982e-05,
+      "loss": 0.0117,
+      "step": 1265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.296621934289681e-05,
+      "loss": 0.3283,
+      "step": 1266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.29604349838038e-05,
+      "loss": 0.0019,
+      "step": 1267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.295465062471079e-05,
+      "loss": 0.0069,
+      "step": 1268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.2948866265617774e-05,
+      "loss": 0.0057,
+      "step": 1269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.2943081906524754e-05,
+      "loss": 0.0093,
+      "step": 1270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.293729754743175e-05,
+      "loss": 0.0024,
+      "step": 1271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.2931513188338734e-05,
+      "loss": 0.0029,
+      "step": 1272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.292572882924572e-05,
+      "loss": 0.0047,
+      "step": 1273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.291994447015271e-05,
+      "loss": 0.0033,
+      "step": 1274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.29141601110597e-05,
+      "loss": 0.0013,
+      "step": 1275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.290837575196668e-05,
+      "loss": 0.0045,
+      "step": 1276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.290259139287367e-05,
+      "loss": 0.0011,
+      "step": 1277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.289680703378066e-05,
+      "loss": 0.0016,
+      "step": 1278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.289102267468765e-05,
+      "loss": 0.0011,
+      "step": 1279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.2885238315594635e-05,
+      "loss": 0.2917,
+      "step": 1280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.287945395650162e-05,
+      "loss": 0.001,
+      "step": 1281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.287366959740861e-05,
+      "loss": 0.6141,
+      "step": 1282
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2867885238315595e-05,
+      "loss": 0.0022,
+      "step": 1283
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.286210087922258e-05,
+      "loss": 0.0018,
+      "step": 1284
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.285631652012957e-05,
+      "loss": 0.0165,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.285053216103656e-05,
+      "loss": 0.0039,
+      "step": 1286
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.284474780194355e-05,
+      "loss": 0.1625,
+      "step": 1287
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2838963442850536e-05,
+      "loss": 0.0016,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.283317908375752e-05,
+      "loss": 0.0086,
+      "step": 1289
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.282739472466451e-05,
+      "loss": 0.0012,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2821610365571496e-05,
+      "loss": 0.0024,
+      "step": 1291
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.281582600647848e-05,
+      "loss": 0.0084,
+      "step": 1292
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2810041647385476e-05,
+      "loss": 0.0013,
+      "step": 1293
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.280425728829246e-05,
+      "loss": 0.0009,
+      "step": 1294
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.279847292919944e-05,
+      "loss": 0.002,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.279268857010643e-05,
+      "loss": 0.0037,
+      "step": 1296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.278690421101342e-05,
+      "loss": 0.0009,
+      "step": 1297
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.278111985192041e-05,
+      "loss": 0.5685,
+      "step": 1298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.27753354928274e-05,
+      "loss": 0.0006,
+      "step": 1299
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2769551133734384e-05,
+      "loss": 0.3527,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.276376677464137e-05,
+      "loss": 0.0358,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.275798241554836e-05,
+      "loss": 0.0009,
+      "step": 1302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2752198056455344e-05,
+      "loss": 0.001,
+      "step": 1303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.274641369736234e-05,
+      "loss": 0.0008,
+      "step": 1304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2740629338269324e-05,
+      "loss": 0.0038,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.273484497917631e-05,
+      "loss": 0.0017,
+      "step": 1306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.27290606200833e-05,
+      "loss": 0.0035,
+      "step": 1307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2723276260990284e-05,
+      "loss": 0.001,
+      "step": 1308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.271749190189727e-05,
+      "loss": 0.0009,
+      "step": 1309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.271170754280426e-05,
+      "loss": 0.001,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.270592318371125e-05,
+      "loss": 0.0012,
+      "step": 1311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.270013882461824e-05,
+      "loss": 0.0205,
+      "step": 1312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.269435446552522e-05,
+      "loss": 0.1617,
+      "step": 1313
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2688570106432205e-05,
+      "loss": 0.003,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.26827857473392e-05,
+      "loss": 0.0041,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2677001388246185e-05,
+      "loss": 0.0098,
+      "step": 1316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.267121702915317e-05,
+      "loss": 0.0007,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.266543267006016e-05,
+      "loss": 0.0014,
+      "step": 1318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2659648310967146e-05,
+      "loss": 0.0382,
+      "step": 1319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.265386395187413e-05,
+      "loss": 0.0013,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.264807959278112e-05,
+      "loss": 0.3076,
+      "step": 1321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.264229523368811e-05,
+      "loss": 0.1285,
+      "step": 1322
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.26365108745951e-05,
+      "loss": 0.032,
+      "step": 1323
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.2630726515502086e-05,
+      "loss": 0.0018,
+      "step": 1324
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.262494215640907e-05,
+      "loss": 0.0082,
+      "step": 1325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.261915779731606e-05,
+      "loss": 0.0008,
+      "step": 1326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2613373438223046e-05,
+      "loss": 0.0015,
+      "step": 1327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.260758907913003e-05,
+      "loss": 0.001,
+      "step": 1328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.260180472003702e-05,
+      "loss": 0.0007,
+      "step": 1329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2596020360944013e-05,
+      "loss": 0.0006,
+      "step": 1330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2590236001851e-05,
+      "loss": 0.1193,
+      "step": 1331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.258445164275798e-05,
+      "loss": 0.3393,
+      "step": 1332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2578667283664974e-05,
+      "loss": 0.003,
+      "step": 1333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.257288292457196e-05,
+      "loss": 0.0057,
+      "step": 1334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.256709856547895e-05,
+      "loss": 0.0042,
+      "step": 1335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2561314206385934e-05,
+      "loss": 0.0007,
+      "step": 1336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.255552984729293e-05,
+      "loss": 0.0012,
+      "step": 1337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.254974548819991e-05,
+      "loss": 0.3864,
+      "step": 1338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2543961129106894e-05,
+      "loss": 0.0156,
+      "step": 1339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.253817677001388e-05,
+      "loss": 0.0061,
+      "step": 1340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2532392410920875e-05,
+      "loss": 0.0007,
+      "step": 1341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.252660805182786e-05,
+      "loss": 0.2082,
+      "step": 1342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.252082369273485e-05,
+      "loss": 0.3213,
+      "step": 1343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2515039333641835e-05,
+      "loss": 0.0021,
+      "step": 1344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.250925497454882e-05,
+      "loss": 0.0212,
+      "step": 1345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.250347061545581e-05,
+      "loss": 0.0046,
+      "step": 1346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2497686256362795e-05,
+      "loss": 0.0044,
+      "step": 1347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.249190189726979e-05,
+      "loss": 0.0147,
+      "step": 1348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2486117538176775e-05,
+      "loss": 0.0347,
+      "step": 1349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.248033317908376e-05,
+      "loss": 0.0015,
+      "step": 1350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.247454881999074e-05,
+      "loss": 0.0032,
+      "step": 1351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2468764460897736e-05,
+      "loss": 0.0081,
+      "step": 1352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.246298010180472e-05,
+      "loss": 0.0018,
+      "step": 1353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.245719574271171e-05,
+      "loss": 0.0185,
+      "step": 1354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2451411383618696e-05,
+      "loss": 0.2711,
+      "step": 1355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.244562702452568e-05,
+      "loss": 0.0035,
+      "step": 1356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.243984266543267e-05,
+      "loss": 0.0063,
+      "step": 1357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2434058306339656e-05,
+      "loss": 0.0025,
+      "step": 1358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.242827394724665e-05,
+      "loss": 0.0014,
+      "step": 1359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2422489588153636e-05,
+      "loss": 0.0021,
+      "step": 1360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.241670522906062e-05,
+      "loss": 0.5234,
+      "step": 1361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.241092086996761e-05,
+      "loss": 0.1541,
+      "step": 1362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.24051365108746e-05,
+      "loss": 0.2366,
+      "step": 1363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.2399352151781583e-05,
+      "loss": 0.0007,
+      "step": 1364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.239356779268857e-05,
+      "loss": 0.3845,
+      "step": 1365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.238778343359556e-05,
+      "loss": 0.0013,
+      "step": 1366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.238199907450255e-05,
+      "loss": 0.0029,
+      "step": 1367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.237621471540954e-05,
+      "loss": 0.0033,
+      "step": 1368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.237043035631652e-05,
+      "loss": 0.0023,
+      "step": 1369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.236464599722351e-05,
+      "loss": 0.0407,
+      "step": 1370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.23588616381305e-05,
+      "loss": 0.2839,
+      "step": 1371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2353077279037484e-05,
+      "loss": 0.0038,
+      "step": 1372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.234729291994447e-05,
+      "loss": 0.0059,
+      "step": 1373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2341508560851465e-05,
+      "loss": 0.0038,
+      "step": 1374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2335724201758445e-05,
+      "loss": 0.239,
+      "step": 1375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.232993984266543e-05,
+      "loss": 0.0047,
+      "step": 1376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.232415548357242e-05,
+      "loss": 0.0153,
+      "step": 1377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.231837112447941e-05,
+      "loss": 0.0113,
+      "step": 1378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.23125867653864e-05,
+      "loss": 0.0084,
+      "step": 1379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2306802406293385e-05,
+      "loss": 0.0114,
+      "step": 1380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.230101804720037e-05,
+      "loss": 0.0071,
+      "step": 1381
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.229523368810736e-05,
+      "loss": 0.2228,
+      "step": 1382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2289449329014345e-05,
+      "loss": 0.011,
+      "step": 1383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.228366496992133e-05,
+      "loss": 0.0076,
+      "step": 1384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2277880610828326e-05,
+      "loss": 0.008,
+      "step": 1385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.227209625173531e-05,
+      "loss": 0.1148,
+      "step": 1386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.22663118926423e-05,
+      "loss": 0.0061,
+      "step": 1387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.226052753354928e-05,
+      "loss": 0.0028,
+      "step": 1388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.225474317445627e-05,
+      "loss": 0.0045,
+      "step": 1389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.224895881536326e-05,
+      "loss": 0.0056,
+      "step": 1390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2243174456270246e-05,
+      "loss": 0.0019,
+      "step": 1391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.223739009717724e-05,
+      "loss": 0.0585,
+      "step": 1392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2231605738084227e-05,
+      "loss": 0.0301,
+      "step": 1393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2225821378991206e-05,
+      "loss": 0.0015,
+      "step": 1394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.222003701989819e-05,
+      "loss": 0.0074,
+      "step": 1395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.221425266080519e-05,
+      "loss": 0.0045,
+      "step": 1396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2208468301712174e-05,
+      "loss": 0.2819,
+      "step": 1397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.220268394261916e-05,
+      "loss": 0.0045,
+      "step": 1398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.219689958352615e-05,
+      "loss": 0.0007,
+      "step": 1399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2191115224433134e-05,
+      "loss": 0.001,
+      "step": 1400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.218533086534012e-05,
+      "loss": 0.0016,
+      "step": 1401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.217954650624711e-05,
+      "loss": 0.0006,
+      "step": 1402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.21737621471541e-05,
+      "loss": 0.0048,
+      "step": 1403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.216797778806109e-05,
+      "loss": 0.0008,
+      "step": 1404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2162193428968074e-05,
+      "loss": 0.0009,
+      "step": 1405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.215640906987506e-05,
+      "loss": 0.3866,
+      "step": 1406
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.215062471078205e-05,
+      "loss": 0.0011,
+      "step": 1407
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2144840351689035e-05,
+      "loss": 0.0007,
+      "step": 1408
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.213905599259602e-05,
+      "loss": 0.0011,
+      "step": 1409
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.213327163350301e-05,
+      "loss": 0.0138,
+      "step": 1410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.212748727441e-05,
+      "loss": 0.0006,
+      "step": 1411
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.212170291531699e-05,
+      "loss": 0.0004,
+      "step": 1412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.211591855622397e-05,
+      "loss": 0.2793,
+      "step": 1413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.211013419713096e-05,
+      "loss": 0.0009,
+      "step": 1414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.210434983803795e-05,
+      "loss": 0.0005,
+      "step": 1415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2098565478944935e-05,
+      "loss": 0.0005,
+      "step": 1416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.209278111985192e-05,
+      "loss": 0.0019,
+      "step": 1417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.208699676075891e-05,
+      "loss": 0.0007,
+      "step": 1418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2081212401665896e-05,
+      "loss": 0.0013,
+      "step": 1419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.207542804257288e-05,
+      "loss": 0.0006,
+      "step": 1420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.206964368347987e-05,
+      "loss": 0.0007,
+      "step": 1421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.206385932438686e-05,
+      "loss": 0.0016,
+      "step": 1422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.205807496529385e-05,
+      "loss": 0.0006,
+      "step": 1423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2052290606200836e-05,
+      "loss": 0.0009,
+      "step": 1424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.204650624710782e-05,
+      "loss": 0.0014,
+      "step": 1425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.204072188801481e-05,
+      "loss": 0.0025,
+      "step": 1426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2034937528921797e-05,
+      "loss": 0.0014,
+      "step": 1427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.202915316982878e-05,
+      "loss": 0.0005,
+      "step": 1428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.202336881073578e-05,
+      "loss": 0.8096,
+      "step": 1429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2017584451642764e-05,
+      "loss": 0.1959,
+      "step": 1430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2011800092549744e-05,
+      "loss": 0.1931,
+      "step": 1431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.200601573345673e-05,
+      "loss": 0.0016,
+      "step": 1432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2000231374363724e-05,
+      "loss": 0.0008,
+      "step": 1433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.199444701527071e-05,
+      "loss": 0.001,
+      "step": 1434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.19886626561777e-05,
+      "loss": 0.0015,
+      "step": 1435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1982878297084684e-05,
+      "loss": 0.0398,
+      "step": 1436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.197709393799167e-05,
+      "loss": 0.0008,
+      "step": 1437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.197130957889866e-05,
+      "loss": 0.0042,
+      "step": 1438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1965525219805644e-05,
+      "loss": 0.0026,
+      "step": 1439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.195974086071264e-05,
+      "loss": 0.0009,
+      "step": 1440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1953956501619625e-05,
+      "loss": 0.0034,
+      "step": 1441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.194817214252661e-05,
+      "loss": 0.0017,
+      "step": 1442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.19423877834336e-05,
+      "loss": 0.001,
+      "step": 1443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1936603424340585e-05,
+      "loss": 0.0036,
+      "step": 1444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.193081906524757e-05,
+      "loss": 0.0093,
+      "step": 1445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.192503470615456e-05,
+      "loss": 0.0013,
+      "step": 1446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1919250347061545e-05,
+      "loss": 0.0033,
+      "step": 1447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.191346598796854e-05,
+      "loss": 0.0006,
+      "step": 1448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1907681628875526e-05,
+      "loss": 0.019,
+      "step": 1449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1901897269782506e-05,
+      "loss": 0.0059,
+      "step": 1450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.18961129106895e-05,
+      "loss": 0.0034,
+      "step": 1451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1890328551596486e-05,
+      "loss": 0.0011,
+      "step": 1452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.188454419250347e-05,
+      "loss": 0.0441,
+      "step": 1453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.187875983341046e-05,
+      "loss": 0.0003,
+      "step": 1454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.187297547431745e-05,
+      "loss": 0.0112,
+      "step": 1455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.186719111522443e-05,
+      "loss": 0.0005,
+      "step": 1456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.186140675613142e-05,
+      "loss": 0.0007,
+      "step": 1457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1855622397038406e-05,
+      "loss": 0.2763,
+      "step": 1458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.18498380379454e-05,
+      "loss": 0.0006,
+      "step": 1459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.184405367885239e-05,
+      "loss": 0.0013,
+      "step": 1460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1838269319759373e-05,
+      "loss": 0.0003,
+      "step": 1461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.183248496066636e-05,
+      "loss": 0.0099,
+      "step": 1462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.182670060157335e-05,
+      "loss": 0.0003,
+      "step": 1463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1820916242480334e-05,
+      "loss": 0.0003,
+      "step": 1464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.181513188338732e-05,
+      "loss": 0.0003,
+      "step": 1465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1809347524294314e-05,
+      "loss": 0.2026,
+      "step": 1466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.18035631652013e-05,
+      "loss": 0.6226,
+      "step": 1467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.179777880610829e-05,
+      "loss": 0.0005,
+      "step": 1468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.179199444701527e-05,
+      "loss": 0.0043,
+      "step": 1469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.178621008792226e-05,
+      "loss": 0.0007,
+      "step": 1470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.178042572882925e-05,
+      "loss": 0.0185,
+      "step": 1471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1774641369736235e-05,
+      "loss": 0.55,
+      "step": 1472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.176885701064323e-05,
+      "loss": 0.0003,
+      "step": 1473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.176307265155021e-05,
+      "loss": 0.0005,
+      "step": 1474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1757288292457195e-05,
+      "loss": 0.0608,
+      "step": 1475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.175150393336418e-05,
+      "loss": 0.8974,
+      "step": 1476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1745719574271175e-05,
+      "loss": 0.0026,
+      "step": 1477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.173993521517816e-05,
+      "loss": 0.015,
+      "step": 1478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.173415085608515e-05,
+      "loss": 0.0137,
+      "step": 1479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1728366496992135e-05,
+      "loss": 0.0017,
+      "step": 1480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.172258213789912e-05,
+      "loss": 0.001,
+      "step": 1481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.171679777880611e-05,
+      "loss": 0.2674,
+      "step": 1482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1711013419713096e-05,
+      "loss": 0.0011,
+      "step": 1483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.170522906062009e-05,
+      "loss": 0.0009,
+      "step": 1484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1699444701527076e-05,
+      "loss": 0.0166,
+      "step": 1485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.169366034243406e-05,
+      "loss": 0.0636,
+      "step": 1486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.168787598334104e-05,
+      "loss": 0.0026,
+      "step": 1487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1682091624248036e-05,
+      "loss": 0.001,
+      "step": 1488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.167630726515502e-05,
+      "loss": 0.0016,
+      "step": 1489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.167052290606201e-05,
+      "loss": 0.291,
+      "step": 1490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1664738546968996e-05,
+      "loss": 0.6097,
+      "step": 1491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.165895418787599e-05,
+      "loss": 0.001,
+      "step": 1492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.165316982878297e-05,
+      "loss": 0.0681,
+      "step": 1493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.164738546968996e-05,
+      "loss": 0.7959,
+      "step": 1494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.164160111059695e-05,
+      "loss": 0.2998,
+      "step": 1495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.163581675150394e-05,
+      "loss": 0.0013,
+      "step": 1496
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1630032392410924e-05,
+      "loss": 0.0042,
+      "step": 1497
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.162424803331791e-05,
+      "loss": 0.0021,
+      "step": 1498
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.16184636742249e-05,
+      "loss": 0.0049,
+      "step": 1499
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1612679315131884e-05,
+      "loss": 0.0032,
+      "step": 1500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.160689495603887e-05,
+      "loss": 0.0028,
+      "step": 1501
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.160111059694586e-05,
+      "loss": 0.0096,
+      "step": 1502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.159532623785285e-05,
+      "loss": 0.1875,
+      "step": 1503
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.158954187875984e-05,
+      "loss": 0.1836,
+      "step": 1504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1583757519666825e-05,
+      "loss": 0.0024,
+      "step": 1505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1577973160573805e-05,
+      "loss": 0.0028,
+      "step": 1506
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.15721888014808e-05,
+      "loss": 0.0207,
+      "step": 1507
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1566404442387785e-05,
+      "loss": 0.0029,
+      "step": 1508
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.156062008329477e-05,
+      "loss": 0.0014,
+      "step": 1509
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1554835724201765e-05,
+      "loss": 0.2233,
+      "step": 1510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.154905136510875e-05,
+      "loss": 0.0014,
+      "step": 1511
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.154326700601573e-05,
+      "loss": 0.037,
+      "step": 1512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.153748264692272e-05,
+      "loss": 0.4093,
+      "step": 1513
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.153169828782971e-05,
+      "loss": 0.0015,
+      "step": 1514
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.15259139287367e-05,
+      "loss": 0.0022,
+      "step": 1515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1520129569643686e-05,
+      "loss": 0.0016,
+      "step": 1516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.151434521055067e-05,
+      "loss": 0.0053,
+      "step": 1517
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.150856085145766e-05,
+      "loss": 0.015,
+      "step": 1518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1502776492364646e-05,
+      "loss": 0.4689,
+      "step": 1519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.149699213327163e-05,
+      "loss": 0.3911,
+      "step": 1520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1491207774178626e-05,
+      "loss": 0.0057,
+      "step": 1521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.148542341508561e-05,
+      "loss": 0.0044,
+      "step": 1522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.14796390559926e-05,
+      "loss": 0.0062,
+      "step": 1523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1473854696899587e-05,
+      "loss": 0.0037,
+      "step": 1524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.146807033780657e-05,
+      "loss": 0.0044,
+      "step": 1525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.146228597871356e-05,
+      "loss": 0.0046,
+      "step": 1526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.145650161962055e-05,
+      "loss": 0.0023,
+      "step": 1527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1450717260527534e-05,
+      "loss": 0.3383,
+      "step": 1528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.144493290143453e-05,
+      "loss": 0.3673,
+      "step": 1529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.143914854234151e-05,
+      "loss": 0.0192,
+      "step": 1530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1433364183248494e-05,
+      "loss": 0.004,
+      "step": 1531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.142757982415549e-05,
+      "loss": 0.0029,
+      "step": 1532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1421795465062474e-05,
+      "loss": 0.0056,
+      "step": 1533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.141601110596946e-05,
+      "loss": 0.0041,
+      "step": 1534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.141022674687645e-05,
+      "loss": 0.1208,
+      "step": 1535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1404442387783434e-05,
+      "loss": 0.2426,
+      "step": 1536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.139865802869042e-05,
+      "loss": 0.0893,
+      "step": 1537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.139287366959741e-05,
+      "loss": 0.0159,
+      "step": 1538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1387089310504395e-05,
+      "loss": 0.012,
+      "step": 1539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.138130495141139e-05,
+      "loss": 0.3107,
+      "step": 1540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1375520592318375e-05,
+      "loss": 0.1791,
+      "step": 1541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.136973623322536e-05,
+      "loss": 0.3775,
+      "step": 1542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.136395187413235e-05,
+      "loss": 0.0074,
+      "step": 1543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1358167515039335e-05,
+      "loss": 0.0589,
+      "step": 1544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.135238315594632e-05,
+      "loss": 0.0564,
+      "step": 1545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.134659879685331e-05,
+      "loss": 0.0237,
+      "step": 1546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.13408144377603e-05,
+      "loss": 0.2995,
+      "step": 1547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.133503007866729e-05,
+      "loss": 0.0107,
+      "step": 1548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.132924571957427e-05,
+      "loss": 0.1168,
+      "step": 1549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1323461360481256e-05,
+      "loss": 0.1376,
+      "step": 1550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.131767700138825e-05,
+      "loss": 0.015,
+      "step": 1551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1311892642295236e-05,
+      "loss": 0.0079,
+      "step": 1552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.130610828320222e-05,
+      "loss": 0.0091,
+      "step": 1553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1300323924109216e-05,
+      "loss": 0.0076,
+      "step": 1554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1294539565016196e-05,
+      "loss": 0.0032,
+      "step": 1555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.128875520592318e-05,
+      "loss": 0.0044,
+      "step": 1556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.128297084683017e-05,
+      "loss": 0.0072,
+      "step": 1557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.127718648773716e-05,
+      "loss": 0.0036,
+      "step": 1558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.127140212864415e-05,
+      "loss": 0.0035,
+      "step": 1559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.126561776955114e-05,
+      "loss": 0.3261,
+      "step": 1560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1259833410458124e-05,
+      "loss": 0.0044,
+      "step": 1561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.125404905136511e-05,
+      "loss": 0.0059,
+      "step": 1562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.12482646922721e-05,
+      "loss": 0.1779,
+      "step": 1563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1242480333179084e-05,
+      "loss": 0.0014,
+      "step": 1564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.123669597408608e-05,
+      "loss": 0.0025,
+      "step": 1565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1230911614993064e-05,
+      "loss": 0.0028,
+      "step": 1566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.122512725590005e-05,
+      "loss": 0.0019,
+      "step": 1567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.121934289680703e-05,
+      "loss": 0.002,
+      "step": 1568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1213558537714024e-05,
+      "loss": 0.0012,
+      "step": 1569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.120777417862101e-05,
+      "loss": 0.0011,
+      "step": 1570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1201989819528e-05,
+      "loss": 0.0086,
+      "step": 1571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1196205460434985e-05,
+      "loss": 0.0028,
+      "step": 1572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.119042110134198e-05,
+      "loss": 0.0012,
+      "step": 1573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.118463674224896e-05,
+      "loss": 0.0195,
+      "step": 1574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1178852383155945e-05,
+      "loss": 0.1627,
+      "step": 1575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.117306802406294e-05,
+      "loss": 0.0062,
+      "step": 1576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1167283664969925e-05,
+      "loss": 0.0008,
+      "step": 1577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.116149930587691e-05,
+      "loss": 0.0008,
+      "step": 1578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.11557149467839e-05,
+      "loss": 0.0029,
+      "step": 1579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1149930587690886e-05,
+      "loss": 0.0006,
+      "step": 1580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.114414622859787e-05,
+      "loss": 0.0007,
+      "step": 1581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.113836186950486e-05,
+      "loss": 0.0021,
+      "step": 1582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1132577510411846e-05,
+      "loss": 0.001,
+      "step": 1583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.112679315131884e-05,
+      "loss": 0.0102,
+      "step": 1584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1121008792225826e-05,
+      "loss": 0.2327,
+      "step": 1585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.111522443313281e-05,
+      "loss": 0.0005,
+      "step": 1586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.11094400740398e-05,
+      "loss": 0.0005,
+      "step": 1587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1103655714946786e-05,
+      "loss": 0.0007,
+      "step": 1588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.109787135585377e-05,
+      "loss": 0.0584,
+      "step": 1589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.109208699676076e-05,
+      "loss": 0.0003,
+      "step": 1590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1086302637667753e-05,
+      "loss": 0.001,
+      "step": 1591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1080518278574733e-05,
+      "loss": 0.0624,
+      "step": 1592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.107473391948172e-05,
+      "loss": 0.0005,
+      "step": 1593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.106894956038871e-05,
+      "loss": 0.0009,
+      "step": 1594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.10631652012957e-05,
+      "loss": 0.0015,
+      "step": 1595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.105738084220269e-05,
+      "loss": 0.1773,
+      "step": 1596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1051596483109674e-05,
+      "loss": 0.0005,
+      "step": 1597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.104581212401666e-05,
+      "loss": 0.0006,
+      "step": 1598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.104002776492365e-05,
+      "loss": 0.0005,
+      "step": 1599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1034243405830634e-05,
+      "loss": 0.0004,
+      "step": 1600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.102845904673762e-05,
+      "loss": 0.0004,
+      "step": 1601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1022674687644615e-05,
+      "loss": 0.0004,
+      "step": 1602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.10168903285516e-05,
+      "loss": 0.0036,
+      "step": 1603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.101110596945859e-05,
+      "loss": 0.0003,
+      "step": 1604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.100532161036557e-05,
+      "loss": 0.0003,
+      "step": 1605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.099953725127256e-05,
+      "loss": 0.0004,
+      "step": 1606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.099375289217955e-05,
+      "loss": 0.0009,
+      "step": 1607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0987968533086535e-05,
+      "loss": 0.0134,
+      "step": 1608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.098218417399352e-05,
+      "loss": 0.8866,
+      "step": 1609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0976399814900515e-05,
+      "loss": 0.2225,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0970615455807495e-05,
+      "loss": 0.0004,
+      "step": 1611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.096483109671448e-05,
+      "loss": 0.0004,
+      "step": 1612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0959046737621476e-05,
+      "loss": 0.0003,
+      "step": 1613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.095326237852846e-05,
+      "loss": 0.0004,
+      "step": 1614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.094747801943545e-05,
+      "loss": 0.0006,
+      "step": 1615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0941693660342436e-05,
+      "loss": 0.0012,
+      "step": 1616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.093590930124942e-05,
+      "loss": 0.0011,
+      "step": 1617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.093012494215641e-05,
+      "loss": 0.0007,
+      "step": 1618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0924340583063396e-05,
+      "loss": 0.0009,
+      "step": 1619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.091855622397038e-05,
+      "loss": 0.0006,
+      "step": 1620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0912771864877376e-05,
+      "loss": 0.0013,
+      "step": 1621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.090698750578436e-05,
+      "loss": 0.0007,
+      "step": 1622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.090120314669135e-05,
+      "loss": 0.0007,
+      "step": 1623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.089541878759834e-05,
+      "loss": 0.1933,
+      "step": 1624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0889634428505324e-05,
+      "loss": 0.0007,
+      "step": 1625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.088385006941231e-05,
+      "loss": 0.0006,
+      "step": 1626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.08780657103193e-05,
+      "loss": 0.0007,
+      "step": 1627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.087228135122629e-05,
+      "loss": 0.0139,
+      "step": 1628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.086649699213328e-05,
+      "loss": 0.0005,
+      "step": 1629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.086071263304026e-05,
+      "loss": 0.0006,
+      "step": 1630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0854928273947244e-05,
+      "loss": 0.0007,
+      "step": 1631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.084914391485424e-05,
+      "loss": 0.0008,
+      "step": 1632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0843359555761224e-05,
+      "loss": 0.0004,
+      "step": 1633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.083757519666821e-05,
+      "loss": 0.0006,
+      "step": 1634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.08317908375752e-05,
+      "loss": 0.0004,
+      "step": 1635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0826006478482185e-05,
+      "loss": 0.0008,
+      "step": 1636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.082022211938917e-05,
+      "loss": 0.0441,
+      "step": 1637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.081443776029616e-05,
+      "loss": 0.0006,
+      "step": 1638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.080865340120315e-05,
+      "loss": 0.1587,
+      "step": 1639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.080286904211014e-05,
+      "loss": 0.0004,
+      "step": 1640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0797084683017125e-05,
+      "loss": 0.0006,
+      "step": 1641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.079130032392411e-05,
+      "loss": 0.0006,
+      "step": 1642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.07855159648311e-05,
+      "loss": 0.007,
+      "step": 1643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0779731605738085e-05,
+      "loss": 0.0085,
+      "step": 1644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.077394724664507e-05,
+      "loss": 0.0015,
+      "step": 1645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0768162887552066e-05,
+      "loss": 0.1794,
+      "step": 1646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.076237852845905e-05,
+      "loss": 0.001,
+      "step": 1647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.075659416936603e-05,
+      "loss": 0.0011,
+      "step": 1648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.075080981027302e-05,
+      "loss": 0.0004,
+      "step": 1649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.074502545118001e-05,
+      "loss": 0.006,
+      "step": 1650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0739241092087e-05,
+      "loss": 0.001,
+      "step": 1651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0733456732993986e-05,
+      "loss": 0.0196,
+      "step": 1652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.072767237390097e-05,
+      "loss": 0.0337,
+      "step": 1653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.072188801480796e-05,
+      "loss": 0.4098,
+      "step": 1654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0716103655714947e-05,
+      "loss": 0.0005,
+      "step": 1655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.071031929662193e-05,
+      "loss": 0.0021,
+      "step": 1656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.070453493752893e-05,
+      "loss": 0.3095,
+      "step": 1657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0698750578435914e-05,
+      "loss": 0.0024,
+      "step": 1658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.06929662193429e-05,
+      "loss": 0.3628,
+      "step": 1659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.068718186024989e-05,
+      "loss": 0.0008,
+      "step": 1660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0681397501156874e-05,
+      "loss": 0.0005,
+      "step": 1661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.067561314206386e-05,
+      "loss": 0.3829,
+      "step": 1662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.066982878297085e-05,
+      "loss": 0.0015,
+      "step": 1663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0664044423877834e-05,
+      "loss": 0.0049,
+      "step": 1664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.065826006478483e-05,
+      "loss": 0.0054,
+      "step": 1665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0652475705691814e-05,
+      "loss": 0.0052,
+      "step": 1666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0646691346598794e-05,
+      "loss": 0.0004,
+      "step": 1667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.064090698750579e-05,
+      "loss": 0.1608,
+      "step": 1668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.0635122628412775e-05,
+      "loss": 0.344,
+      "step": 1669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.062933826931976e-05,
+      "loss": 0.0085,
+      "step": 1670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.062355391022675e-05,
+      "loss": 0.0008,
+      "step": 1671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.061776955113374e-05,
+      "loss": 0.0016,
+      "step": 1672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.061198519204072e-05,
+      "loss": 0.0012,
+      "step": 1673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.060620083294771e-05,
+      "loss": 0.0406,
+      "step": 1674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0600416473854695e-05,
+      "loss": 0.0033,
+      "step": 1675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.059463211476169e-05,
+      "loss": 0.0039,
+      "step": 1676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0588847755668676e-05,
+      "loss": 0.0007,
+      "step": 1677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.058306339657566e-05,
+      "loss": 0.0008,
+      "step": 1678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.057727903748265e-05,
+      "loss": 0.3779,
+      "step": 1679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0571494678389636e-05,
+      "loss": 0.0008,
+      "step": 1680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.056571031929662e-05,
+      "loss": 0.0007,
+      "step": 1681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.055992596020361e-05,
+      "loss": 0.0008,
+      "step": 1682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.05541416011106e-05,
+      "loss": 0.3698,
+      "step": 1683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.054835724201759e-05,
+      "loss": 0.001,
+      "step": 1684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0542572882924576e-05,
+      "loss": 0.0009,
+      "step": 1685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0536788523831556e-05,
+      "loss": 0.0034,
+      "step": 1686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.053100416473855e-05,
+      "loss": 0.0011,
+      "step": 1687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.052521980564554e-05,
+      "loss": 0.3359,
+      "step": 1688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0519435446552523e-05,
+      "loss": 0.0074,
+      "step": 1689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.051365108745951e-05,
+      "loss": 0.0012,
+      "step": 1690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0507866728366504e-05,
+      "loss": 0.4057,
+      "step": 1691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0502082369273484e-05,
+      "loss": 0.1219,
+      "step": 1692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.049629801018047e-05,
+      "loss": 0.0036,
+      "step": 1693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0490513651087464e-05,
+      "loss": 0.016,
+      "step": 1694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.048472929199445e-05,
+      "loss": 0.0017,
+      "step": 1695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.047894493290144e-05,
+      "loss": 0.1324,
+      "step": 1696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0473160573808424e-05,
+      "loss": 0.3762,
+      "step": 1697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.046737621471541e-05,
+      "loss": 0.0023,
+      "step": 1698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.04615918556224e-05,
+      "loss": 0.0274,
+      "step": 1699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0455807496529384e-05,
+      "loss": 0.0032,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.045002313743637e-05,
+      "loss": 0.0027,
+      "step": 1701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0444238778343365e-05,
+      "loss": 0.3197,
+      "step": 1702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.043845441925035e-05,
+      "loss": 0.0288,
+      "step": 1703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.043267006015734e-05,
+      "loss": 0.0266,
+      "step": 1704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0426885701064325e-05,
+      "loss": 0.0081,
+      "step": 1705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.042110134197131e-05,
+      "loss": 0.0028,
+      "step": 1706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.04153169828783e-05,
+      "loss": 0.0041,
+      "step": 1707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0409532623785285e-05,
+      "loss": 0.0277,
+      "step": 1708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.040374826469228e-05,
+      "loss": 0.0024,
+      "step": 1709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.039796390559926e-05,
+      "loss": 0.2437,
+      "step": 1710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0392179546506246e-05,
+      "loss": 0.1158,
+      "step": 1711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.038639518741323e-05,
+      "loss": 0.0041,
+      "step": 1712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0380610828320226e-05,
+      "loss": 0.0037,
+      "step": 1713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.037482646922721e-05,
+      "loss": 0.0085,
+      "step": 1714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.03690421101342e-05,
+      "loss": 0.1003,
+      "step": 1715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.0363257751041186e-05,
+      "loss": 0.0025,
+      "step": 1716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.035747339194817e-05,
+      "loss": 0.0024,
+      "step": 1717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.035168903285516e-05,
+      "loss": 0.0054,
+      "step": 1718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0345904673762146e-05,
+      "loss": 0.0017,
+      "step": 1719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.034012031466914e-05,
+      "loss": 0.2748,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.033433595557613e-05,
+      "loss": 0.0033,
+      "step": 1721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0328551596483113e-05,
+      "loss": 0.2261,
+      "step": 1722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0322767237390093e-05,
+      "loss": 0.0022,
+      "step": 1723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.031698287829709e-05,
+      "loss": 0.4486,
+      "step": 1724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0311198519204074e-05,
+      "loss": 0.0028,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.030541416011106e-05,
+      "loss": 0.0035,
+      "step": 1726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0299629801018054e-05,
+      "loss": 0.0018,
+      "step": 1727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.029384544192504e-05,
+      "loss": 0.0022,
+      "step": 1728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.028806108283202e-05,
+      "loss": 0.0005,
+      "step": 1729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.028227672373901e-05,
+      "loss": 0.0012,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0276492364646e-05,
+      "loss": 0.0013,
+      "step": 1731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.027070800555299e-05,
+      "loss": 0.0007,
+      "step": 1732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0264923646459975e-05,
+      "loss": 0.0011,
+      "step": 1733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.025913928736696e-05,
+      "loss": 0.0015,
+      "step": 1734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.025335492827395e-05,
+      "loss": 0.0012,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0247570569180935e-05,
+      "loss": 0.0005,
+      "step": 1736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.024178621008792e-05,
+      "loss": 0.2234,
+      "step": 1737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0236001850994915e-05,
+      "loss": 0.001,
+      "step": 1738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.02302174919019e-05,
+      "loss": 0.0023,
+      "step": 1739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.022443313280889e-05,
+      "loss": 0.0084,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0218648773715875e-05,
+      "loss": 0.024,
+      "step": 1741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.021286441462286e-05,
+      "loss": 0.0008,
+      "step": 1742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.020708005552985e-05,
+      "loss": 0.0029,
+      "step": 1743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0201295696436836e-05,
+      "loss": 0.0019,
+      "step": 1744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.019551133734382e-05,
+      "loss": 0.0068,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0189726978250816e-05,
+      "loss": 0.1318,
+      "step": 1746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.01839426191578e-05,
+      "loss": 0.0029,
+      "step": 1747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.017815826006478e-05,
+      "loss": 0.0007,
+      "step": 1748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0172373900971776e-05,
+      "loss": 0.0024,
+      "step": 1749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.016658954187876e-05,
+      "loss": 0.0004,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.016080518278575e-05,
+      "loss": 0.0035,
+      "step": 1751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0155020823692737e-05,
+      "loss": 0.0005,
+      "step": 1752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.014923646459972e-05,
+      "loss": 0.0003,
+      "step": 1753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.014345210550671e-05,
+      "loss": 0.0003,
+      "step": 1754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.01376677464137e-05,
+      "loss": 0.0002,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0131883387320684e-05,
+      "loss": 0.0003,
+      "step": 1756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.012609902822768e-05,
+      "loss": 0.0006,
+      "step": 1757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.0120314669134664e-05,
+      "loss": 0.0003,
+      "step": 1758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.011453031004165e-05,
+      "loss": 0.1856,
+      "step": 1759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.010874595094864e-05,
+      "loss": 0.0006,
+      "step": 1760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0102961591855624e-05,
+      "loss": 0.0004,
+      "step": 1761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.009717723276261e-05,
+      "loss": 0.0004,
+      "step": 1762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.00913928736696e-05,
+      "loss": 0.0004,
+      "step": 1763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.008560851457659e-05,
+      "loss": 0.4128,
+      "step": 1764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.007982415548358e-05,
+      "loss": 0.5177,
+      "step": 1765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.007403979639056e-05,
+      "loss": 0.5498,
+      "step": 1766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0068255437297545e-05,
+      "loss": 0.0008,
+      "step": 1767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.006247107820454e-05,
+      "loss": 0.0017,
+      "step": 1768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0056686719111525e-05,
+      "loss": 0.0011,
+      "step": 1769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.005090236001851e-05,
+      "loss": 0.0038,
+      "step": 1770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.00451180009255e-05,
+      "loss": 0.6525,
+      "step": 1771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0039333641832485e-05,
+      "loss": 0.0017,
+      "step": 1772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.003354928273947e-05,
+      "loss": 0.0019,
+      "step": 1773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.002776492364646e-05,
+      "loss": 0.0014,
+      "step": 1774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.002198056455345e-05,
+      "loss": 0.0072,
+      "step": 1775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.001619620546044e-05,
+      "loss": 0.0029,
+      "step": 1776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.0010411846367426e-05,
+      "loss": 0.0087,
+      "step": 1777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.000462748727441e-05,
+      "loss": 0.2238,
+      "step": 1778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.99988431281814e-05,
+      "loss": 0.0051,
+      "step": 1779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9993058769088386e-05,
+      "loss": 0.38,
+      "step": 1780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998727440999537e-05,
+      "loss": 0.0073,
+      "step": 1781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998149005090236e-05,
+      "loss": 0.0025,
+      "step": 1782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.997570569180935e-05,
+      "loss": 0.1464,
+      "step": 1783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.996992133271634e-05,
+      "loss": 0.0033,
+      "step": 1784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.996413697362332e-05,
+      "loss": 0.005,
+      "step": 1785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.995835261453031e-05,
+      "loss": 0.0081,
+      "step": 1786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.99525682554373e-05,
+      "loss": 0.0029,
+      "step": 1787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.994678389634429e-05,
+      "loss": 0.0021,
+      "step": 1788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9940999537251274e-05,
+      "loss": 0.0026,
+      "step": 1789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.993521517815827e-05,
+      "loss": 0.0548,
+      "step": 1790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.992943081906525e-05,
+      "loss": 0.0024,
+      "step": 1791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9923646459972234e-05,
+      "loss": 0.0033,
+      "step": 1792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.991786210087922e-05,
+      "loss": 0.0029,
+      "step": 1793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9912077741786214e-05,
+      "loss": 0.0011,
+      "step": 1794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.99062933826932e-05,
+      "loss": 0.002,
+      "step": 1795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.990050902360019e-05,
+      "loss": 0.2584,
+      "step": 1796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9894724664507174e-05,
+      "loss": 0.0019,
+      "step": 1797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.988894030541416e-05,
+      "loss": 0.0022,
+      "step": 1798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.988315594632115e-05,
+      "loss": 0.0017,
+      "step": 1799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9877371587228135e-05,
+      "loss": 0.0017,
+      "step": 1800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.987158722813513e-05,
+      "loss": 0.0011,
+      "step": 1801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9865802869042115e-05,
+      "loss": 0.0065,
+      "step": 1802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.98600185099491e-05,
+      "loss": 0.0014,
+      "step": 1803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.985423415085608e-05,
+      "loss": 0.0018,
+      "step": 1804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9848449791763075e-05,
+      "loss": 0.0011,
+      "step": 1805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.984266543267006e-05,
+      "loss": 0.0006,
+      "step": 1806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.983688107357705e-05,
+      "loss": 0.0014,
+      "step": 1807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.983109671448404e-05,
+      "loss": 0.0008,
+      "step": 1808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.982531235539102e-05,
+      "loss": 0.0109,
+      "step": 1809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.981952799629801e-05,
+      "loss": 0.0006,
+      "step": 1810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9813743637204996e-05,
+      "loss": 0.0007,
+      "step": 1811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.980795927811199e-05,
+      "loss": 0.4597,
+      "step": 1812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9802174919018976e-05,
+      "loss": 0.0009,
+      "step": 1813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.979639055992596e-05,
+      "loss": 0.0007,
+      "step": 1814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.979060620083295e-05,
+      "loss": 0.0006,
+      "step": 1815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9784821841739936e-05,
+      "loss": 0.0011,
+      "step": 1816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.977903748264692e-05,
+      "loss": 0.1585,
+      "step": 1817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.977325312355391e-05,
+      "loss": 0.6212,
+      "step": 1818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9767468764460903e-05,
+      "loss": 0.0009,
+      "step": 1819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.976168440536789e-05,
+      "loss": 0.0014,
+      "step": 1820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.975590004627488e-05,
+      "loss": 0.0014,
+      "step": 1821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.975011568718186e-05,
+      "loss": 0.0076,
+      "step": 1822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.974433132808885e-05,
+      "loss": 0.2041,
+      "step": 1823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.973854696899584e-05,
+      "loss": 0.0012,
+      "step": 1824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9732762609902824e-05,
+      "loss": 0.0031,
+      "step": 1825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.972697825080981e-05,
+      "loss": 0.1621,
+      "step": 1826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9721193891716804e-05,
+      "loss": 0.001,
+      "step": 1827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9715409532623784e-05,
+      "loss": 0.002,
+      "step": 1828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.970962517353077e-05,
+      "loss": 0.0024,
+      "step": 1829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9703840814437765e-05,
+      "loss": 0.0158,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.969805645534475e-05,
+      "loss": 0.002,
+      "step": 1831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.969227209625174e-05,
+      "loss": 0.0103,
+      "step": 1832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9686487737158725e-05,
+      "loss": 0.0024,
+      "step": 1833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.968070337806571e-05,
+      "loss": 0.0016,
+      "step": 1834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.96749190189727e-05,
+      "loss": 0.0262,
+      "step": 1835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9669134659879685e-05,
+      "loss": 0.0013,
+      "step": 1836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.966335030078667e-05,
+      "loss": 0.0102,
+      "step": 1837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9657565941693665e-05,
+      "loss": 0.3573,
+      "step": 1838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.965178158260065e-05,
+      "loss": 0.0017,
+      "step": 1839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.964599722350764e-05,
+      "loss": 0.0018,
+      "step": 1840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.964021286441462e-05,
+      "loss": 0.0021,
+      "step": 1841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.963442850532161e-05,
+      "loss": 0.0011,
+      "step": 1842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.96286441462286e-05,
+      "loss": 0.0016,
+      "step": 1843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9622859787135586e-05,
+      "loss": 0.4682,
+      "step": 1844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.961707542804258e-05,
+      "loss": 0.0023,
+      "step": 1845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9611291068949566e-05,
+      "loss": 0.0046,
+      "step": 1846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9605506709856546e-05,
+      "loss": 0.0039,
+      "step": 1847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.959972235076353e-05,
+      "loss": 0.1009,
+      "step": 1848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9593937991670526e-05,
+      "loss": 0.0023,
+      "step": 1849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.958815363257751e-05,
+      "loss": 0.0017,
+      "step": 1850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.95823692734845e-05,
+      "loss": 0.0014,
+      "step": 1851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.957658491439149e-05,
+      "loss": 0.003,
+      "step": 1852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9570800555298473e-05,
+      "loss": 0.0013,
+      "step": 1853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.956501619620546e-05,
+      "loss": 0.0013,
+      "step": 1854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.955923183711245e-05,
+      "loss": 0.0019,
+      "step": 1855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.955344747801944e-05,
+      "loss": 0.0008,
+      "step": 1856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.954766311892643e-05,
+      "loss": 0.0025,
+      "step": 1857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9541878759833414e-05,
+      "loss": 0.0007,
+      "step": 1858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.95360944007404e-05,
+      "loss": 0.0036,
+      "step": 1859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.953031004164739e-05,
+      "loss": 0.001,
+      "step": 1860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9524525682554374e-05,
+      "loss": 0.0013,
+      "step": 1861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.951874132346136e-05,
+      "loss": 0.0005,
+      "step": 1862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.951295696436835e-05,
+      "loss": 0.0005,
+      "step": 1863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.950717260527534e-05,
+      "loss": 0.0007,
+      "step": 1864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.950138824618233e-05,
+      "loss": 0.0016,
+      "step": 1865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.949560388708931e-05,
+      "loss": 0.0009,
+      "step": 1866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.94898195279963e-05,
+      "loss": 0.001,
+      "step": 1867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.948403516890329e-05,
+      "loss": 0.0006,
+      "step": 1868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9478250809810275e-05,
+      "loss": 0.0007,
+      "step": 1869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.947246645071726e-05,
+      "loss": 0.0007,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.946668209162425e-05,
+      "loss": 0.0004,
+      "step": 1871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9460897732531235e-05,
+      "loss": 0.0011,
+      "step": 1872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.945511337343822e-05,
+      "loss": 0.0028,
+      "step": 1873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.944932901434521e-05,
+      "loss": 0.1398,
+      "step": 1874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.94435446552522e-05,
+      "loss": 0.0005,
+      "step": 1875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.943776029615919e-05,
+      "loss": 0.1008,
+      "step": 1876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9431975937066176e-05,
+      "loss": 0.002,
+      "step": 1877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.942619157797316e-05,
+      "loss": 0.0003,
+      "step": 1878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.942040721888015e-05,
+      "loss": 0.0451,
+      "step": 1879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9414622859787136e-05,
+      "loss": 0.0006,
+      "step": 1880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.940883850069412e-05,
+      "loss": 0.0004,
+      "step": 1881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9403054141601117e-05,
+      "loss": 0.0592,
+      "step": 1882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.93972697825081e-05,
+      "loss": 0.0006,
+      "step": 1883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.939148542341508e-05,
+      "loss": 0.0004,
+      "step": 1884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.938570106432207e-05,
+      "loss": 0.0004,
+      "step": 1885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9379916705229064e-05,
+      "loss": 0.0003,
+      "step": 1886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.937413234613605e-05,
+      "loss": 0.0003,
+      "step": 1887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.936834798704304e-05,
+      "loss": 0.001,
+      "step": 1888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.936256362795003e-05,
+      "loss": 0.0034,
+      "step": 1889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.935677926885701e-05,
+      "loss": 0.0005,
+      "step": 1890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9350994909764e-05,
+      "loss": 0.0493,
+      "step": 1891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9345210550670984e-05,
+      "loss": 0.0003,
+      "step": 1892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.933942619157798e-05,
+      "loss": 0.0004,
+      "step": 1893
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9333641832484964e-05,
+      "loss": 0.0005,
+      "step": 1894
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.932785747339195e-05,
+      "loss": 0.0005,
+      "step": 1895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.932207311429894e-05,
+      "loss": 0.0003,
+      "step": 1896
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9316288755205925e-05,
+      "loss": 0.0003,
+      "step": 1897
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.931050439611291e-05,
+      "loss": 0.0004,
+      "step": 1898
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.93047200370199e-05,
+      "loss": 0.0005,
+      "step": 1899
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.929893567792689e-05,
+      "loss": 0.0003,
+      "step": 1900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.929315131883388e-05,
+      "loss": 0.0003,
+      "step": 1901
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9287366959740865e-05,
+      "loss": 0.0151,
+      "step": 1902
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9281582600647845e-05,
+      "loss": 0.0002,
+      "step": 1903
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.927579824155484e-05,
+      "loss": 0.4797,
+      "step": 1904
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9270013882461825e-05,
+      "loss": 0.7987,
+      "step": 1905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.926422952336881e-05,
+      "loss": 0.0003,
+      "step": 1906
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.92584451642758e-05,
+      "loss": 0.0002,
+      "step": 1907
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.925266080518279e-05,
+      "loss": 0.0003,
+      "step": 1908
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.924687644608977e-05,
+      "loss": 0.3136,
+      "step": 1909
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.924109208699676e-05,
+      "loss": 0.0003,
+      "step": 1910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.923530772790375e-05,
+      "loss": 0.0007,
+      "step": 1911
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.922952336881074e-05,
+      "loss": 0.0003,
+      "step": 1912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9223739009717726e-05,
+      "loss": 0.205,
+      "step": 1913
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.921795465062471e-05,
+      "loss": 0.0005,
+      "step": 1914
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.92121702915317e-05,
+      "loss": 0.0018,
+      "step": 1915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9206385932438687e-05,
+      "loss": 0.0028,
+      "step": 1916
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.920060157334567e-05,
+      "loss": 0.0007,
+      "step": 1917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.919481721425266e-05,
+      "loss": 0.0029,
+      "step": 1918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9189032855159654e-05,
+      "loss": 0.007,
+      "step": 1919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.918324849606664e-05,
+      "loss": 0.0011,
+      "step": 1920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.917746413697363e-05,
+      "loss": 0.0005,
+      "step": 1921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.917167977788061e-05,
+      "loss": 0.165,
+      "step": 1922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.91658954187876e-05,
+      "loss": 0.0004,
+      "step": 1923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.916011105969459e-05,
+      "loss": 0.0008,
+      "step": 1924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9154326700601574e-05,
+      "loss": 0.0029,
+      "step": 1925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.914854234150857e-05,
+      "loss": 0.0035,
+      "step": 1926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.914275798241555e-05,
+      "loss": 0.0005,
+      "step": 1927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9136973623322534e-05,
+      "loss": 0.0005,
+      "step": 1928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.913118926422952e-05,
+      "loss": 0.6259,
+      "step": 1929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9125404905136515e-05,
+      "loss": 0.0005,
+      "step": 1930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.91196205460435e-05,
+      "loss": 0.0003,
+      "step": 1931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.911383618695049e-05,
+      "loss": 0.0008,
+      "step": 1932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9108051827857475e-05,
+      "loss": 0.0004,
+      "step": 1933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.910226746876446e-05,
+      "loss": 0.0004,
+      "step": 1934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.909648310967145e-05,
+      "loss": 0.0004,
+      "step": 1935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9090698750578435e-05,
+      "loss": 0.0011,
+      "step": 1936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.908491439148543e-05,
+      "loss": 0.2063,
+      "step": 1937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9079130032392416e-05,
+      "loss": 0.0005,
+      "step": 1938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.90733456732994e-05,
+      "loss": 0.0021,
+      "step": 1939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.906756131420638e-05,
+      "loss": 0.0005,
+      "step": 1940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9061776955113376e-05,
+      "loss": 0.0019,
+      "step": 1941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.905599259602036e-05,
+      "loss": 0.0004,
+      "step": 1942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.905020823692735e-05,
+      "loss": 0.0007,
+      "step": 1943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9044423877834336e-05,
+      "loss": 0.0006,
+      "step": 1944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.903863951874133e-05,
+      "loss": 0.0009,
+      "step": 1945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.903285515964831e-05,
+      "loss": 0.0011,
+      "step": 1946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9027070800555296e-05,
+      "loss": 0.0009,
+      "step": 1947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.902128644146229e-05,
+      "loss": 0.0002,
+      "step": 1948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.901550208236928e-05,
+      "loss": 0.6755,
+      "step": 1949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9009717723276263e-05,
+      "loss": 0.0011,
+      "step": 1950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.900393336418325e-05,
+      "loss": 0.001,
+      "step": 1951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.899814900509024e-05,
+      "loss": 0.0138,
+      "step": 1952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.8992364645997224e-05,
+      "loss": 0.2476,
+      "step": 1953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.898658028690421e-05,
+      "loss": 0.138,
+      "step": 1954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.89807959278112e-05,
+      "loss": 0.0821,
+      "step": 1955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.897501156871819e-05,
+      "loss": 0.2727,
+      "step": 1956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.896922720962518e-05,
+      "loss": 0.0006,
+      "step": 1957
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.8963442850532164e-05,
+      "loss": 0.0047,
+      "step": 1958
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.895765849143915e-05,
+      "loss": 0.0007,
+      "step": 1959
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.895187413234614e-05,
+      "loss": 0.0048,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.8946089773253125e-05,
+      "loss": 0.0008,
+      "step": 1961
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.894030541416011e-05,
+      "loss": 0.0021,
+      "step": 1962
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.8934521055067105e-05,
+      "loss": 0.008,
+      "step": 1963
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.892873669597409e-05,
+      "loss": 0.0013,
+      "step": 1964
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.892295233688107e-05,
+      "loss": 0.0098,
+      "step": 1965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.891716797778806e-05,
+      "loss": 0.0004,
+      "step": 1966
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.891138361869505e-05,
+      "loss": 0.0054,
+      "step": 1967
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.890559925960204e-05,
+      "loss": 0.0049,
+      "step": 1968
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.8899814900509025e-05,
+      "loss": 0.1734,
+      "step": 1969
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.889403054141602e-05,
+      "loss": 0.0005,
+      "step": 1970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.8888246182323e-05,
+      "loss": 0.0072,
+      "step": 1971
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.8882461823229986e-05,
+      "loss": 0.0004,
+      "step": 1972
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.887667746413697e-05,
+      "loss": 0.2543,
+      "step": 1973
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.8870893105043966e-05,
+      "loss": 0.0011,
+      "step": 1974
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.886510874595095e-05,
+      "loss": 0.0004,
+      "step": 1975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.885932438685794e-05,
+      "loss": 0.0004,
+      "step": 1976
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.8853540027764926e-05,
+      "loss": 0.0004,
+      "step": 1977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.884775566867191e-05,
+      "loss": 0.0005,
+      "step": 1978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.88419713095789e-05,
+      "loss": 0.1642,
+      "step": 1979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8836186950485886e-05,
+      "loss": 0.0056,
+      "step": 1980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.883040259139288e-05,
+      "loss": 0.2036,
+      "step": 1981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.882461823229987e-05,
+      "loss": 0.0005,
+      "step": 1982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8818833873206854e-05,
+      "loss": 0.0009,
+      "step": 1983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8813049514113834e-05,
+      "loss": 0.0267,
+      "step": 1984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.880726515502083e-05,
+      "loss": 0.0005,
+      "step": 1985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8801480795927814e-05,
+      "loss": 0.0003,
+      "step": 1986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.87956964368348e-05,
+      "loss": 0.0003,
+      "step": 1987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.878991207774179e-05,
+      "loss": 0.0013,
+      "step": 1988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8784127718648774e-05,
+      "loss": 0.0025,
+      "step": 1989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.877834335955576e-05,
+      "loss": 0.0013,
+      "step": 1990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.877255900046275e-05,
+      "loss": 0.1733,
+      "step": 1991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.876677464136974e-05,
+      "loss": 0.1769,
+      "step": 1992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.876099028227673e-05,
+      "loss": 0.0006,
+      "step": 1993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8755205923183715e-05,
+      "loss": 0.0013,
+      "step": 1994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.87494215640907e-05,
+      "loss": 0.0012,
+      "step": 1995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.874363720499769e-05,
+      "loss": 0.0012,
+      "step": 1996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8737852845904675e-05,
+      "loss": 0.0061,
+      "step": 1997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.873206848681166e-05,
+      "loss": 0.0008,
+      "step": 1998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.872628412771865e-05,
+      "loss": 0.0003,
+      "step": 1999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.872049976862564e-05,
+      "loss": 0.0004,
+      "step": 2000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.871471540953263e-05,
+      "loss": 0.1671,
+      "step": 2001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.870893105043961e-05,
+      "loss": 0.0004,
+      "step": 2002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.87031466913466e-05,
+      "loss": 0.0011,
+      "step": 2003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.869736233225359e-05,
+      "loss": 0.0005,
+      "step": 2004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8691577973160576e-05,
+      "loss": 0.0004,
+      "step": 2005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.868579361406756e-05,
+      "loss": 0.0008,
+      "step": 2006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8680009254974556e-05,
+      "loss": 0.0003,
+      "step": 2007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8674224895881536e-05,
+      "loss": 0.0004,
+      "step": 2008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.866844053678852e-05,
+      "loss": 0.0002,
+      "step": 2009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.866265617769551e-05,
+      "loss": 0.0003,
+      "step": 2010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.86568718186025e-05,
+      "loss": 0.0002,
+      "step": 2011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.865108745950949e-05,
+      "loss": 0.1383,
+      "step": 2012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8645303100416477e-05,
+      "loss": 0.0003,
+      "step": 2013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.863951874132346e-05,
+      "loss": 0.0004,
+      "step": 2014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.863373438223045e-05,
+      "loss": 0.0003,
+      "step": 2015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.862795002313744e-05,
+      "loss": 0.0008,
+      "step": 2016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8622165664044424e-05,
+      "loss": 0.012,
+      "step": 2017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.861638130495142e-05,
+      "loss": 0.0004,
+      "step": 2018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.8610596945858404e-05,
+      "loss": 0.0002,
+      "step": 2019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.860481258676539e-05,
+      "loss": 0.0003,
+      "step": 2020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.859902822767237e-05,
+      "loss": 0.1133,
+      "step": 2021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8593243868579364e-05,
+      "loss": 0.0002,
+      "step": 2022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.858745950948635e-05,
+      "loss": 0.001,
+      "step": 2023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.858167515039334e-05,
+      "loss": 0.0026,
+      "step": 2024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8575890791300324e-05,
+      "loss": 0.0003,
+      "step": 2025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.857010643220732e-05,
+      "loss": 0.0002,
+      "step": 2026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.85643220731143e-05,
+      "loss": 0.0005,
+      "step": 2027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8558537714021285e-05,
+      "loss": 0.0269,
+      "step": 2028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.855275335492828e-05,
+      "loss": 0.0015,
+      "step": 2029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8546968995835265e-05,
+      "loss": 0.0059,
+      "step": 2030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.854118463674225e-05,
+      "loss": 0.0003,
+      "step": 2031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.853540027764924e-05,
+      "loss": 0.0004,
+      "step": 2032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8529615918556225e-05,
+      "loss": 0.0002,
+      "step": 2033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.852383155946321e-05,
+      "loss": 0.023,
+      "step": 2034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.85180472003702e-05,
+      "loss": 0.2956,
+      "step": 2035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8512262841277186e-05,
+      "loss": 0.0004,
+      "step": 2036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.850647848218418e-05,
+      "loss": 0.0006,
+      "step": 2037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8500694123091166e-05,
+      "loss": 0.0002,
+      "step": 2038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.849490976399815e-05,
+      "loss": 0.0039,
+      "step": 2039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.848912540490514e-05,
+      "loss": 0.0003,
+      "step": 2040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8483341045812126e-05,
+      "loss": 0.0114,
+      "step": 2041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.847755668671911e-05,
+      "loss": 0.0002,
+      "step": 2042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.84717723276261e-05,
+      "loss": 0.0318,
+      "step": 2043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.846598796853309e-05,
+      "loss": 0.0001,
+      "step": 2044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.846020360944007e-05,
+      "loss": 0.5115,
+      "step": 2045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.845441925034706e-05,
+      "loss": 0.0002,
+      "step": 2046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.844863489125405e-05,
+      "loss": 0.0019,
+      "step": 2047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.844285053216104e-05,
+      "loss": 0.0003,
+      "step": 2048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.843706617306803e-05,
+      "loss": 0.0036,
+      "step": 2049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8431281813975014e-05,
+      "loss": 0.4949,
+      "step": 2050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8425497454882e-05,
+      "loss": 0.0002,
+      "step": 2051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.841971309578899e-05,
+      "loss": 0.0056,
+      "step": 2052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8413928736695974e-05,
+      "loss": 0.0003,
+      "step": 2053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.840814437760296e-05,
+      "loss": 0.0007,
+      "step": 2054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8402360018509954e-05,
+      "loss": 0.0005,
+      "step": 2055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.839657565941694e-05,
+      "loss": 0.0006,
+      "step": 2056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.839079130032393e-05,
+      "loss": 0.0025,
+      "step": 2057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.838500694123091e-05,
+      "loss": 0.0008,
+      "step": 2058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.83792225821379e-05,
+      "loss": 0.0007,
+      "step": 2059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.837343822304489e-05,
+      "loss": 0.0011,
+      "step": 2060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8367653863951875e-05,
+      "loss": 0.001,
+      "step": 2061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.836186950485887e-05,
+      "loss": 0.0027,
+      "step": 2062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8356085145765855e-05,
+      "loss": 0.0011,
+      "step": 2063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.8350300786672835e-05,
+      "loss": 0.0019,
+      "step": 2064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.834451642757982e-05,
+      "loss": 0.0022,
+      "step": 2065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8338732068486815e-05,
+      "loss": 0.0009,
+      "step": 2066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.83329477093938e-05,
+      "loss": 0.001,
+      "step": 2067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.832716335030079e-05,
+      "loss": 0.0024,
+      "step": 2068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8321378991207776e-05,
+      "loss": 0.0012,
+      "step": 2069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.831559463211476e-05,
+      "loss": 0.0006,
+      "step": 2070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.830981027302175e-05,
+      "loss": 0.0013,
+      "step": 2071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8304025913928736e-05,
+      "loss": 0.0007,
+      "step": 2072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.829824155483573e-05,
+      "loss": 0.0008,
+      "step": 2073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8292457195742716e-05,
+      "loss": 0.0009,
+      "step": 2074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.82866728366497e-05,
+      "loss": 0.0009,
+      "step": 2075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.828088847755669e-05,
+      "loss": 0.0216,
+      "step": 2076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8275104118463676e-05,
+      "loss": 0.001,
+      "step": 2077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.826931975937066e-05,
+      "loss": 0.0009,
+      "step": 2078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.826353540027765e-05,
+      "loss": 0.0007,
+      "step": 2079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.825775104118464e-05,
+      "loss": 0.0008,
+      "step": 2080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.825196668209163e-05,
+      "loss": 0.001,
+      "step": 2081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.824618232299862e-05,
+      "loss": 0.0005,
+      "step": 2082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.82403979639056e-05,
+      "loss": 0.0005,
+      "step": 2083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.823461360481259e-05,
+      "loss": 0.0006,
+      "step": 2084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.822882924571958e-05,
+      "loss": 0.0003,
+      "step": 2085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8223044886626564e-05,
+      "loss": 0.0016,
+      "step": 2086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.821726052753355e-05,
+      "loss": 0.0051,
+      "step": 2087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.821147616844054e-05,
+      "loss": 0.0003,
+      "step": 2088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8205691809347524e-05,
+      "loss": 0.0006,
+      "step": 2089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.819990745025451e-05,
+      "loss": 0.0004,
+      "step": 2090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.81941230911615e-05,
+      "loss": 0.0003,
+      "step": 2091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.818833873206849e-05,
+      "loss": 0.0005,
+      "step": 2092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.818255437297548e-05,
+      "loss": 0.0004,
+      "step": 2093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8176770013882465e-05,
+      "loss": 0.0005,
+      "step": 2094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.817098565478945e-05,
+      "loss": 0.0003,
+      "step": 2095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.816520129569644e-05,
+      "loss": 0.0002,
+      "step": 2096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8159416936603425e-05,
+      "loss": 0.0002,
+      "step": 2097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.815363257751041e-05,
+      "loss": 0.0004,
+      "step": 2098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8147848218417405e-05,
+      "loss": 0.4098,
+      "step": 2099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.814206385932439e-05,
+      "loss": 0.4717,
+      "step": 2100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.813627950023137e-05,
+      "loss": 0.0019,
+      "step": 2101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.813049514113836e-05,
+      "loss": 0.0006,
+      "step": 2102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.812471078204535e-05,
+      "loss": 0.399,
+      "step": 2103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.811892642295234e-05,
+      "loss": 0.0019,
+      "step": 2104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8113142063859326e-05,
+      "loss": 0.6418,
+      "step": 2105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.810735770476631e-05,
+      "loss": 0.0031,
+      "step": 2106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.81015733456733e-05,
+      "loss": 0.7201,
+      "step": 2107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.8095788986580286e-05,
+      "loss": 0.005,
+      "step": 2108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.809000462748727e-05,
+      "loss": 0.0056,
+      "step": 2109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.8084220268394267e-05,
+      "loss": 0.2618,
+      "step": 2110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.807843590930125e-05,
+      "loss": 0.004,
+      "step": 2111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.807265155020824e-05,
+      "loss": 0.0056,
+      "step": 2112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.806686719111523e-05,
+      "loss": 0.0053,
+      "step": 2113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.8061082832022214e-05,
+      "loss": 0.0114,
+      "step": 2114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.80552984729292e-05,
+      "loss": 0.1976,
+      "step": 2115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.804951411383619e-05,
+      "loss": 0.0046,
+      "step": 2116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.8043729754743174e-05,
+      "loss": 0.0193,
+      "step": 2117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.803794539565017e-05,
+      "loss": 0.5146,
+      "step": 2118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.8032161036557154e-05,
+      "loss": 0.2259,
+      "step": 2119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.8026376677464134e-05,
+      "loss": 0.0069,
+      "step": 2120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.802059231837113e-05,
+      "loss": 0.0061,
+      "step": 2121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.8014807959278114e-05,
+      "loss": 0.0078,
+      "step": 2122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.80090236001851e-05,
+      "loss": 0.0126,
+      "step": 2123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.800323924109209e-05,
+      "loss": 0.0045,
+      "step": 2124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.799745488199908e-05,
+      "loss": 0.0082,
+      "step": 2125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.799167052290606e-05,
+      "loss": 0.0026,
+      "step": 2126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.798588616381305e-05,
+      "loss": 0.0022,
+      "step": 2127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7980101804720035e-05,
+      "loss": 0.0033,
+      "step": 2128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.797431744562703e-05,
+      "loss": 0.0015,
+      "step": 2129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7968533086534015e-05,
+      "loss": 0.0018,
+      "step": 2130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7962748727441e-05,
+      "loss": 0.4015,
+      "step": 2131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.795696436834799e-05,
+      "loss": 0.0009,
+      "step": 2132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7951180009254975e-05,
+      "loss": 0.0026,
+      "step": 2133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.794539565016196e-05,
+      "loss": 0.0015,
+      "step": 2134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.793961129106895e-05,
+      "loss": 0.2338,
+      "step": 2135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.793382693197594e-05,
+      "loss": 0.0331,
+      "step": 2136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.792804257288293e-05,
+      "loss": 0.0029,
+      "step": 2137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7922258213789916e-05,
+      "loss": 0.0021,
+      "step": 2138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7916473854696896e-05,
+      "loss": 0.2038,
+      "step": 2139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.791068949560389e-05,
+      "loss": 0.0021,
+      "step": 2140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7904905136510876e-05,
+      "loss": 0.004,
+      "step": 2141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.789912077741786e-05,
+      "loss": 0.059,
+      "step": 2142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7893336418324857e-05,
+      "loss": 0.0017,
+      "step": 2143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.788755205923184e-05,
+      "loss": 0.0052,
+      "step": 2144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.788176770013882e-05,
+      "loss": 0.0022,
+      "step": 2145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.787598334104581e-05,
+      "loss": 0.0026,
+      "step": 2146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7870198981952804e-05,
+      "loss": 0.0075,
+      "step": 2147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.786441462285979e-05,
+      "loss": 0.7179,
+      "step": 2148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.785863026376678e-05,
+      "loss": 0.7893,
+      "step": 2149
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.7852845904673764e-05,
+      "loss": 0.007,
+      "step": 2150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.784706154558075e-05,
+      "loss": 0.0007,
+      "step": 2151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.784127718648774e-05,
+      "loss": 0.0041,
+      "step": 2152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7835492827394724e-05,
+      "loss": 0.0029,
+      "step": 2153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.782970846830172e-05,
+      "loss": 0.2133,
+      "step": 2154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7823924109208704e-05,
+      "loss": 0.0022,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.781813975011569e-05,
+      "loss": 0.0062,
+      "step": 2156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.781235539102268e-05,
+      "loss": 0.0015,
+      "step": 2157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7806571031929665e-05,
+      "loss": 0.0026,
+      "step": 2158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.780078667283665e-05,
+      "loss": 0.0058,
+      "step": 2159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.779500231374364e-05,
+      "loss": 0.0011,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7789217954650625e-05,
+      "loss": 0.0345,
+      "step": 2161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.778343359555762e-05,
+      "loss": 0.0016,
+      "step": 2162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.77776492364646e-05,
+      "loss": 0.0028,
+      "step": 2163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7771864877371585e-05,
+      "loss": 0.0023,
+      "step": 2164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.776608051827858e-05,
+      "loss": 0.0058,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7760296159185566e-05,
+      "loss": 0.3859,
+      "step": 2166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.775451180009255e-05,
+      "loss": 0.0022,
+      "step": 2167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.774872744099954e-05,
+      "loss": 0.0019,
+      "step": 2168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7742943081906526e-05,
+      "loss": 0.0034,
+      "step": 2169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.773715872281351e-05,
+      "loss": 0.0021,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.77313743637205e-05,
+      "loss": 0.0019,
+      "step": 2171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7725590004627486e-05,
+      "loss": 0.002,
+      "step": 2172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.771980564553448e-05,
+      "loss": 0.001,
+      "step": 2173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7714021286441466e-05,
+      "loss": 0.0031,
+      "step": 2174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.770823692734845e-05,
+      "loss": 0.0025,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.770245256825543e-05,
+      "loss": 0.0036,
+      "step": 2176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.769666820916243e-05,
+      "loss": 0.001,
+      "step": 2177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7690883850069413e-05,
+      "loss": 0.019,
+      "step": 2178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.76850994909764e-05,
+      "loss": 0.3225,
+      "step": 2179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7679315131883394e-05,
+      "loss": 0.0017,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.767353077279038e-05,
+      "loss": 0.0008,
+      "step": 2181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.766774641369736e-05,
+      "loss": 0.0006,
+      "step": 2182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.766196205460435e-05,
+      "loss": 0.0011,
+      "step": 2183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.765617769551134e-05,
+      "loss": 0.0009,
+      "step": 2184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.765039333641833e-05,
+      "loss": 0.0009,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7644608977325314e-05,
+      "loss": 0.2827,
+      "step": 2186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.76388246182323e-05,
+      "loss": 0.0007,
+      "step": 2187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.763304025913929e-05,
+      "loss": 0.0018,
+      "step": 2188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7627255900046275e-05,
+      "loss": 0.0072,
+      "step": 2189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.762147154095326e-05,
+      "loss": 0.0006,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7615687181860255e-05,
+      "loss": 0.0006,
+      "step": 2191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.760990282276724e-05,
+      "loss": 0.0252,
+      "step": 2192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.760411846367423e-05,
+      "loss": 0.0111,
+      "step": 2193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.7598334104581215e-05,
+      "loss": 0.0007,
+      "step": 2194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.75925497454882e-05,
+      "loss": 0.0016,
+      "step": 2195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.758676538639519e-05,
+      "loss": 0.0089,
+      "step": 2196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7580981027302175e-05,
+      "loss": 0.0945,
+      "step": 2197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.757519666820916e-05,
+      "loss": 0.0013,
+      "step": 2198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7569412309116156e-05,
+      "loss": 0.0006,
+      "step": 2199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.756362795002314e-05,
+      "loss": 0.3651,
+      "step": 2200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.755784359093012e-05,
+      "loss": 0.0009,
+      "step": 2201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7552059231837116e-05,
+      "loss": 0.0028,
+      "step": 2202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.75462748727441e-05,
+      "loss": 0.0007,
+      "step": 2203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.754049051365109e-05,
+      "loss": 0.0233,
+      "step": 2204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7534706154558076e-05,
+      "loss": 0.0005,
+      "step": 2205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.752892179546506e-05,
+      "loss": 0.3604,
+      "step": 2206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.752313743637205e-05,
+      "loss": 0.0071,
+      "step": 2207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7517353077279036e-05,
+      "loss": 0.0007,
+      "step": 2208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.751156871818602e-05,
+      "loss": 0.0013,
+      "step": 2209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.750578435909302e-05,
+      "loss": 0.0034,
+      "step": 2210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.0018,
+      "step": 2211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.749421564090699e-05,
+      "loss": 0.315,
+      "step": 2212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.748843128181398e-05,
+      "loss": 0.0013,
+      "step": 2213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7482646922720964e-05,
+      "loss": 0.001,
+      "step": 2214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.747686256362795e-05,
+      "loss": 0.0045,
+      "step": 2215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.747107820453494e-05,
+      "loss": 0.3567,
+      "step": 2216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.746529384544193e-05,
+      "loss": 0.0013,
+      "step": 2217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.745950948634892e-05,
+      "loss": 0.0018,
+      "step": 2218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.74537251272559e-05,
+      "loss": 0.0009,
+      "step": 2219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7447940768162884e-05,
+      "loss": 0.0107,
+      "step": 2220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.744215640906988e-05,
+      "loss": 0.092,
+      "step": 2221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7436372049976865e-05,
+      "loss": 0.003,
+      "step": 2222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.743058769088385e-05,
+      "loss": 0.0026,
+      "step": 2223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7424803331790845e-05,
+      "loss": 0.3424,
+      "step": 2224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7419018972697825e-05,
+      "loss": 0.0018,
+      "step": 2225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.741323461360481e-05,
+      "loss": 0.0011,
+      "step": 2226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.74074502545118e-05,
+      "loss": 0.275,
+      "step": 2227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.740166589541879e-05,
+      "loss": 0.0012,
+      "step": 2228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.739588153632578e-05,
+      "loss": 0.0026,
+      "step": 2229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7390097177232765e-05,
+      "loss": 0.1595,
+      "step": 2230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.738431281813975e-05,
+      "loss": 0.0146,
+      "step": 2231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.737852845904674e-05,
+      "loss": 0.0011,
+      "step": 2232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7372744099953726e-05,
+      "loss": 0.0026,
+      "step": 2233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.736695974086071e-05,
+      "loss": 0.301,
+      "step": 2234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.7361175381767706e-05,
+      "loss": 0.0014,
+      "step": 2235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.735539102267469e-05,
+      "loss": 0.0029,
+      "step": 2236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.734960666358168e-05,
+      "loss": 0.0035,
+      "step": 2237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.734382230448866e-05,
+      "loss": 0.002,
+      "step": 2238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.733803794539565e-05,
+      "loss": 0.0053,
+      "step": 2239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.733225358630264e-05,
+      "loss": 0.0016,
+      "step": 2240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7326469227209627e-05,
+      "loss": 0.0034,
+      "step": 2241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.732068486811661e-05,
+      "loss": 0.0052,
+      "step": 2242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.731490050902361e-05,
+      "loss": 0.0059,
+      "step": 2243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.730911614993059e-05,
+      "loss": 0.0015,
+      "step": 2244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7303331790837574e-05,
+      "loss": 0.0015,
+      "step": 2245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.729754743174457e-05,
+      "loss": 0.0016,
+      "step": 2246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7291763072651554e-05,
+      "loss": 0.0014,
+      "step": 2247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.728597871355854e-05,
+      "loss": 0.0027,
+      "step": 2248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.728019435446553e-05,
+      "loss": 0.0103,
+      "step": 2249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7274409995372514e-05,
+      "loss": 0.0196,
+      "step": 2250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.72686256362795e-05,
+      "loss": 0.0009,
+      "step": 2251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.726284127718649e-05,
+      "loss": 0.0012,
+      "step": 2252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7257056918093474e-05,
+      "loss": 0.0008,
+      "step": 2253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.725127255900047e-05,
+      "loss": 0.0007,
+      "step": 2254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7245488199907455e-05,
+      "loss": 0.0009,
+      "step": 2255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.723970384081444e-05,
+      "loss": 0.001,
+      "step": 2256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.723391948172142e-05,
+      "loss": 0.0005,
+      "step": 2257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7228135122628415e-05,
+      "loss": 0.0005,
+      "step": 2258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.72223507635354e-05,
+      "loss": 0.0005,
+      "step": 2259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.721656640444239e-05,
+      "loss": 0.0012,
+      "step": 2260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.721078204534938e-05,
+      "loss": 0.0015,
+      "step": 2261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.720499768625637e-05,
+      "loss": 0.0004,
+      "step": 2262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.719921332716335e-05,
+      "loss": 0.2428,
+      "step": 2263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7193428968070335e-05,
+      "loss": 0.0007,
+      "step": 2264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.718764460897733e-05,
+      "loss": 0.0005,
+      "step": 2265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7181860249884316e-05,
+      "loss": 0.0007,
+      "step": 2266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.71760758907913e-05,
+      "loss": 0.0003,
+      "step": 2267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.717029153169829e-05,
+      "loss": 0.0031,
+      "step": 2268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7164507172605276e-05,
+      "loss": 0.001,
+      "step": 2269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.715872281351226e-05,
+      "loss": 0.0005,
+      "step": 2270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.715293845441925e-05,
+      "loss": 0.0003,
+      "step": 2271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.714715409532624e-05,
+      "loss": 0.0005,
+      "step": 2272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.714136973623323e-05,
+      "loss": 0.0006,
+      "step": 2273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7135585377140217e-05,
+      "loss": 0.0016,
+      "step": 2274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.71298010180472e-05,
+      "loss": 0.0009,
+      "step": 2275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.712401665895419e-05,
+      "loss": 0.0014,
+      "step": 2276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.711823229986118e-05,
+      "loss": 0.0002,
+      "step": 2277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7112447940768164e-05,
+      "loss": 0.0174,
+      "step": 2278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.710666358167515e-05,
+      "loss": 0.0019,
+      "step": 2279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7100879222582144e-05,
+      "loss": 0.0012,
+      "step": 2280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7095094863489124e-05,
+      "loss": 0.3031,
+      "step": 2281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.708931050439611e-05,
+      "loss": 0.0005,
+      "step": 2282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7083526145303104e-05,
+      "loss": 0.0003,
+      "step": 2283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.707774178621009e-05,
+      "loss": 0.0004,
+      "step": 2284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.707195742711708e-05,
+      "loss": 0.0021,
+      "step": 2285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7066173068024064e-05,
+      "loss": 0.0003,
+      "step": 2286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.706038870893105e-05,
+      "loss": 0.0003,
+      "step": 2287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.705460434983804e-05,
+      "loss": 0.0753,
+      "step": 2288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7048819990745025e-05,
+      "loss": 0.0003,
+      "step": 2289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.704303563165201e-05,
+      "loss": 0.2286,
+      "step": 2290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7037251272559005e-05,
+      "loss": 0.4431,
+      "step": 2291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.703146691346599e-05,
+      "loss": 0.0283,
+      "step": 2292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.702568255437298e-05,
+      "loss": 0.0006,
+      "step": 2293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7019898195279965e-05,
+      "loss": 0.0003,
+      "step": 2294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.701411383618695e-05,
+      "loss": 0.0004,
+      "step": 2295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.700832947709394e-05,
+      "loss": 0.2364,
+      "step": 2296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7002545118000926e-05,
+      "loss": 0.0003,
+      "step": 2297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.699676075890792e-05,
+      "loss": 0.001,
+      "step": 2298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.6990976399814906e-05,
+      "loss": 0.0006,
+      "step": 2299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.6985192040721886e-05,
+      "loss": 0.0004,
+      "step": 2300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.697940768162887e-05,
+      "loss": 0.0024,
+      "step": 2301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.6973623322535866e-05,
+      "loss": 0.0015,
+      "step": 2302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.696783896344285e-05,
+      "loss": 0.3821,
+      "step": 2303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.696205460434984e-05,
+      "loss": 0.3565,
+      "step": 2304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.695627024525683e-05,
+      "loss": 0.2008,
+      "step": 2305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.695048588616381e-05,
+      "loss": 0.0007,
+      "step": 2306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.69447015270708e-05,
+      "loss": 0.0029,
+      "step": 2307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.693891716797779e-05,
+      "loss": 0.0048,
+      "step": 2308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.693313280888478e-05,
+      "loss": 0.0028,
+      "step": 2309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.692734844979177e-05,
+      "loss": 0.001,
+      "step": 2310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.6921564090698754e-05,
+      "loss": 0.0033,
+      "step": 2311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.691577973160574e-05,
+      "loss": 0.0061,
+      "step": 2312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.690999537251273e-05,
+      "loss": 0.0061,
+      "step": 2313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.6904211013419714e-05,
+      "loss": 0.0112,
+      "step": 2314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.68984266543267e-05,
+      "loss": 0.0109,
+      "step": 2315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.6892642295233694e-05,
+      "loss": 0.0145,
+      "step": 2316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.688685793614068e-05,
+      "loss": 0.0038,
+      "step": 2317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.688107357704767e-05,
+      "loss": 0.0124,
+      "step": 2318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.687528921795465e-05,
+      "loss": 0.0046,
+      "step": 2319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.686950485886164e-05,
+      "loss": 0.0006,
+      "step": 2320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.686372049976863e-05,
+      "loss": 0.5091,
+      "step": 2321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.6857936140675615e-05,
+      "loss": 0.0007,
+      "step": 2322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.68521517815826e-05,
+      "loss": 0.0018,
+      "step": 2323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.684636742248959e-05,
+      "loss": 0.0008,
+      "step": 2324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.6840583063396575e-05,
+      "loss": 0.0006,
+      "step": 2325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.683479870430356e-05,
+      "loss": 0.0004,
+      "step": 2326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6829014345210555e-05,
+      "loss": 0.0012,
+      "step": 2327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.682322998611754e-05,
+      "loss": 0.0006,
+      "step": 2328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.681744562702453e-05,
+      "loss": 0.0005,
+      "step": 2329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6811661267931516e-05,
+      "loss": 0.0005,
+      "step": 2330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.68058769088385e-05,
+      "loss": 0.0007,
+      "step": 2331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.680009254974549e-05,
+      "loss": 0.0006,
+      "step": 2332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6794308190652476e-05,
+      "loss": 0.0005,
+      "step": 2333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.678852383155946e-05,
+      "loss": 0.0002,
+      "step": 2334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6782739472466456e-05,
+      "loss": 0.0004,
+      "step": 2335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.677695511337344e-05,
+      "loss": 0.0005,
+      "step": 2336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.677117075428042e-05,
+      "loss": 0.0004,
+      "step": 2337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.676538639518741e-05,
+      "loss": 0.6871,
+      "step": 2338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.67596020360944e-05,
+      "loss": 0.0006,
+      "step": 2339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.675381767700139e-05,
+      "loss": 0.0007,
+      "step": 2340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.674803331790838e-05,
+      "loss": 0.0004,
+      "step": 2341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.674224895881537e-05,
+      "loss": 0.0007,
+      "step": 2342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.673646459972235e-05,
+      "loss": 0.0002,
+      "step": 2343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.673068024062934e-05,
+      "loss": 0.0035,
+      "step": 2344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6724895881536324e-05,
+      "loss": 0.0002,
+      "step": 2345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.671911152244332e-05,
+      "loss": 0.0006,
+      "step": 2346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6713327163350304e-05,
+      "loss": 0.0005,
+      "step": 2347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.670754280425729e-05,
+      "loss": 0.0316,
+      "step": 2348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.670175844516428e-05,
+      "loss": 0.0007,
+      "step": 2349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6695974086071264e-05,
+      "loss": 0.0005,
+      "step": 2350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.669018972697825e-05,
+      "loss": 0.0004,
+      "step": 2351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.668440536788524e-05,
+      "loss": 0.0004,
+      "step": 2352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.667862100879223e-05,
+      "loss": 0.0003,
+      "step": 2353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.667283664969922e-05,
+      "loss": 0.0003,
+      "step": 2354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6667052290606205e-05,
+      "loss": 0.0003,
+      "step": 2355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6661267931513185e-05,
+      "loss": 0.1332,
+      "step": 2356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.665548357242018e-05,
+      "loss": 0.0002,
+      "step": 2357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6649699213327165e-05,
+      "loss": 0.0006,
+      "step": 2358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.664391485423415e-05,
+      "loss": 0.0002,
+      "step": 2359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.663813049514114e-05,
+      "loss": 0.0003,
+      "step": 2360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.663234613604813e-05,
+      "loss": 0.0004,
+      "step": 2361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.662656177695511e-05,
+      "loss": 0.0002,
+      "step": 2362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.66207774178621e-05,
+      "loss": 0.0004,
+      "step": 2363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.661499305876909e-05,
+      "loss": 0.0005,
+      "step": 2364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.660920869967608e-05,
+      "loss": 0.2099,
+      "step": 2365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6603424340583066e-05,
+      "loss": 0.0264,
+      "step": 2366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.659763998149005e-05,
+      "loss": 0.1036,
+      "step": 2367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.659185562239704e-05,
+      "loss": 0.1849,
+      "step": 2368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6586071263304026e-05,
+      "loss": 0.4458,
+      "step": 2369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.658028690421101e-05,
+      "loss": 0.0005,
+      "step": 2370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6574502545118e-05,
+      "loss": 0.2086,
+      "step": 2371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.656871818602499e-05,
+      "loss": 0.0067,
+      "step": 2372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.656293382693198e-05,
+      "loss": 0.0035,
+      "step": 2373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.655714946783897e-05,
+      "loss": 0.0013,
+      "step": 2374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6551365108745954e-05,
+      "loss": 0.0003,
+      "step": 2375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.654558074965294e-05,
+      "loss": 0.0022,
+      "step": 2376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.653979639055993e-05,
+      "loss": 0.0073,
+      "step": 2377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6534012031466914e-05,
+      "loss": 0.0003,
+      "step": 2378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.652822767237391e-05,
+      "loss": 0.0016,
+      "step": 2379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.652244331328089e-05,
+      "loss": 0.0004,
+      "step": 2380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6516658954187874e-05,
+      "loss": 0.0003,
+      "step": 2381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.651087459509486e-05,
+      "loss": 0.0003,
+      "step": 2382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6505090236001854e-05,
+      "loss": 0.0008,
+      "step": 2383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.649930587690884e-05,
+      "loss": 0.0003,
+      "step": 2384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.649352151781583e-05,
+      "loss": 0.0025,
+      "step": 2385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6487737158722815e-05,
+      "loss": 0.0004,
+      "step": 2386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.64819527996298e-05,
+      "loss": 0.0004,
+      "step": 2387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.647616844053679e-05,
+      "loss": 0.2714,
+      "step": 2388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6470384081443775e-05,
+      "loss": 0.0043,
+      "step": 2389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.646459972235077e-05,
+      "loss": 0.0006,
+      "step": 2390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6458815363257755e-05,
+      "loss": 0.2682,
+      "step": 2391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.645303100416474e-05,
+      "loss": 0.0013,
+      "step": 2392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.644724664507172e-05,
+      "loss": 0.0012,
+      "step": 2393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6441462285978716e-05,
+      "loss": 0.0006,
+      "step": 2394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.64356779268857e-05,
+      "loss": 0.0043,
+      "step": 2395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.642989356779269e-05,
+      "loss": 0.0003,
+      "step": 2396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.642410920869968e-05,
+      "loss": 0.0005,
+      "step": 2397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.641832484960667e-05,
+      "loss": 0.0026,
+      "step": 2398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.641254049051365e-05,
+      "loss": 0.0694,
+      "step": 2399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6406756131420636e-05,
+      "loss": 0.0056,
+      "step": 2400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.640097177232763e-05,
+      "loss": 0.0004,
+      "step": 2401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6395187413234616e-05,
+      "loss": 0.0004,
+      "step": 2402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.63894030541416e-05,
+      "loss": 0.0011,
+      "step": 2403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.638361869504859e-05,
+      "loss": 0.0003,
+      "step": 2404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.637783433595558e-05,
+      "loss": 0.2778,
+      "step": 2405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.637204997686256e-05,
+      "loss": 0.4104,
+      "step": 2406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.636626561776955e-05,
+      "loss": 0.0003,
+      "step": 2407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6360481258676544e-05,
+      "loss": 0.0007,
+      "step": 2408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.635469689958353e-05,
+      "loss": 0.562,
+      "step": 2409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.634891254049052e-05,
+      "loss": 0.0008,
+      "step": 2410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6343128181397504e-05,
+      "loss": 0.2646,
+      "step": 2411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.633734382230449e-05,
+      "loss": 0.0018,
+      "step": 2412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.633155946321148e-05,
+      "loss": 0.0018,
+      "step": 2413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6325775104118464e-05,
+      "loss": 0.0019,
+      "step": 2414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.631999074502545e-05,
+      "loss": 0.0011,
+      "step": 2415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6314206385932445e-05,
+      "loss": 0.0007,
+      "step": 2416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.630842202683943e-05,
+      "loss": 0.2709,
+      "step": 2417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.630263766774641e-05,
+      "loss": 0.0022,
+      "step": 2418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.62968533086534e-05,
+      "loss": 0.0071,
+      "step": 2419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.629106894956039e-05,
+      "loss": 0.0109,
+      "step": 2420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.628528459046738e-05,
+      "loss": 0.0064,
+      "step": 2421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6279500231374365e-05,
+      "loss": 0.0131,
+      "step": 2422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.627371587228136e-05,
+      "loss": 0.0055,
+      "step": 2423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.626793151318834e-05,
+      "loss": 0.0444,
+      "step": 2424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6262147154095325e-05,
+      "loss": 0.0019,
+      "step": 2425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.625636279500231e-05,
+      "loss": 0.006,
+      "step": 2426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6250578435909306e-05,
+      "loss": 0.0077,
+      "step": 2427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.624479407681629e-05,
+      "loss": 0.1485,
+      "step": 2428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.623900971772328e-05,
+      "loss": 0.0028,
+      "step": 2429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6233225358630266e-05,
+      "loss": 0.0051,
+      "step": 2430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.622744099953725e-05,
+      "loss": 0.0028,
+      "step": 2431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.622165664044424e-05,
+      "loss": 0.0242,
+      "step": 2432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6215872281351226e-05,
+      "loss": 0.001,
+      "step": 2433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.621008792225822e-05,
+      "loss": 0.0015,
+      "step": 2434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6204303563165206e-05,
+      "loss": 0.5959,
+      "step": 2435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.619851920407219e-05,
+      "loss": 0.1141,
+      "step": 2436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.619273484497917e-05,
+      "loss": 0.0011,
+      "step": 2437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.618695048588617e-05,
+      "loss": 0.0008,
+      "step": 2438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6181166126793153e-05,
+      "loss": 0.0009,
+      "step": 2439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.617538176770014e-05,
+      "loss": 0.2476,
+      "step": 2440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.616959740860713e-05,
+      "loss": 0.3882,
+      "step": 2441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6163813049514114e-05,
+      "loss": 0.0008,
+      "step": 2442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.61580286904211e-05,
+      "loss": 0.0003,
+      "step": 2443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.615224433132809e-05,
+      "loss": 0.0241,
+      "step": 2444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.614645997223508e-05,
+      "loss": 0.0016,
+      "step": 2445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.614067561314207e-05,
+      "loss": 0.0007,
+      "step": 2446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6134891254049054e-05,
+      "loss": 0.0046,
+      "step": 2447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.612910689495604e-05,
+      "loss": 0.004,
+      "step": 2448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.612332253586303e-05,
+      "loss": 0.0018,
+      "step": 2449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6117538176770015e-05,
+      "loss": 0.0021,
+      "step": 2450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6111753817677e-05,
+      "loss": 0.166,
+      "step": 2451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.610596945858399e-05,
+      "loss": 0.2036,
+      "step": 2452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.610018509949098e-05,
+      "loss": 0.0037,
+      "step": 2453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.609440074039797e-05,
+      "loss": 0.0013,
+      "step": 2454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.608861638130495e-05,
+      "loss": 0.0028,
+      "step": 2455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.608283202221194e-05,
+      "loss": 0.0007,
+      "step": 2456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.607704766311893e-05,
+      "loss": 0.0053,
+      "step": 2457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6071263304025915e-05,
+      "loss": 0.0021,
+      "step": 2458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.60654789449329e-05,
+      "loss": 0.0009,
+      "step": 2459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6059694585839896e-05,
+      "loss": 0.089,
+      "step": 2460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6053910226746876e-05,
+      "loss": 0.0009,
+      "step": 2461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.604812586765386e-05,
+      "loss": 0.0012,
+      "step": 2462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.604234150856085e-05,
+      "loss": 0.0009,
+      "step": 2463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.603655714946784e-05,
+      "loss": 0.0091,
+      "step": 2464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.603077279037483e-05,
+      "loss": 0.0006,
+      "step": 2465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6024988431281816e-05,
+      "loss": 0.0391,
+      "step": 2466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.60192040721888e-05,
+      "loss": 0.051,
+      "step": 2467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.601341971309579e-05,
+      "loss": 0.0011,
+      "step": 2468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6007635354002776e-05,
+      "loss": 0.1313,
+      "step": 2469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.600185099490976e-05,
+      "loss": 0.0011,
+      "step": 2470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.599606663581676e-05,
+      "loss": 0.0008,
+      "step": 2471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.5990282276723744e-05,
+      "loss": 0.003,
+      "step": 2472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.598449791763073e-05,
+      "loss": 0.0556,
+      "step": 2473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.597871355853771e-05,
+      "loss": 0.0005,
+      "step": 2474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.5972929199444704e-05,
+      "loss": 0.0007,
+      "step": 2475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.596714484035169e-05,
+      "loss": 0.0661,
+      "step": 2476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.596136048125868e-05,
+      "loss": 0.0014,
+      "step": 2477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.595557612216567e-05,
+      "loss": 0.002,
+      "step": 2478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.594979176307266e-05,
+      "loss": 0.1893,
+      "step": 2479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.594400740397964e-05,
+      "loss": 0.0005,
+      "step": 2480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.5938223044886624e-05,
+      "loss": 0.2769,
+      "step": 2481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.593243868579362e-05,
+      "loss": 0.0008,
+      "step": 2482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.5926654326700605e-05,
+      "loss": 0.0005,
+      "step": 2483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.592086996760759e-05,
+      "loss": 0.0004,
+      "step": 2484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.591508560851458e-05,
+      "loss": 0.0008,
+      "step": 2485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.5909301249421565e-05,
+      "loss": 0.0015,
+      "step": 2486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.590351689032855e-05,
+      "loss": 0.2464,
+      "step": 2487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.589773253123554e-05,
+      "loss": 0.0005,
+      "step": 2488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.589194817214253e-05,
+      "loss": 0.0004,
+      "step": 2489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.588616381304952e-05,
+      "loss": 0.0053,
+      "step": 2490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.5880379453956505e-05,
+      "loss": 0.3136,
+      "step": 2491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.587459509486349e-05,
+      "loss": 0.0055,
+      "step": 2492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.586881073577048e-05,
+      "loss": 0.0014,
+      "step": 2493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.5863026376677466e-05,
+      "loss": 0.0007,
+      "step": 2494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.585724201758445e-05,
+      "loss": 0.0005,
+      "step": 2495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.585145765849144e-05,
+      "loss": 0.0444,
+      "step": 2496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.584567329939843e-05,
+      "loss": 0.0004,
+      "step": 2497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.583988894030541e-05,
+      "loss": 0.0004,
+      "step": 2498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.58341045812124e-05,
+      "loss": 0.0005,
+      "step": 2499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.582832022211939e-05,
+      "loss": 0.4297,
+      "step": 2500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.582253586302638e-05,
+      "loss": 0.0006,
+      "step": 2501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5816751503933367e-05,
+      "loss": 0.0006,
+      "step": 2502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.581096714484035e-05,
+      "loss": 0.0013,
+      "step": 2503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.580518278574734e-05,
+      "loss": 0.0011,
+      "step": 2504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.579939842665433e-05,
+      "loss": 0.0005,
+      "step": 2505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5793614067561314e-05,
+      "loss": 0.0005,
+      "step": 2506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.57878297084683e-05,
+      "loss": 0.001,
+      "step": 2507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5782045349375294e-05,
+      "loss": 0.0006,
+      "step": 2508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.577626099028228e-05,
+      "loss": 0.0004,
+      "step": 2509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.577047663118927e-05,
+      "loss": 0.0014,
+      "step": 2510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.576469227209625e-05,
+      "loss": 0.0005,
+      "step": 2511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.575890791300324e-05,
+      "loss": 0.0008,
+      "step": 2512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.575312355391023e-05,
+      "loss": 0.0006,
+      "step": 2513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5747339194817214e-05,
+      "loss": 0.0057,
+      "step": 2514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.574155483572421e-05,
+      "loss": 0.0522,
+      "step": 2515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5735770476631195e-05,
+      "loss": 0.2093,
+      "step": 2516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5729986117538175e-05,
+      "loss": 0.0009,
+      "step": 2517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.572420175844516e-05,
+      "loss": 0.0024,
+      "step": 2518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5718417399352155e-05,
+      "loss": 0.0005,
+      "step": 2519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.571263304025914e-05,
+      "loss": 0.0209,
+      "step": 2520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.570684868116613e-05,
+      "loss": 0.0007,
+      "step": 2521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5701064322073115e-05,
+      "loss": 0.0005,
+      "step": 2522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.56952799629801e-05,
+      "loss": 0.0016,
+      "step": 2523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.568949560388709e-05,
+      "loss": 0.0007,
+      "step": 2524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5683711244794076e-05,
+      "loss": 0.0013,
+      "step": 2525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.567792688570107e-05,
+      "loss": 0.0027,
+      "step": 2526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5672142526608056e-05,
+      "loss": 0.4544,
+      "step": 2527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.566635816751504e-05,
+      "loss": 0.3356,
+      "step": 2528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.566057380842203e-05,
+      "loss": 0.0012,
+      "step": 2529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5654789449329016e-05,
+      "loss": 0.0018,
+      "step": 2530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5649005090236e-05,
+      "loss": 0.0008,
+      "step": 2531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.564322073114299e-05,
+      "loss": 0.0004,
+      "step": 2532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5637436372049976e-05,
+      "loss": 0.1647,
+      "step": 2533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.563165201295697e-05,
+      "loss": 0.001,
+      "step": 2534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.562586765386396e-05,
+      "loss": 0.0127,
+      "step": 2535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.562008329477094e-05,
+      "loss": 0.0025,
+      "step": 2536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.561429893567793e-05,
+      "loss": 0.2661,
+      "step": 2537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.560851457658492e-05,
+      "loss": 0.0007,
+      "step": 2538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5602730217491904e-05,
+      "loss": 0.0004,
+      "step": 2539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.559694585839889e-05,
+      "loss": 0.0029,
+      "step": 2540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5591161499305884e-05,
+      "loss": 0.1085,
+      "step": 2541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5585377140212864e-05,
+      "loss": 0.0016,
+      "step": 2542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.557959278111985e-05,
+      "loss": 0.0222,
+      "step": 2543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.557380842202684e-05,
+      "loss": 0.0004,
+      "step": 2544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.556802406293383e-05,
+      "loss": 0.0082,
+      "step": 2545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.556223970384082e-05,
+      "loss": 0.0009,
+      "step": 2546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5556455344747805e-05,
+      "loss": 0.0041,
+      "step": 2547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.555067098565479e-05,
+      "loss": 0.0011,
+      "step": 2548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.554488662656178e-05,
+      "loss": 0.0102,
+      "step": 2549
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5539102267468765e-05,
+      "loss": 0.0004,
+      "step": 2550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.553331790837575e-05,
+      "loss": 0.0011,
+      "step": 2551
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5527533549282745e-05,
+      "loss": 0.4171,
+      "step": 2552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.552174919018973e-05,
+      "loss": 0.0036,
+      "step": 2553
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.551596483109672e-05,
+      "loss": 0.0007,
+      "step": 2554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.55101804720037e-05,
+      "loss": 0.0006,
+      "step": 2555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.550439611291069e-05,
+      "loss": 0.0004,
+      "step": 2556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.549861175381768e-05,
+      "loss": 0.0007,
+      "step": 2557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5492827394724666e-05,
+      "loss": 0.0034,
+      "step": 2558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.548704303563166e-05,
+      "loss": 0.002,
+      "step": 2559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.548125867653864e-05,
+      "loss": 0.0009,
+      "step": 2560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5475474317445626e-05,
+      "loss": 0.0007,
+      "step": 2561
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.546968995835261e-05,
+      "loss": 0.0016,
+      "step": 2562
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5463905599259606e-05,
+      "loss": 0.001,
+      "step": 2563
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.545812124016659e-05,
+      "loss": 0.0008,
+      "step": 2564
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.545233688107358e-05,
+      "loss": 0.2714,
+      "step": 2565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5446552521980566e-05,
+      "loss": 0.0012,
+      "step": 2566
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.544076816288755e-05,
+      "loss": 0.0009,
+      "step": 2567
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.543498380379454e-05,
+      "loss": 0.0254,
+      "step": 2568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.542919944470153e-05,
+      "loss": 0.0016,
+      "step": 2569
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.542341508560852e-05,
+      "loss": 0.0011,
+      "step": 2570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.541763072651551e-05,
+      "loss": 0.0008,
+      "step": 2571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5411846367422494e-05,
+      "loss": 0.0015,
+      "step": 2572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5406062008329474e-05,
+      "loss": 0.0012,
+      "step": 2573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.540027764923647e-05,
+      "loss": 0.0031,
+      "step": 2574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5394493290143454e-05,
+      "loss": 0.0014,
+      "step": 2575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.538870893105044e-05,
+      "loss": 0.0006,
+      "step": 2576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.538292457195743e-05,
+      "loss": 0.004,
+      "step": 2577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.537714021286442e-05,
+      "loss": 0.0015,
+      "step": 2578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.53713558537714e-05,
+      "loss": 0.458,
+      "step": 2579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.536557149467839e-05,
+      "loss": 0.0036,
+      "step": 2580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.535978713558538e-05,
+      "loss": 0.0313,
+      "step": 2581
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.535400277649237e-05,
+      "loss": 0.0004,
+      "step": 2582
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5348218417399355e-05,
+      "loss": 0.261,
+      "step": 2583
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.534243405830634e-05,
+      "loss": 0.0028,
+      "step": 2584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.533664969921333e-05,
+      "loss": 0.0005,
+      "step": 2585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5330865340120315e-05,
+      "loss": 0.0009,
+      "step": 2586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.53250809810273e-05,
+      "loss": 0.0009,
+      "step": 2587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.531929662193429e-05,
+      "loss": 0.0057,
+      "step": 2588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.531351226284128e-05,
+      "loss": 0.001,
+      "step": 2589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.530772790374827e-05,
+      "loss": 0.003,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5301943544655256e-05,
+      "loss": 0.0017,
+      "step": 2591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5296159185562236e-05,
+      "loss": 0.0005,
+      "step": 2592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.529037482646923e-05,
+      "loss": 0.0058,
+      "step": 2593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5284590467376216e-05,
+      "loss": 0.0014,
+      "step": 2594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.52788061082832e-05,
+      "loss": 0.0016,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5273021749190196e-05,
+      "loss": 0.0066,
+      "step": 2596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.526723739009718e-05,
+      "loss": 0.0006,
+      "step": 2597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.526145303100416e-05,
+      "loss": 0.0007,
+      "step": 2598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.525566867191115e-05,
+      "loss": 0.0004,
+      "step": 2599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.524988431281814e-05,
+      "loss": 0.0008,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.524409995372513e-05,
+      "loss": 0.0003,
+      "step": 2601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.523831559463212e-05,
+      "loss": 0.0017,
+      "step": 2602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5232531235539104e-05,
+      "loss": 0.1327,
+      "step": 2603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.522674687644609e-05,
+      "loss": 0.0002,
+      "step": 2604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.522096251735308e-05,
+      "loss": 0.1182,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5215178158260064e-05,
+      "loss": 0.001,
+      "step": 2606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.520939379916706e-05,
+      "loss": 0.0029,
+      "step": 2607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5203609440074044e-05,
+      "loss": 0.5148,
+      "step": 2608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.519782508098103e-05,
+      "loss": 0.0576,
+      "step": 2609
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.519204072188802e-05,
+      "loss": 0.0002,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5186256362795004e-05,
+      "loss": 0.0005,
+      "step": 2611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.518047200370199e-05,
+      "loss": 0.0004,
+      "step": 2612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.517468764460898e-05,
+      "loss": 0.0003,
+      "step": 2613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5168903285515965e-05,
+      "loss": 0.0003,
+      "step": 2614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.516311892642296e-05,
+      "loss": 0.0007,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.515733456732994e-05,
+      "loss": 0.0014,
+      "step": 2616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5151550208236925e-05,
+      "loss": 0.0006,
+      "step": 2617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.514576584914392e-05,
+      "loss": 0.0021,
+      "step": 2618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5139981490050905e-05,
+      "loss": 0.0013,
+      "step": 2619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.513419713095789e-05,
+      "loss": 0.4152,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.512841277186488e-05,
+      "loss": 0.0009,
+      "step": 2621
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5122628412771865e-05,
+      "loss": 0.0006,
+      "step": 2622
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.511684405367885e-05,
+      "loss": 0.0008,
+      "step": 2623
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.511105969458584e-05,
+      "loss": 0.0008,
+      "step": 2624
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5105275335492826e-05,
+      "loss": 0.0033,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.509949097639982e-05,
+      "loss": 0.0012,
+      "step": 2626
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5093706617306806e-05,
+      "loss": 0.2543,
+      "step": 2627
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.508792225821379e-05,
+      "loss": 0.0007,
+      "step": 2628
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.508213789912078e-05,
+      "loss": 0.0023,
+      "step": 2629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5076353540027766e-05,
+      "loss": 0.0437,
+      "step": 2630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.507056918093475e-05,
+      "loss": 0.0009,
+      "step": 2631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.506478482184174e-05,
+      "loss": 0.6772,
+      "step": 2632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.505900046274873e-05,
+      "loss": 0.005,
+      "step": 2633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.505321610365572e-05,
+      "loss": 0.0012,
+      "step": 2634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.50474317445627e-05,
+      "loss": 0.0017,
+      "step": 2635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.504164738546969e-05,
+      "loss": 0.0014,
+      "step": 2636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.503586302637668e-05,
+      "loss": 0.0144,
+      "step": 2637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.503007866728367e-05,
+      "loss": 0.0027,
+      "step": 2638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5024294308190654e-05,
+      "loss": 0.0013,
+      "step": 2639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.501850994909765e-05,
+      "loss": 0.0009,
+      "step": 2640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.501272559000463e-05,
+      "loss": 0.0034,
+      "step": 2641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5006941230911614e-05,
+      "loss": 0.0011,
+      "step": 2642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.50011568718186e-05,
+      "loss": 0.0018,
+      "step": 2643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4995372512725594e-05,
+      "loss": 0.003,
+      "step": 2644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.498958815363258e-05,
+      "loss": 0.0004,
+      "step": 2645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.498380379453957e-05,
+      "loss": 0.1108,
+      "step": 2646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4978019435446555e-05,
+      "loss": 0.0044,
+      "step": 2647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497223507635354e-05,
+      "loss": 0.1077,
+      "step": 2648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.496645071726053e-05,
+      "loss": 0.0008,
+      "step": 2649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4960666358167515e-05,
+      "loss": 0.2997,
+      "step": 2650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.495488199907451e-05,
+      "loss": 0.0008,
+      "step": 2651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4949097639981495e-05,
+      "loss": 0.4109,
+      "step": 2652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.494331328088848e-05,
+      "loss": 0.0014,
+      "step": 2653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.493752892179546e-05,
+      "loss": 0.2278,
+      "step": 2654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4931744562702456e-05,
+      "loss": 0.0494,
+      "step": 2655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.492596020360944e-05,
+      "loss": 0.0013,
+      "step": 2656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.492017584451643e-05,
+      "loss": 0.0019,
+      "step": 2657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4914391485423416e-05,
+      "loss": 0.0033,
+      "step": 2658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.49086071263304e-05,
+      "loss": 0.0095,
+      "step": 2659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.490282276723739e-05,
+      "loss": 0.0085,
+      "step": 2660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4897038408144376e-05,
+      "loss": 0.0046,
+      "step": 2661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.489125404905137e-05,
+      "loss": 0.0066,
+      "step": 2662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4885469689958356e-05,
+      "loss": 0.002,
+      "step": 2663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.487968533086534e-05,
+      "loss": 0.0005,
+      "step": 2664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.487390097177233e-05,
+      "loss": 0.0018,
+      "step": 2665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.486811661267932e-05,
+      "loss": 0.0011,
+      "step": 2666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4862332253586303e-05,
+      "loss": 0.0015,
+      "step": 2667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.485654789449329e-05,
+      "loss": 0.2268,
+      "step": 2668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.485076353540028e-05,
+      "loss": 0.0017,
+      "step": 2669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.484497917630727e-05,
+      "loss": 0.2508,
+      "step": 2670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.483919481721426e-05,
+      "loss": 0.0045,
+      "step": 2671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.483341045812124e-05,
+      "loss": 0.001,
+      "step": 2672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4827626099028224e-05,
+      "loss": 0.0009,
+      "step": 2673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.482184173993522e-05,
+      "loss": 0.0013,
+      "step": 2674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4816057380842204e-05,
+      "loss": 0.0257,
+      "step": 2675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.481027302174919e-05,
+      "loss": 0.1801,
+      "step": 2676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4804488662656185e-05,
+      "loss": 0.0009,
+      "step": 2677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4798704303563165e-05,
+      "loss": 0.0025,
+      "step": 2678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.479291994447015e-05,
+      "loss": 0.0007,
+      "step": 2679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.478713558537714e-05,
+      "loss": 0.0391,
+      "step": 2680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.478135122628413e-05,
+      "loss": 0.0011,
+      "step": 2681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.477556686719112e-05,
+      "loss": 0.0092,
+      "step": 2682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4769782508098105e-05,
+      "loss": 0.0006,
+      "step": 2683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.476399814900509e-05,
+      "loss": 0.0047,
+      "step": 2684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.475821378991208e-05,
+      "loss": 0.0006,
+      "step": 2685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4752429430819065e-05,
+      "loss": 0.0007,
+      "step": 2686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.474664507172605e-05,
+      "loss": 0.0006,
+      "step": 2687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4740860712633046e-05,
+      "loss": 0.0005,
+      "step": 2688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.473507635354003e-05,
+      "loss": 0.0005,
+      "step": 2689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.472929199444702e-05,
+      "loss": 0.0005,
+      "step": 2690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4723507635354e-05,
+      "loss": 0.0007,
+      "step": 2691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.471772327626099e-05,
+      "loss": 0.0011,
+      "step": 2692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.471193891716798e-05,
+      "loss": 0.0005,
+      "step": 2693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4706154558074966e-05,
+      "loss": 0.0761,
+      "step": 2694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.470037019898195e-05,
+      "loss": 0.0006,
+      "step": 2695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4694585839888946e-05,
+      "loss": 0.0007,
+      "step": 2696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4688801480795926e-05,
+      "loss": 0.0011,
+      "step": 2697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.468301712170291e-05,
+      "loss": 0.0005,
+      "step": 2698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.467723276260991e-05,
+      "loss": 0.0006,
+      "step": 2699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4671448403516894e-05,
+      "loss": 0.0004,
+      "step": 2700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.466566404442388e-05,
+      "loss": 0.0211,
+      "step": 2701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.465987968533087e-05,
+      "loss": 0.0004,
+      "step": 2702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4654095326237854e-05,
+      "loss": 0.0005,
+      "step": 2703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.464831096714484e-05,
+      "loss": 0.0013,
+      "step": 2704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.464252660805183e-05,
+      "loss": 0.0007,
+      "step": 2705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4636742248958814e-05,
+      "loss": 0.0012,
+      "step": 2706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.463095788986581e-05,
+      "loss": 0.0135,
+      "step": 2707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4625173530772794e-05,
+      "loss": 0.3716,
+      "step": 2708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.461938917167978e-05,
+      "loss": 0.0003,
+      "step": 2709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.461360481258677e-05,
+      "loss": 0.0019,
+      "step": 2710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4607820453493755e-05,
+      "loss": 0.0004,
+      "step": 2711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.460203609440074e-05,
+      "loss": 0.329,
+      "step": 2712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.459625173530773e-05,
+      "loss": 0.0009,
+      "step": 2713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.459046737621472e-05,
+      "loss": 0.0004,
+      "step": 2714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.458468301712171e-05,
+      "loss": 0.0013,
+      "step": 2715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.457889865802869e-05,
+      "loss": 0.0007,
+      "step": 2716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4573114298935675e-05,
+      "loss": 0.0008,
+      "step": 2717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.456732993984267e-05,
+      "loss": 0.0035,
+      "step": 2718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4561545580749655e-05,
+      "loss": 0.0069,
+      "step": 2719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.455576122165664e-05,
+      "loss": 0.0022,
+      "step": 2720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.454997686256363e-05,
+      "loss": 0.0329,
+      "step": 2721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4544192503470616e-05,
+      "loss": 0.262,
+      "step": 2722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.45384081443776e-05,
+      "loss": 0.0006,
+      "step": 2723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.453262378528459e-05,
+      "loss": 0.0079,
+      "step": 2724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.452683942619158e-05,
+      "loss": 0.0008,
+      "step": 2725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.452105506709857e-05,
+      "loss": 0.0007,
+      "step": 2726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4515270708005556e-05,
+      "loss": 0.0004,
+      "step": 2727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.450948634891254e-05,
+      "loss": 0.0006,
+      "step": 2728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.450370198981953e-05,
+      "loss": 0.0006,
+      "step": 2729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4497917630726517e-05,
+      "loss": 0.0003,
+      "step": 2730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.44921332716335e-05,
+      "loss": 0.5481,
+      "step": 2731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.44863489125405e-05,
+      "loss": 0.0006,
+      "step": 2732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4480564553447484e-05,
+      "loss": 0.0013,
+      "step": 2733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4474780194354464e-05,
+      "loss": 0.0006,
+      "step": 2734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.446899583526145e-05,
+      "loss": 0.0005,
+      "step": 2735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4463211476168444e-05,
+      "loss": 0.0006,
+      "step": 2736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.445742711707543e-05,
+      "loss": 0.0005,
+      "step": 2737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.445164275798242e-05,
+      "loss": 0.0052,
+      "step": 2738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4445858398889404e-05,
+      "loss": 0.1485,
+      "step": 2739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.444007403979639e-05,
+      "loss": 0.0005,
+      "step": 2740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.443428968070338e-05,
+      "loss": 0.0012,
+      "step": 2741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4428505321610364e-05,
+      "loss": 0.004,
+      "step": 2742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.442272096251736e-05,
+      "loss": 0.0011,
+      "step": 2743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4416936603424345e-05,
+      "loss": 0.0005,
+      "step": 2744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.441115224433133e-05,
+      "loss": 0.0013,
+      "step": 2745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.440536788523832e-05,
+      "loss": 0.0012,
+      "step": 2746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4399583526145305e-05,
+      "loss": 0.0021,
+      "step": 2747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.439379916705229e-05,
+      "loss": 0.0006,
+      "step": 2748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.438801480795928e-05,
+      "loss": 0.0118,
+      "step": 2749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4382230448866265e-05,
+      "loss": 0.0009,
+      "step": 2750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.437644608977326e-05,
+      "loss": 0.2571,
+      "step": 2751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4370661730680246e-05,
+      "loss": 0.001,
+      "step": 2752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4364877371587226e-05,
+      "loss": 0.0007,
+      "step": 2753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.435909301249421e-05,
+      "loss": 0.0033,
+      "step": 2754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4353308653401206e-05,
+      "loss": 0.0038,
+      "step": 2755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.434752429430819e-05,
+      "loss": 0.0008,
+      "step": 2756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.434173993521518e-05,
+      "loss": 0.0006,
+      "step": 2757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.433595557612217e-05,
+      "loss": 0.0043,
+      "step": 2758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.433017121702915e-05,
+      "loss": 0.0007,
+      "step": 2759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.432438685793614e-05,
+      "loss": 0.4425,
+      "step": 2760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4318602498843126e-05,
+      "loss": 0.0007,
+      "step": 2761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.431281813975012e-05,
+      "loss": 0.6889,
+      "step": 2762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.430703378065711e-05,
+      "loss": 0.3767,
+      "step": 2763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.430124942156409e-05,
+      "loss": 0.0338,
+      "step": 2764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.429546506247108e-05,
+      "loss": 0.0012,
+      "step": 2765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.428968070337807e-05,
+      "loss": 0.0018,
+      "step": 2766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4283896344285054e-05,
+      "loss": 0.0027,
+      "step": 2767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.427811198519204e-05,
+      "loss": 0.0026,
+      "step": 2768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4272327626099034e-05,
+      "loss": 0.0084,
+      "step": 2769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.426654326700602e-05,
+      "loss": 0.0052,
+      "step": 2770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.426075890791301e-05,
+      "loss": 0.0068,
+      "step": 2771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.425497454881999e-05,
+      "loss": 0.0011,
+      "step": 2772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.424919018972698e-05,
+      "loss": 0.1723,
+      "step": 2773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.424340583063397e-05,
+      "loss": 0.0038,
+      "step": 2774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4237621471540954e-05,
+      "loss": 0.0037,
+      "step": 2775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.423183711244794e-05,
+      "loss": 0.0031,
+      "step": 2776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.422605275335493e-05,
+      "loss": 0.0039,
+      "step": 2777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4220268394261915e-05,
+      "loss": 0.0035,
+      "step": 2778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.42144840351689e-05,
+      "loss": 0.3212,
+      "step": 2779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4208699676075895e-05,
+      "loss": 0.0055,
+      "step": 2780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.420291531698288e-05,
+      "loss": 0.007,
+      "step": 2781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.419713095788987e-05,
+      "loss": 0.0018,
+      "step": 2782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4191346598796855e-05,
+      "loss": 0.0016,
+      "step": 2783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.418556223970384e-05,
+      "loss": 0.0107,
+      "step": 2784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.417977788061083e-05,
+      "loss": 0.0016,
+      "step": 2785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4173993521517816e-05,
+      "loss": 0.0013,
+      "step": 2786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.41682091624248e-05,
+      "loss": 0.0024,
+      "step": 2787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4162424803331796e-05,
+      "loss": 0.002,
+      "step": 2788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.415664044423878e-05,
+      "loss": 0.0009,
+      "step": 2789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.415085608514576e-05,
+      "loss": 0.0013,
+      "step": 2790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4145071726052756e-05,
+      "loss": 0.0009,
+      "step": 2791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.413928736695974e-05,
+      "loss": 0.0156,
+      "step": 2792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.413350300786673e-05,
+      "loss": 0.0358,
+      "step": 2793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4127718648773716e-05,
+      "loss": 0.0011,
+      "step": 2794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.412193428968071e-05,
+      "loss": 0.0007,
+      "step": 2795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.411614993058769e-05,
+      "loss": 0.0005,
+      "step": 2796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.411036557149468e-05,
+      "loss": 0.001,
+      "step": 2797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4104581212401663e-05,
+      "loss": 0.0006,
+      "step": 2798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.409879685330866e-05,
+      "loss": 0.0005,
+      "step": 2799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4093012494215644e-05,
+      "loss": 0.001,
+      "step": 2800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.408722813512263e-05,
+      "loss": 0.2248,
+      "step": 2801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.408144377602962e-05,
+      "loss": 0.0005,
+      "step": 2802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4075659416936604e-05,
+      "loss": 0.2936,
+      "step": 2803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.406987505784359e-05,
+      "loss": 0.0005,
+      "step": 2804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.406409069875058e-05,
+      "loss": 0.0007,
+      "step": 2805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.405830633965757e-05,
+      "loss": 0.0009,
+      "step": 2806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.405252198056456e-05,
+      "loss": 0.0027,
+      "step": 2807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4046737621471545e-05,
+      "loss": 0.0012,
+      "step": 2808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4040953262378525e-05,
+      "loss": 0.0012,
+      "step": 2809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.403516890328552e-05,
+      "loss": 0.4305,
+      "step": 2810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4029384544192505e-05,
+      "loss": 0.001,
+      "step": 2811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.402360018509949e-05,
+      "loss": 0.0005,
+      "step": 2812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4017815826006485e-05,
+      "loss": 0.002,
+      "step": 2813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.401203146691347e-05,
+      "loss": 0.0154,
+      "step": 2814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.400624710782045e-05,
+      "loss": 0.0054,
+      "step": 2815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.400046274872744e-05,
+      "loss": 0.0012,
+      "step": 2816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.399467838963443e-05,
+      "loss": 0.0187,
+      "step": 2817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.398889403054142e-05,
+      "loss": 0.0064,
+      "step": 2818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3983109671448406e-05,
+      "loss": 0.0525,
+      "step": 2819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.397732531235539e-05,
+      "loss": 0.0109,
+      "step": 2820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.397154095326238e-05,
+      "loss": 0.0095,
+      "step": 2821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3965756594169366e-05,
+      "loss": 0.0015,
+      "step": 2822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.395997223507635e-05,
+      "loss": 0.0023,
+      "step": 2823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3954187875983346e-05,
+      "loss": 0.0059,
+      "step": 2824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.394840351689033e-05,
+      "loss": 0.0019,
+      "step": 2825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.394261915779732e-05,
+      "loss": 0.0006,
+      "step": 2826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3936834798704307e-05,
+      "loss": 0.0012,
+      "step": 2827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.393105043961129e-05,
+      "loss": 0.2058,
+      "step": 2828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.392526608051828e-05,
+      "loss": 0.0005,
+      "step": 2829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.391948172142527e-05,
+      "loss": 0.0097,
+      "step": 2830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3913697362332254e-05,
+      "loss": 0.0007,
+      "step": 2831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.390791300323925e-05,
+      "loss": 0.0005,
+      "step": 2832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3902128644146234e-05,
+      "loss": 0.0379,
+      "step": 2833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3896344285053214e-05,
+      "loss": 0.0011,
+      "step": 2834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.38905599259602e-05,
+      "loss": 0.0016,
+      "step": 2835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3884775566867194e-05,
+      "loss": 0.0012,
+      "step": 2836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.387899120777418e-05,
+      "loss": 0.0008,
+      "step": 2837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.387320684868117e-05,
+      "loss": 0.0004,
+      "step": 2838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3867422489588154e-05,
+      "loss": 0.0009,
+      "step": 2839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.386163813049514e-05,
+      "loss": 0.0052,
+      "step": 2840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.385585377140213e-05,
+      "loss": 0.0083,
+      "step": 2841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3850069412309115e-05,
+      "loss": 0.0096,
+      "step": 2842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.384428505321611e-05,
+      "loss": 0.0005,
+      "step": 2843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3838500694123095e-05,
+      "loss": 0.0003,
+      "step": 2844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.383271633503008e-05,
+      "loss": 0.001,
+      "step": 2845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.382693197593707e-05,
+      "loss": 0.0005,
+      "step": 2846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3821147616844055e-05,
+      "loss": 0.0004,
+      "step": 2847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.381536325775104e-05,
+      "loss": 0.0004,
+      "step": 2848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.380957889865803e-05,
+      "loss": 0.0004,
+      "step": 2849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.380379453956502e-05,
+      "loss": 0.0004,
+      "step": 2850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.379801018047201e-05,
+      "loss": 0.0002,
+      "step": 2851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.379222582137899e-05,
+      "loss": 0.0008,
+      "step": 2852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3786441462285976e-05,
+      "loss": 0.0006,
+      "step": 2853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.378065710319297e-05,
+      "loss": 0.0002,
+      "step": 2854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3774872744099956e-05,
+      "loss": 0.0003,
+      "step": 2855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.376908838500694e-05,
+      "loss": 0.0013,
+      "step": 2856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.376330402591393e-05,
+      "loss": 0.0004,
+      "step": 2857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3757519666820916e-05,
+      "loss": 0.0003,
+      "step": 2858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.37517353077279e-05,
+      "loss": 0.0003,
+      "step": 2859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.374595094863489e-05,
+      "loss": 0.0004,
+      "step": 2860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.374016658954188e-05,
+      "loss": 0.0025,
+      "step": 2861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.373438223044887e-05,
+      "loss": 0.0009,
+      "step": 2862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.372859787135586e-05,
+      "loss": 0.0013,
+      "step": 2863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3722813512262844e-05,
+      "loss": 0.0013,
+      "step": 2864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.371702915316983e-05,
+      "loss": 0.0002,
+      "step": 2865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.371124479407682e-05,
+      "loss": 0.0002,
+      "step": 2866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3705460434983804e-05,
+      "loss": 0.0002,
+      "step": 2867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.369967607589079e-05,
+      "loss": 0.5153,
+      "step": 2868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3693891716797784e-05,
+      "loss": 0.0003,
+      "step": 2869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.368810735770477e-05,
+      "loss": 0.0002,
+      "step": 2870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.368232299861175e-05,
+      "loss": 0.0002,
+      "step": 2871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3676538639518744e-05,
+      "loss": 0.001,
+      "step": 2872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.367075428042573e-05,
+      "loss": 0.0003,
+      "step": 2873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.366496992133272e-05,
+      "loss": 0.0007,
+      "step": 2874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3659185562239705e-05,
+      "loss": 0.0003,
+      "step": 2875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.36534012031467e-05,
+      "loss": 0.0002,
+      "step": 2876
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.364761684405368e-05,
+      "loss": 0.0011,
+      "step": 2877
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3641832484960665e-05,
+      "loss": 0.0063,
+      "step": 2878
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.363604812586765e-05,
+      "loss": 0.0003,
+      "step": 2879
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3630263766774645e-05,
+      "loss": 0.0004,
+      "step": 2880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.362447940768163e-05,
+      "loss": 0.4572,
+      "step": 2881
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.361869504858862e-05,
+      "loss": 0.093,
+      "step": 2882
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3612910689495606e-05,
+      "loss": 0.0006,
+      "step": 2883
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.360712633040259e-05,
+      "loss": 0.0004,
+      "step": 2884
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.360134197130958e-05,
+      "loss": 0.0012,
+      "step": 2885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3595557612216566e-05,
+      "loss": 0.0005,
+      "step": 2886
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.358977325312356e-05,
+      "loss": 0.0004,
+      "step": 2887
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3583988894030546e-05,
+      "loss": 0.0005,
+      "step": 2888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.357820453493753e-05,
+      "loss": 0.0004,
+      "step": 2889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.357242017584451e-05,
+      "loss": 0.0004,
+      "step": 2890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3566635816751506e-05,
+      "loss": 0.0006,
+      "step": 2891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.356085145765849e-05,
+      "loss": 0.0006,
+      "step": 2892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.355506709856548e-05,
+      "loss": 0.0242,
+      "step": 2893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3549282739472473e-05,
+      "loss": 0.0004,
+      "step": 2894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3543498380379453e-05,
+      "loss": 0.0007,
+      "step": 2895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.353771402128644e-05,
+      "loss": 0.0008,
+      "step": 2896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.353192966219343e-05,
+      "loss": 0.2905,
+      "step": 2897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.352614530310042e-05,
+      "loss": 0.0013,
+      "step": 2898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.352036094400741e-05,
+      "loss": 0.0014,
+      "step": 2899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3514576584914394e-05,
+      "loss": 0.0007,
+      "step": 2900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.350879222582138e-05,
+      "loss": 0.0009,
+      "step": 2901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.350300786672837e-05,
+      "loss": 0.0007,
+      "step": 2902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3497223507635354e-05,
+      "loss": 0.0028,
+      "step": 2903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.349143914854234e-05,
+      "loss": 0.0087,
+      "step": 2904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3485654789449335e-05,
+      "loss": 0.0005,
+      "step": 2905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.347987043035632e-05,
+      "loss": 0.0011,
+      "step": 2906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.347408607126331e-05,
+      "loss": 0.0004,
+      "step": 2907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.346830171217029e-05,
+      "loss": 0.0006,
+      "step": 2908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.346251735307728e-05,
+      "loss": 0.0003,
+      "step": 2909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.345673299398427e-05,
+      "loss": 0.0006,
+      "step": 2910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3450948634891255e-05,
+      "loss": 0.3367,
+      "step": 2911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.344516427579824e-05,
+      "loss": 0.0011,
+      "step": 2912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3439379916705235e-05,
+      "loss": 0.0023,
+      "step": 2913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3433595557612215e-05,
+      "loss": 0.0008,
+      "step": 2914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.34278111985192e-05,
+      "loss": 0.0019,
+      "step": 2915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3422026839426196e-05,
+      "loss": 0.0024,
+      "step": 2916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.341624248033318e-05,
+      "loss": 0.0005,
+      "step": 2917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.341045812124017e-05,
+      "loss": 0.0036,
+      "step": 2918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3404673762147156e-05,
+      "loss": 0.0007,
+      "step": 2919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.339888940305414e-05,
+      "loss": 0.0007,
+      "step": 2920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.339310504396113e-05,
+      "loss": 0.0004,
+      "step": 2921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3387320684868116e-05,
+      "loss": 0.0013,
+      "step": 2922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.33815363257751e-05,
+      "loss": 0.0008,
+      "step": 2923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3375751966682096e-05,
+      "loss": 0.0012,
+      "step": 2924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.336996760758908e-05,
+      "loss": 0.0014,
+      "step": 2925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.336418324849607e-05,
+      "loss": 0.0012,
+      "step": 2926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.335839888940305e-05,
+      "loss": 0.0003,
+      "step": 2927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3352614530310043e-05,
+      "loss": 0.0023,
+      "step": 2928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.334683017121703e-05,
+      "loss": 0.0313,
+      "step": 2929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.334104581212402e-05,
+      "loss": 0.0008,
+      "step": 2930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.333526145303101e-05,
+      "loss": 0.0036,
+      "step": 2931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3329477093938e-05,
+      "loss": 0.3495,
+      "step": 2932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.332369273484498e-05,
+      "loss": 0.0005,
+      "step": 2933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3317908375751964e-05,
+      "loss": 0.0005,
+      "step": 2934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.331212401665896e-05,
+      "loss": 0.004,
+      "step": 2935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3306339657565944e-05,
+      "loss": 0.2865,
+      "step": 2936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.330055529847293e-05,
+      "loss": 0.0023,
+      "step": 2937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.329477093937992e-05,
+      "loss": 0.0007,
+      "step": 2938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3288986580286905e-05,
+      "loss": 0.0002,
+      "step": 2939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.328320222119389e-05,
+      "loss": 0.0004,
+      "step": 2940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.327741786210088e-05,
+      "loss": 0.0004,
+      "step": 2941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.327163350300787e-05,
+      "loss": 0.0003,
+      "step": 2942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.326584914391486e-05,
+      "loss": 0.0014,
+      "step": 2943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3260064784821845e-05,
+      "loss": 0.0008,
+      "step": 2944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.325428042572883e-05,
+      "loss": 0.0013,
+      "step": 2945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.324849606663582e-05,
+      "loss": 0.0004,
+      "step": 2946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3242711707542805e-05,
+      "loss": 0.0002,
+      "step": 2947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.323692734844979e-05,
+      "loss": 0.0022,
+      "step": 2948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.323114298935678e-05,
+      "loss": 0.0035,
+      "step": 2949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.322535863026377e-05,
+      "loss": 0.0011,
+      "step": 2950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.321957427117075e-05,
+      "loss": 0.0003,
+      "step": 2951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.321378991207774e-05,
+      "loss": 0.001,
+      "step": 2952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.320800555298473e-05,
+      "loss": 0.0005,
+      "step": 2953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.320222119389172e-05,
+      "loss": 0.0003,
+      "step": 2954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3196436834798706e-05,
+      "loss": 0.0003,
+      "step": 2955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.319065247570569e-05,
+      "loss": 0.0004,
+      "step": 2956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.318486811661268e-05,
+      "loss": 0.0709,
+      "step": 2957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3179083757519667e-05,
+      "loss": 0.0026,
+      "step": 2958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.317329939842665e-05,
+      "loss": 0.0003,
+      "step": 2959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.316751503933364e-05,
+      "loss": 0.0009,
+      "step": 2960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3161730680240634e-05,
+      "loss": 0.0003,
+      "step": 2961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.315594632114762e-05,
+      "loss": 0.1508,
+      "step": 2962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.315016196205461e-05,
+      "loss": 0.001,
+      "step": 2963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3144377602961594e-05,
+      "loss": 0.0006,
+      "step": 2964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.313859324386858e-05,
+      "loss": 0.0004,
+      "step": 2965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.313280888477557e-05,
+      "loss": 0.0053,
+      "step": 2966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3127024525682554e-05,
+      "loss": 0.0002,
+      "step": 2967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.312124016658955e-05,
+      "loss": 0.0006,
+      "step": 2968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3115455807496534e-05,
+      "loss": 0.0033,
+      "step": 2969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3109671448403514e-05,
+      "loss": 0.0015,
+      "step": 2970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.31038870893105e-05,
+      "loss": 0.0011,
+      "step": 2971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3098102730217495e-05,
+      "loss": 0.0018,
+      "step": 2972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.309231837112448e-05,
+      "loss": 0.0006,
+      "step": 2973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.308653401203147e-05,
+      "loss": 0.0004,
+      "step": 2974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.308074965293846e-05,
+      "loss": 0.0006,
+      "step": 2975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.307496529384544e-05,
+      "loss": 0.2597,
+      "step": 2976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.306918093475243e-05,
+      "loss": 0.0002,
+      "step": 2977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.3063396575659415e-05,
+      "loss": 0.0007,
+      "step": 2978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.305761221656641e-05,
+      "loss": 0.5603,
+      "step": 2979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.3051827857473395e-05,
+      "loss": 0.0008,
+      "step": 2980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.304604349838038e-05,
+      "loss": 0.0013,
+      "step": 2981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.304025913928737e-05,
+      "loss": 0.2402,
+      "step": 2982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.3034474780194356e-05,
+      "loss": 0.0038,
+      "step": 2983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.302869042110134e-05,
+      "loss": 0.0012,
+      "step": 2984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.302290606200833e-05,
+      "loss": 0.0002,
+      "step": 2985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.301712170291532e-05,
+      "loss": 0.0005,
+      "step": 2986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.301133734382231e-05,
+      "loss": 0.0082,
+      "step": 2987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.3005552984729296e-05,
+      "loss": 0.0009,
+      "step": 2988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2999768625636276e-05,
+      "loss": 0.0012,
+      "step": 2989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.299398426654327e-05,
+      "loss": 0.0013,
+      "step": 2990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2988199907450257e-05,
+      "loss": 0.0003,
+      "step": 2991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.298241554835724e-05,
+      "loss": 0.0011,
+      "step": 2992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.297663118926423e-05,
+      "loss": 0.0005,
+      "step": 2993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2970846830171224e-05,
+      "loss": 0.0004,
+      "step": 2994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2965062471078204e-05,
+      "loss": 0.0007,
+      "step": 2995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.295927811198519e-05,
+      "loss": 0.002,
+      "step": 2996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2953493752892184e-05,
+      "loss": 0.0056,
+      "step": 2997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.294770939379917e-05,
+      "loss": 0.0004,
+      "step": 2998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.294192503470616e-05,
+      "loss": 0.0005,
+      "step": 2999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2936140675613144e-05,
+      "loss": 0.0002,
+      "step": 3000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.293035631652013e-05,
+      "loss": 0.2094,
+      "step": 3001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.292457195742712e-05,
+      "loss": 0.0002,
+      "step": 3002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2918787598334104e-05,
+      "loss": 0.0022,
+      "step": 3003
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.291300323924109e-05,
+      "loss": 0.0008,
+      "step": 3004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2907218880148085e-05,
+      "loss": 0.2892,
+      "step": 3005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.290143452105507e-05,
+      "loss": 0.0248,
+      "step": 3006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.289565016196206e-05,
+      "loss": 0.0003,
+      "step": 3007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.288986580286904e-05,
+      "loss": 0.0251,
+      "step": 3008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.288408144377603e-05,
+      "loss": 0.0006,
+      "step": 3009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.287829708468302e-05,
+      "loss": 0.0004,
+      "step": 3010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2872512725590005e-05,
+      "loss": 0.0011,
+      "step": 3011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2866728366497e-05,
+      "loss": 0.0074,
+      "step": 3012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.286094400740398e-05,
+      "loss": 0.0887,
+      "step": 3013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2855159648310966e-05,
+      "loss": 0.0003,
+      "step": 3014
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.284937528921795e-05,
+      "loss": 0.0003,
+      "step": 3015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2843590930124946e-05,
+      "loss": 0.0007,
+      "step": 3016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.283780657103193e-05,
+      "loss": 0.0008,
+      "step": 3017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.283202221193892e-05,
+      "loss": 0.0002,
+      "step": 3018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2826237852845906e-05,
+      "loss": 0.0002,
+      "step": 3019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.282045349375289e-05,
+      "loss": 0.0007,
+      "step": 3020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.281466913465988e-05,
+      "loss": 0.0004,
+      "step": 3021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2808884775566866e-05,
+      "loss": 0.0002,
+      "step": 3022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.280310041647386e-05,
+      "loss": 0.0002,
+      "step": 3023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.279731605738085e-05,
+      "loss": 0.0003,
+      "step": 3024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2791531698287833e-05,
+      "loss": 0.0002,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2785747339194813e-05,
+      "loss": 0.0002,
+      "step": 3026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.277996298010181e-05,
+      "loss": 0.0002,
+      "step": 3027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2774178621008794e-05,
+      "loss": 0.0003,
+      "step": 3028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.276839426191578e-05,
+      "loss": 0.0002,
+      "step": 3029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.276260990282277e-05,
+      "loss": 0.0005,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.275682554372976e-05,
+      "loss": 0.0003,
+      "step": 3031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.275104118463674e-05,
+      "loss": 0.0002,
+      "step": 3032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.274525682554373e-05,
+      "loss": 0.0002,
+      "step": 3033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.273947246645072e-05,
+      "loss": 0.0436,
+      "step": 3034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.273368810735771e-05,
+      "loss": 0.0004,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2727903748264695e-05,
+      "loss": 0.0009,
+      "step": 3036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.272211938917168e-05,
+      "loss": 0.004,
+      "step": 3037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.271633503007867e-05,
+      "loss": 0.0746,
+      "step": 3038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2710550670985655e-05,
+      "loss": 0.0024,
+      "step": 3039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.270476631189264e-05,
+      "loss": 0.0002,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.269898195279963e-05,
+      "loss": 0.248,
+      "step": 3041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.269319759370662e-05,
+      "loss": 0.0002,
+      "step": 3042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.268741323461361e-05,
+      "loss": 0.001,
+      "step": 3043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2681628875520595e-05,
+      "loss": 0.0004,
+      "step": 3044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.267584451642758e-05,
+      "loss": 0.001,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.267006015733457e-05,
+      "loss": 0.0097,
+      "step": 3046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2664275798241556e-05,
+      "loss": 0.012,
+      "step": 3047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.265849143914854e-05,
+      "loss": 0.0003,
+      "step": 3048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2652707080055536e-05,
+      "loss": 0.0188,
+      "step": 3049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.264692272096252e-05,
+      "loss": 0.0004,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.26411383618695e-05,
+      "loss": 0.2991,
+      "step": 3051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.263535400277649e-05,
+      "loss": 0.0023,
+      "step": 3052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.262956964368348e-05,
+      "loss": 0.001,
+      "step": 3053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.262378528459047e-05,
+      "loss": 0.002,
+      "step": 3054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2618000925497456e-05,
+      "loss": 0.0002,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.261221656640444e-05,
+      "loss": 0.0006,
+      "step": 3056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.260643220731143e-05,
+      "loss": 0.0005,
+      "step": 3057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.260064784821842e-05,
+      "loss": 0.0006,
+      "step": 3058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2594863489125403e-05,
+      "loss": 0.0035,
+      "step": 3059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.25890791300324e-05,
+      "loss": 0.0019,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2583294770939384e-05,
+      "loss": 0.4474,
+      "step": 3061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.257751041184637e-05,
+      "loss": 0.0025,
+      "step": 3062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.257172605275336e-05,
+      "loss": 0.0005,
+      "step": 3063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2565941693660344e-05,
+      "loss": 0.0003,
+      "step": 3064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.256015733456733e-05,
+      "loss": 0.0007,
+      "step": 3065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.255437297547432e-05,
+      "loss": 0.0002,
+      "step": 3066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.254858861638131e-05,
+      "loss": 0.0046,
+      "step": 3067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.25428042572883e-05,
+      "loss": 0.0004,
+      "step": 3068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.253701989819528e-05,
+      "loss": 0.0027,
+      "step": 3069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2531235539102265e-05,
+      "loss": 0.0009,
+      "step": 3070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.252545118000926e-05,
+      "loss": 0.0017,
+      "step": 3071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2519666820916245e-05,
+      "loss": 0.0007,
+      "step": 3072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.251388246182323e-05,
+      "loss": 0.0002,
+      "step": 3073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.250809810273022e-05,
+      "loss": 0.0001,
+      "step": 3074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2502313743637205e-05,
+      "loss": 0.4101,
+      "step": 3075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.249652938454419e-05,
+      "loss": 0.0005,
+      "step": 3076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.249074502545118e-05,
+      "loss": 0.0019,
+      "step": 3077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.248496066635817e-05,
+      "loss": 0.0005,
+      "step": 3078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.247917630726516e-05,
+      "loss": 0.0001,
+      "step": 3079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2473391948172146e-05,
+      "loss": 0.0007,
+      "step": 3080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.246760758907913e-05,
+      "loss": 0.0005,
+      "step": 3081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.246182322998612e-05,
+      "loss": 0.0044,
+      "step": 3082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2456038870893106e-05,
+      "loss": 0.2837,
+      "step": 3083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.245025451180009e-05,
+      "loss": 0.0002,
+      "step": 3084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.244447015270708e-05,
+      "loss": 0.0005,
+      "step": 3085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.243868579361407e-05,
+      "loss": 0.0014,
+      "step": 3086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.243290143452106e-05,
+      "loss": 0.334,
+      "step": 3087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.242711707542804e-05,
+      "loss": 0.0013,
+      "step": 3088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2421332716335027e-05,
+      "loss": 0.2289,
+      "step": 3089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.241554835724202e-05,
+      "loss": 0.0069,
+      "step": 3090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.240976399814901e-05,
+      "loss": 0.002,
+      "step": 3091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2403979639055994e-05,
+      "loss": 0.0039,
+      "step": 3092
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.239819527996299e-05,
+      "loss": 0.2535,
+      "step": 3093
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.239241092086997e-05,
+      "loss": 0.0025,
+      "step": 3094
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2386626561776954e-05,
+      "loss": 0.0003,
+      "step": 3095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.238084220268394e-05,
+      "loss": 0.002,
+      "step": 3096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2375057843590934e-05,
+      "loss": 0.0062,
+      "step": 3097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.236927348449792e-05,
+      "loss": 0.0035,
+      "step": 3098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.236348912540491e-05,
+      "loss": 0.0064,
+      "step": 3099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2357704766311894e-05,
+      "loss": 0.0398,
+      "step": 3100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.235192040721888e-05,
+      "loss": 0.0028,
+      "step": 3101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.234613604812587e-05,
+      "loss": 0.204,
+      "step": 3102
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2340351689032855e-05,
+      "loss": 0.0023,
+      "step": 3103
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.233456732993985e-05,
+      "loss": 0.0007,
+      "step": 3104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2328782970846835e-05,
+      "loss": 0.0006,
+      "step": 3105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.232299861175382e-05,
+      "loss": 0.0015,
+      "step": 3106
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.23172142526608e-05,
+      "loss": 0.0629,
+      "step": 3107
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2311429893567795e-05,
+      "loss": 0.2115,
+      "step": 3108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.230564553447478e-05,
+      "loss": 0.0093,
+      "step": 3109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.229986117538177e-05,
+      "loss": 0.0382,
+      "step": 3110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2294076816288756e-05,
+      "loss": 0.0007,
+      "step": 3111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.228829245719575e-05,
+      "loss": 0.0004,
+      "step": 3112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.228250809810273e-05,
+      "loss": 0.0008,
+      "step": 3113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2276723739009716e-05,
+      "loss": 0.0007,
+      "step": 3114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.227093937991671e-05,
+      "loss": 0.1062,
+      "step": 3115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2265155020823696e-05,
+      "loss": 0.002,
+      "step": 3116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.225937066173068e-05,
+      "loss": 0.0003,
+      "step": 3117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.225358630263767e-05,
+      "loss": 0.0003,
+      "step": 3118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2247801943544656e-05,
+      "loss": 0.0011,
+      "step": 3119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.224201758445164e-05,
+      "loss": 0.0013,
+      "step": 3120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.223623322535863e-05,
+      "loss": 0.0006,
+      "step": 3121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.223044886626562e-05,
+      "loss": 0.0007,
+      "step": 3122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.222466450717261e-05,
+      "loss": 0.003,
+      "step": 3123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.22188801480796e-05,
+      "loss": 0.0014,
+      "step": 3124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2213095788986584e-05,
+      "loss": 0.0008,
+      "step": 3125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.220731142989357e-05,
+      "loss": 0.0004,
+      "step": 3126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.220152707080056e-05,
+      "loss": 0.0024,
+      "step": 3127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2195742711707544e-05,
+      "loss": 0.0002,
+      "step": 3128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.218995835261453e-05,
+      "loss": 0.0023,
+      "step": 3129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2184173993521524e-05,
+      "loss": 0.0019,
+      "step": 3130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2178389634428504e-05,
+      "loss": 0.1374,
+      "step": 3131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.217260527533549e-05,
+      "loss": 0.0006,
+      "step": 3132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.216682091624248e-05,
+      "loss": 0.0004,
+      "step": 3133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.216103655714947e-05,
+      "loss": 0.0019,
+      "step": 3134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.215525219805646e-05,
+      "loss": 0.0003,
+      "step": 3135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2149467838963445e-05,
+      "loss": 0.7092,
+      "step": 3136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.214368347987043e-05,
+      "loss": 0.0011,
+      "step": 3137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.213789912077742e-05,
+      "loss": 0.0004,
+      "step": 3138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2132114761684405e-05,
+      "loss": 0.0006,
+      "step": 3139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.212633040259139e-05,
+      "loss": 0.0012,
+      "step": 3140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2120546043498385e-05,
+      "loss": 0.0007,
+      "step": 3141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.211476168440537e-05,
+      "loss": 0.0008,
+      "step": 3142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.210897732531236e-05,
+      "loss": 0.0003,
+      "step": 3143
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.210319296621934e-05,
+      "loss": 0.0098,
+      "step": 3144
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.209740860712633e-05,
+      "loss": 0.0007,
+      "step": 3145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.209162424803332e-05,
+      "loss": 0.0002,
+      "step": 3146
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2085839888940306e-05,
+      "loss": 0.0001,
+      "step": 3147
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.20800555298473e-05,
+      "loss": 0.0002,
+      "step": 3148
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2074271170754286e-05,
+      "loss": 0.0002,
+      "step": 3149
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2068486811661266e-05,
+      "loss": 0.0013,
+      "step": 3150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.206270245256825e-05,
+      "loss": 0.0005,
+      "step": 3151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2056918093475246e-05,
+      "loss": 0.0078,
+      "step": 3152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.205113373438223e-05,
+      "loss": 0.2499,
+      "step": 3153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.204534937528922e-05,
+      "loss": 0.3456,
+      "step": 3154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.203956501619621e-05,
+      "loss": 0.0001,
+      "step": 3155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2033780657103193e-05,
+      "loss": 0.0003,
+      "step": 3156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.202799629801018e-05,
+      "loss": 0.0004,
+      "step": 3157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.202221193891717e-05,
+      "loss": 0.0006,
+      "step": 3158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.201642757982416e-05,
+      "loss": 0.0009,
+      "step": 3159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.201064322073115e-05,
+      "loss": 0.0003,
+      "step": 3160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2004858861638134e-05,
+      "loss": 0.1619,
+      "step": 3161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.199907450254512e-05,
+      "loss": 0.0001,
+      "step": 3162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.199329014345211e-05,
+      "loss": 0.0017,
+      "step": 3163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1987505784359094e-05,
+      "loss": 0.0003,
+      "step": 3164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.198172142526608e-05,
+      "loss": 0.0018,
+      "step": 3165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.197593706617307e-05,
+      "loss": 0.0004,
+      "step": 3166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.197015270708006e-05,
+      "loss": 0.0012,
+      "step": 3167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.196436834798705e-05,
+      "loss": 0.0003,
+      "step": 3168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.195858398889403e-05,
+      "loss": 0.0002,
+      "step": 3169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1952799629801015e-05,
+      "loss": 0.0032,
+      "step": 3170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.194701527070801e-05,
+      "loss": 0.0012,
+      "step": 3171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1941230911614995e-05,
+      "loss": 0.0016,
+      "step": 3172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.193544655252198e-05,
+      "loss": 0.0001,
+      "step": 3173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.192966219342897e-05,
+      "loss": 0.0002,
+      "step": 3174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1923877834335955e-05,
+      "loss": 0.0004,
+      "step": 3175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.191809347524294e-05,
+      "loss": 0.1832,
+      "step": 3176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.191230911614993e-05,
+      "loss": 0.0002,
+      "step": 3177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.190652475705692e-05,
+      "loss": 0.0027,
+      "step": 3178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.190074039796391e-05,
+      "loss": 0.0005,
+      "step": 3179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1894956038870896e-05,
+      "loss": 0.0001,
+      "step": 3180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.188917167977788e-05,
+      "loss": 0.0016,
+      "step": 3181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.188338732068487e-05,
+      "loss": 0.0002,
+      "step": 3182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1877602961591856e-05,
+      "loss": 0.0002,
+      "step": 3183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.187181860249884e-05,
+      "loss": 0.0007,
+      "step": 3184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1866034243405837e-05,
+      "loss": 0.0004,
+      "step": 3185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.186024988431282e-05,
+      "loss": 0.0002,
+      "step": 3186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.18544655252198e-05,
+      "loss": 0.0007,
+      "step": 3187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.184868116612679e-05,
+      "loss": 0.3269,
+      "step": 3188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1842896807033784e-05,
+      "loss": 0.0003,
+      "step": 3189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.183711244794077e-05,
+      "loss": 0.0001,
+      "step": 3190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.183132808884776e-05,
+      "loss": 0.0001,
+      "step": 3191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1825543729754744e-05,
+      "loss": 0.0008,
+      "step": 3192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.181975937066173e-05,
+      "loss": 0.2455,
+      "step": 3193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.181397501156872e-05,
+      "loss": 0.0005,
+      "step": 3194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1808190652475704e-05,
+      "loss": 0.0005,
+      "step": 3195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.18024062933827e-05,
+      "loss": 0.0004,
+      "step": 3196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1796621934289684e-05,
+      "loss": 0.1102,
+      "step": 3197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.179083757519667e-05,
+      "loss": 0.2547,
+      "step": 3198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.178505321610366e-05,
+      "loss": 0.0069,
+      "step": 3199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1779268857010645e-05,
+      "loss": 0.0369,
+      "step": 3200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.177348449791763e-05,
+      "loss": 0.0006,
+      "step": 3201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.176770013882462e-05,
+      "loss": 0.0038,
+      "step": 3202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1761915779731605e-05,
+      "loss": 0.0003,
+      "step": 3203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.17561314206386e-05,
+      "loss": 0.0002,
+      "step": 3204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1750347061545585e-05,
+      "loss": 0.0002,
+      "step": 3205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1744562702452565e-05,
+      "loss": 0.0028,
+      "step": 3206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.173877834335956e-05,
+      "loss": 0.0003,
+      "step": 3207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1732993984266545e-05,
+      "loss": 0.113,
+      "step": 3208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.172720962517353e-05,
+      "loss": 0.001,
+      "step": 3209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.172142526608052e-05,
+      "loss": 0.0003,
+      "step": 3210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.171564090698751e-05,
+      "loss": 0.0006,
+      "step": 3211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.170985654789449e-05,
+      "loss": 0.0002,
+      "step": 3212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.170407218880148e-05,
+      "loss": 0.0004,
+      "step": 3213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1698287829708466e-05,
+      "loss": 0.0001,
+      "step": 3214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.169250347061546e-05,
+      "loss": 0.0003,
+      "step": 3215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1686719111522446e-05,
+      "loss": 0.0005,
+      "step": 3216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.168093475242943e-05,
+      "loss": 0.002,
+      "step": 3217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.167515039333642e-05,
+      "loss": 0.0114,
+      "step": 3218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1669366034243407e-05,
+      "loss": 0.0002,
+      "step": 3219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.166358167515039e-05,
+      "loss": 0.0002,
+      "step": 3220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.165779731605738e-05,
+      "loss": 0.0001,
+      "step": 3221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1652012956964374e-05,
+      "loss": 0.0005,
+      "step": 3222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.164622859787136e-05,
+      "loss": 0.0048,
+      "step": 3223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.164044423877835e-05,
+      "loss": 0.0003,
+      "step": 3224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.163465987968533e-05,
+      "loss": 0.0039,
+      "step": 3225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.162887552059232e-05,
+      "loss": 0.6784,
+      "step": 3226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.162309116149931e-05,
+      "loss": 0.0001,
+      "step": 3227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1617306802406294e-05,
+      "loss": 0.0002,
+      "step": 3228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.161152244331329e-05,
+      "loss": 0.0006,
+      "step": 3229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.160573808422027e-05,
+      "loss": 0.0001,
+      "step": 3230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1599953725127254e-05,
+      "loss": 0.137,
+      "step": 3231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.159416936603424e-05,
+      "loss": 0.0022,
+      "step": 3232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1588385006941235e-05,
+      "loss": 0.0001,
+      "step": 3233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.158260064784822e-05,
+      "loss": 0.0006,
+      "step": 3234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.157681628875521e-05,
+      "loss": 0.0001,
+      "step": 3235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1571031929662195e-05,
+      "loss": 0.0001,
+      "step": 3236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.156524757056918e-05,
+      "loss": 0.0002,
+      "step": 3237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.155946321147617e-05,
+      "loss": 0.0002,
+      "step": 3238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1553678852383155e-05,
+      "loss": 0.001,
+      "step": 3239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.154789449329015e-05,
+      "loss": 0.001,
+      "step": 3240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1542110134197136e-05,
+      "loss": 0.0005,
+      "step": 3241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.153632577510412e-05,
+      "loss": 0.1373,
+      "step": 3242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.15305414160111e-05,
+      "loss": 0.0014,
+      "step": 3243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1524757056918096e-05,
+      "loss": 0.0002,
+      "step": 3244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.151897269782508e-05,
+      "loss": 0.0001,
+      "step": 3245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.151318833873207e-05,
+      "loss": 0.3257,
+      "step": 3246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1507403979639056e-05,
+      "loss": 0.0002,
+      "step": 3247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.150161962054605e-05,
+      "loss": 0.0002,
+      "step": 3248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149583526145303e-05,
+      "loss": 0.0001,
+      "step": 3249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1490050902360016e-05,
+      "loss": 0.0648,
+      "step": 3250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1484266543267e-05,
+      "loss": 0.0004,
+      "step": 3251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1478482184174e-05,
+      "loss": 0.1828,
+      "step": 3252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1472697825080983e-05,
+      "loss": 0.1568,
+      "step": 3253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.146691346598797e-05,
+      "loss": 0.0022,
+      "step": 3254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.146112910689496e-05,
+      "loss": 0.0002,
+      "step": 3255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1455344747801944e-05,
+      "loss": 0.8318,
+      "step": 3256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.144956038870893e-05,
+      "loss": 0.002,
+      "step": 3257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.144377602961592e-05,
+      "loss": 0.2965,
+      "step": 3258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.143799167052291e-05,
+      "loss": 0.1796,
+      "step": 3259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.14322073114299e-05,
+      "loss": 0.0003,
+      "step": 3260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1426422952336884e-05,
+      "loss": 0.0011,
+      "step": 3261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1420638593243864e-05,
+      "loss": 0.0278,
+      "step": 3262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141485423415086e-05,
+      "loss": 0.0003,
+      "step": 3263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1409069875057845e-05,
+      "loss": 0.1548,
+      "step": 3264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.140328551596483e-05,
+      "loss": 0.0081,
+      "step": 3265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1397501156871825e-05,
+      "loss": 0.0013,
+      "step": 3266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.139171679777881e-05,
+      "loss": 0.1067,
+      "step": 3267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.138593243868579e-05,
+      "loss": 0.0012,
+      "step": 3268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.138014807959278e-05,
+      "loss": 0.0018,
+      "step": 3269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137436372049977e-05,
+      "loss": 0.0009,
+      "step": 3270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.136857936140676e-05,
+      "loss": 0.0007,
+      "step": 3271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1362795002313745e-05,
+      "loss": 0.004,
+      "step": 3272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135701064322073e-05,
+      "loss": 0.1454,
+      "step": 3273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135122628412772e-05,
+      "loss": 0.0015,
+      "step": 3274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1345441925034706e-05,
+      "loss": 0.0005,
+      "step": 3275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.133965756594169e-05,
+      "loss": 0.0032,
+      "step": 3276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1333873206848686e-05,
+      "loss": 0.0166,
+      "step": 3277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132808884775567e-05,
+      "loss": 0.0076,
+      "step": 3278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132230448866266e-05,
+      "loss": 0.0003,
+      "step": 3279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1316520129569646e-05,
+      "loss": 0.0005,
+      "step": 3280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131073577047663e-05,
+      "loss": 0.0014,
+      "step": 3281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.130495141138362e-05,
+      "loss": 0.0001,
+      "step": 3282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1299167052290606e-05,
+      "loss": 0.0772,
+      "step": 3283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.129338269319759e-05,
+      "loss": 0.007,
+      "step": 3284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.128759833410459e-05,
+      "loss": 0.0001,
+      "step": 3285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1281813975011573e-05,
+      "loss": 0.0026,
+      "step": 3286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1276029615918553e-05,
+      "loss": 0.0012,
+      "step": 3287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.127024525682555e-05,
+      "loss": 0.3031,
+      "step": 3288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1264460897732534e-05,
+      "loss": 0.0001,
+      "step": 3289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.125867653863952e-05,
+      "loss": 0.0235,
+      "step": 3290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.125289217954651e-05,
+      "loss": 0.2858,
+      "step": 3291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1247107820453494e-05,
+      "loss": 0.0002,
+      "step": 3292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.124132346136048e-05,
+      "loss": 0.0001,
+      "step": 3293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.123553910226747e-05,
+      "loss": 0.1505,
+      "step": 3294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1229754743174454e-05,
+      "loss": 0.2227,
+      "step": 3295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.122397038408145e-05,
+      "loss": 0.0007,
+      "step": 3296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1218186024988435e-05,
+      "loss": 0.0071,
+      "step": 3297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.121240166589542e-05,
+      "loss": 0.0011,
+      "step": 3298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.120661730680241e-05,
+      "loss": 0.0002,
+      "step": 3299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1200832947709395e-05,
+      "loss": 0.0102,
+      "step": 3300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.119504858861638e-05,
+      "loss": 0.0003,
+      "step": 3301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.118926422952337e-05,
+      "loss": 0.0018,
+      "step": 3302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.118347987043036e-05,
+      "loss": 0.0002,
+      "step": 3303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.117769551133735e-05,
+      "loss": 0.0007,
+      "step": 3304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.117191115224433e-05,
+      "loss": 0.0012,
+      "step": 3305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1166126793151315e-05,
+      "loss": 0.0001,
+      "step": 3306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.116034243405831e-05,
+      "loss": 0.0015,
+      "step": 3307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1154558074965296e-05,
+      "loss": 0.0002,
+      "step": 3308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.114877371587228e-05,
+      "loss": 0.0001,
+      "step": 3309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1142989356779276e-05,
+      "loss": 0.0001,
+      "step": 3310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1137204997686256e-05,
+      "loss": 0.0002,
+      "step": 3311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.113142063859324e-05,
+      "loss": 0.0002,
+      "step": 3312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.112563627950023e-05,
+      "loss": 0.0001,
+      "step": 3313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.111985192040722e-05,
+      "loss": 0.0129,
+      "step": 3314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.111406756131421e-05,
+      "loss": 0.0002,
+      "step": 3315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1108283202221197e-05,
+      "loss": 0.0001,
+      "step": 3316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.110249884312818e-05,
+      "loss": 0.0002,
+      "step": 3317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.109671448403517e-05,
+      "loss": 0.002,
+      "step": 3318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.109093012494216e-05,
+      "loss": 0.0001,
+      "step": 3319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1085145765849144e-05,
+      "loss": 0.0001,
+      "step": 3320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.107936140675614e-05,
+      "loss": 0.0001,
+      "step": 3321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1073577047663124e-05,
+      "loss": 0.0001,
+      "step": 3322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.106779268857011e-05,
+      "loss": 0.0228,
+      "step": 3323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.106200832947709e-05,
+      "loss": 0.0002,
+      "step": 3324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1056223970384084e-05,
+      "loss": 0.1286,
+      "step": 3325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.105043961129107e-05,
+      "loss": 0.0001,
+      "step": 3326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.104465525219806e-05,
+      "loss": 0.0002,
+      "step": 3327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1038870893105044e-05,
+      "loss": 0.0001,
+      "step": 3328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.103308653401204e-05,
+      "loss": 0.0019,
+      "step": 3329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.102730217491902e-05,
+      "loss": 0.0001,
+      "step": 3330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1021517815826005e-05,
+      "loss": 0.0029,
+      "step": 3331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1015733456733e-05,
+      "loss": 0.0001,
+      "step": 3332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1009949097639985e-05,
+      "loss": 0.0001,
+      "step": 3333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.100416473854697e-05,
+      "loss": 0.0001,
+      "step": 3334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.099838037945396e-05,
+      "loss": 0.0005,
+      "step": 3335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0992596020360945e-05,
+      "loss": 0.0004,
+      "step": 3336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.098681166126793e-05,
+      "loss": 0.0014,
+      "step": 3337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.098102730217492e-05,
+      "loss": 0.0005,
+      "step": 3338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0975242943081905e-05,
+      "loss": 0.002,
+      "step": 3339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.09694585839889e-05,
+      "loss": 0.0001,
+      "step": 3340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0963674224895886e-05,
+      "loss": 0.0001,
+      "step": 3341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.095788986580287e-05,
+      "loss": 0.0008,
+      "step": 3342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.095210550670985e-05,
+      "loss": 0.0005,
+      "step": 3343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0946321147616846e-05,
+      "loss": 0.0001,
+      "step": 3344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.094053678852383e-05,
+      "loss": 0.0001,
+      "step": 3345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.093475242943082e-05,
+      "loss": 0.0001,
+      "step": 3346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.092896807033781e-05,
+      "loss": 0.0001,
+      "step": 3347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.092318371124479e-05,
+      "loss": 0.0001,
+      "step": 3348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.091739935215178e-05,
+      "loss": 0.003,
+      "step": 3349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0911614993058767e-05,
+      "loss": 0.0001,
+      "step": 3350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.090583063396576e-05,
+      "loss": 0.0001,
+      "step": 3351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.090004627487275e-05,
+      "loss": 0.0012,
+      "step": 3352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0894261915779734e-05,
+      "loss": 0.0001,
+      "step": 3353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.088847755668672e-05,
+      "loss": 0.0003,
+      "step": 3354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.088269319759371e-05,
+      "loss": 0.0001,
+      "step": 3355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0876908838500694e-05,
+      "loss": 0.0001,
+      "step": 3356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.087112447940768e-05,
+      "loss": 0.0001,
+      "step": 3357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0865340120314674e-05,
+      "loss": 0.0001,
+      "step": 3358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.085955576122166e-05,
+      "loss": 0.0001,
+      "step": 3359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.085377140212865e-05,
+      "loss": 0.0001,
+      "step": 3360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.084798704303563e-05,
+      "loss": 0.0001,
+      "step": 3361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.084220268394262e-05,
+      "loss": 0.0002,
+      "step": 3362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.083641832484961e-05,
+      "loss": 0.3603,
+      "step": 3363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0830633965756595e-05,
+      "loss": 0.0001,
+      "step": 3364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.082484960666358e-05,
+      "loss": 0.0097,
+      "step": 3365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0819065247570575e-05,
+      "loss": 0.0001,
+      "step": 3366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0813280888477555e-05,
+      "loss": 0.0001,
+      "step": 3367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.080749652938454e-05,
+      "loss": 0.0001,
+      "step": 3368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0801712170291535e-05,
+      "loss": 0.5104,
+      "step": 3369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.079592781119852e-05,
+      "loss": 0.0469,
+      "step": 3370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.079014345210551e-05,
+      "loss": 0.0115,
+      "step": 3371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0784359093012496e-05,
+      "loss": 0.0001,
+      "step": 3372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.077857473391948e-05,
+      "loss": 0.0003,
+      "step": 3373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.077279037482647e-05,
+      "loss": 0.0001,
+      "step": 3374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0767006015733456e-05,
+      "loss": 0.0002,
+      "step": 3375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.076122165664044e-05,
+      "loss": 0.1829,
+      "step": 3376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0755437297547436e-05,
+      "loss": 0.0001,
+      "step": 3377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.074965293845442e-05,
+      "loss": 0.0003,
+      "step": 3378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.074386857936141e-05,
+      "loss": 0.0001,
+      "step": 3379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0738084220268396e-05,
+      "loss": 0.0001,
+      "step": 3380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.073229986117538e-05,
+      "loss": 0.0001,
+      "step": 3381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.072651550208237e-05,
+      "loss": 0.4448,
+      "step": 3382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.072073114298936e-05,
+      "loss": 0.0001,
+      "step": 3383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.071494678389635e-05,
+      "loss": 0.0034,
+      "step": 3384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.070916242480334e-05,
+      "loss": 0.0002,
+      "step": 3385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.070337806571032e-05,
+      "loss": 0.0001,
+      "step": 3386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0697593706617304e-05,
+      "loss": 0.0002,
+      "step": 3387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.06918093475243e-05,
+      "loss": 0.0002,
+      "step": 3388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0686024988431284e-05,
+      "loss": 0.0701,
+      "step": 3389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.068024062933827e-05,
+      "loss": 0.0007,
+      "step": 3390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0674456270245264e-05,
+      "loss": 0.0004,
+      "step": 3391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0668671911152244e-05,
+      "loss": 0.0029,
+      "step": 3392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.066288755205923e-05,
+      "loss": 0.0003,
+      "step": 3393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.065710319296622e-05,
+      "loss": 0.0001,
+      "step": 3394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.065131883387321e-05,
+      "loss": 0.0002,
+      "step": 3395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.06455344747802e-05,
+      "loss": 0.0577,
+      "step": 3396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0639750115687185e-05,
+      "loss": 0.0002,
+      "step": 3397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.063396575659417e-05,
+      "loss": 0.0118,
+      "step": 3398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.062818139750116e-05,
+      "loss": 0.0002,
+      "step": 3399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0622397038408145e-05,
+      "loss": 0.0003,
+      "step": 3400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.061661267931513e-05,
+      "loss": 0.0038,
+      "step": 3401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0610828320222125e-05,
+      "loss": 0.002,
+      "step": 3402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.060504396112911e-05,
+      "loss": 0.0007,
+      "step": 3403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.05992596020361e-05,
+      "loss": 0.0002,
+      "step": 3404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.059347524294308e-05,
+      "loss": 0.0062,
+      "step": 3405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.058769088385007e-05,
+      "loss": 0.0004,
+      "step": 3406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.058190652475706e-05,
+      "loss": 0.0002,
+      "step": 3407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0576122165664046e-05,
+      "loss": 0.0002,
+      "step": 3408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.057033780657103e-05,
+      "loss": 0.6162,
+      "step": 3409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.056455344747802e-05,
+      "loss": 0.0013,
+      "step": 3410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0558769088385006e-05,
+      "loss": 0.0003,
+      "step": 3411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.055298472929199e-05,
+      "loss": 0.0004,
+      "step": 3412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0547200370198986e-05,
+      "loss": 0.0003,
+      "step": 3413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.054141601110597e-05,
+      "loss": 0.0002,
+      "step": 3414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.053563165201296e-05,
+      "loss": 0.0001,
+      "step": 3415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.052984729291995e-05,
+      "loss": 0.0325,
+      "step": 3416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0524062933826934e-05,
+      "loss": 0.0002,
+      "step": 3417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.051827857473392e-05,
+      "loss": 0.0001,
+      "step": 3418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0512494215640907e-05,
+      "loss": 0.0002,
+      "step": 3419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0506709856547894e-05,
+      "loss": 0.0014,
+      "step": 3420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0500925497454884e-05,
+      "loss": 0.0002,
+      "step": 3421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.049514113836187e-05,
+      "loss": 0.0001,
+      "step": 3422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0489356779268857e-05,
+      "loss": 0.0001,
+      "step": 3423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0483572420175844e-05,
+      "loss": 0.0002,
+      "step": 3424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0477788061082834e-05,
+      "loss": 0.0001,
+      "step": 3425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.047200370198982e-05,
+      "loss": 0.0001,
+      "step": 3426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0466219342896808e-05,
+      "loss": 0.0001,
+      "step": 3427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0460434983803798e-05,
+      "loss": 0.0001,
+      "step": 3428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0454650624710785e-05,
+      "loss": 0.0004,
+      "step": 3429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.044886626561777e-05,
+      "loss": 0.0002,
+      "step": 3430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0443081906524755e-05,
+      "loss": 0.001,
+      "step": 3431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.043729754743175e-05,
+      "loss": 0.0011,
+      "step": 3432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0431513188338735e-05,
+      "loss": 0.0005,
+      "step": 3433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.042572882924572e-05,
+      "loss": 0.0001,
+      "step": 3434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0419944470152705e-05,
+      "loss": 0.0014,
+      "step": 3435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.04141601110597e-05,
+      "loss": 0.0001,
+      "step": 3436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0408375751966682e-05,
+      "loss": 0.0001,
+      "step": 3437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.040259139287367e-05,
+      "loss": 0.2882,
+      "step": 3438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0396807033780662e-05,
+      "loss": 0.0003,
+      "step": 3439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0391022674687646e-05,
+      "loss": 0.608,
+      "step": 3440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0385238315594633e-05,
+      "loss": 0.0001,
+      "step": 3441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.037945395650162e-05,
+      "loss": 0.0002,
+      "step": 3442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.037366959740861e-05,
+      "loss": 0.0002,
+      "step": 3443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0367885238315596e-05,
+      "loss": 0.3127,
+      "step": 3444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0362100879222583e-05,
+      "loss": 0.0003,
+      "step": 3445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.035631652012957e-05,
+      "loss": 0.0005,
+      "step": 3446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.035053216103656e-05,
+      "loss": 0.0004,
+      "step": 3447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0344747801943547e-05,
+      "loss": 0.001,
+      "step": 3448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0338963442850533e-05,
+      "loss": 0.0031,
+      "step": 3449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0333179083757524e-05,
+      "loss": 0.0014,
+      "step": 3450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.032739472466451e-05,
+      "loss": 0.0012,
+      "step": 3451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0321610365571497e-05,
+      "loss": 0.0009,
+      "step": 3452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.031582600647848e-05,
+      "loss": 0.0027,
+      "step": 3453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0310041647385474e-05,
+      "loss": 0.0058,
+      "step": 3454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.030425728829246e-05,
+      "loss": 0.0029,
+      "step": 3455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0298472929199444e-05,
+      "loss": 0.0012,
+      "step": 3456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.029268857010643e-05,
+      "loss": 0.0018,
+      "step": 3457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.028690421101342e-05,
+      "loss": 0.0026,
+      "step": 3458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0281119851920408e-05,
+      "loss": 0.0023,
+      "step": 3459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0275335492827395e-05,
+      "loss": 0.0009,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0269551133734385e-05,
+      "loss": 0.0013,
+      "step": 3461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.026376677464137e-05,
+      "loss": 0.0146,
+      "step": 3462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0257982415548358e-05,
+      "loss": 0.0009,
+      "step": 3463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0252198056455345e-05,
+      "loss": 0.0021,
+      "step": 3464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0246413697362335e-05,
+      "loss": 0.0011,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0240629338269322e-05,
+      "loss": 0.0004,
+      "step": 3466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.023484497917631e-05,
+      "loss": 0.0006,
+      "step": 3467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0229060620083295e-05,
+      "loss": 0.0012,
+      "step": 3468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0223276260990286e-05,
+      "loss": 0.0005,
+      "step": 3469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0217491901897272e-05,
+      "loss": 0.0003,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0211707542804256e-05,
+      "loss": 0.0634,
+      "step": 3471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.020592318371125e-05,
+      "loss": 0.0977,
+      "step": 3472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0200138824618236e-05,
+      "loss": 0.0866,
+      "step": 3473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.019435446552522e-05,
+      "loss": 0.0004,
+      "step": 3474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0188570106432206e-05,
+      "loss": 0.0004,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.01827857473392e-05,
+      "loss": 0.0115,
+      "step": 3476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0177001388246183e-05,
+      "loss": 0.0003,
+      "step": 3477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.017121702915317e-05,
+      "loss": 0.0002,
+      "step": 3478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0165432670060156e-05,
+      "loss": 0.0003,
+      "step": 3479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0159648310967147e-05,
+      "loss": 0.0003,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0153863951874133e-05,
+      "loss": 0.409,
+      "step": 3481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.014807959278112e-05,
+      "loss": 0.0003,
+      "step": 3482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.014229523368811e-05,
+      "loss": 0.2433,
+      "step": 3483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0136510874595097e-05,
+      "loss": 0.0002,
+      "step": 3484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0130726515502084e-05,
+      "loss": 0.2212,
+      "step": 3485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.012494215640907e-05,
+      "loss": 0.0003,
+      "step": 3486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.011915779731606e-05,
+      "loss": 0.0003,
+      "step": 3487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0113373438223047e-05,
+      "loss": 0.0005,
+      "step": 3488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0107589079130034e-05,
+      "loss": 0.0005,
+      "step": 3489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0101804720037018e-05,
+      "loss": 0.4989,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.009602036094401e-05,
+      "loss": 0.0003,
+      "step": 3491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0090236001850998e-05,
+      "loss": 0.0005,
+      "step": 3492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.008445164275798e-05,
+      "loss": 0.0007,
+      "step": 3493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0078667283664975e-05,
+      "loss": 0.106,
+      "step": 3494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.007288292457196e-05,
+      "loss": 0.0003,
+      "step": 3495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0067098565478945e-05,
+      "loss": 0.0025,
+      "step": 3496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.006131420638593e-05,
+      "loss": 0.0007,
+      "step": 3497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0055529847292925e-05,
+      "loss": 0.0024,
+      "step": 3498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.004974548819991e-05,
+      "loss": 0.0006,
+      "step": 3499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0043961129106895e-05,
+      "loss": 0.0005,
+      "step": 3500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0038176770013882e-05,
+      "loss": 0.0013,
+      "step": 3501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0032392410920872e-05,
+      "loss": 0.0003,
+      "step": 3502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.002660805182786e-05,
+      "loss": 0.0011,
+      "step": 3503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0020823692734846e-05,
+      "loss": 0.0005,
+      "step": 3504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0015039333641832e-05,
+      "loss": 0.0013,
+      "step": 3505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0009254974548823e-05,
+      "loss": 0.0005,
+      "step": 3506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.000347061545581e-05,
+      "loss": 0.0004,
+      "step": 3507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9997686256362796e-05,
+      "loss": 0.0003,
+      "step": 3508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9991901897269786e-05,
+      "loss": 0.0885,
+      "step": 3509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9986117538176773e-05,
+      "loss": 0.0006,
+      "step": 3510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.998033317908376e-05,
+      "loss": 0.0012,
+      "step": 3511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9974548819990743e-05,
+      "loss": 0.0005,
+      "step": 3512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9968764460897737e-05,
+      "loss": 0.3278,
+      "step": 3513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9962980101804723e-05,
+      "loss": 0.4553,
+      "step": 3514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9957195742711707e-05,
+      "loss": 0.0004,
+      "step": 3515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9951411383618694e-05,
+      "loss": 0.0003,
+      "step": 3516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9945627024525684e-05,
+      "loss": 0.001,
+      "step": 3517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.993984266543267e-05,
+      "loss": 0.2346,
+      "step": 3518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9934058306339657e-05,
+      "loss": 0.0006,
+      "step": 3519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9928273947246647e-05,
+      "loss": 0.0012,
+      "step": 3520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9922489588153634e-05,
+      "loss": 0.0091,
+      "step": 3521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.991670522906062e-05,
+      "loss": 0.0062,
+      "step": 3522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9910920869967608e-05,
+      "loss": 0.0004,
+      "step": 3523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9905136510874598e-05,
+      "loss": 0.0005,
+      "step": 3524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9899352151781585e-05,
+      "loss": 0.0005,
+      "step": 3525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.989356779268857e-05,
+      "loss": 0.0007,
+      "step": 3526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9887783433595558e-05,
+      "loss": 0.0004,
+      "step": 3527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9881999074502548e-05,
+      "loss": 0.1801,
+      "step": 3528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9876214715409535e-05,
+      "loss": 0.0021,
+      "step": 3529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.987043035631652e-05,
+      "loss": 0.0013,
+      "step": 3530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9864645997223512e-05,
+      "loss": 0.0011,
+      "step": 3531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.98588616381305e-05,
+      "loss": 0.0007,
+      "step": 3532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9853077279037482e-05,
+      "loss": 0.0006,
+      "step": 3533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.984729291994447e-05,
+      "loss": 0.001,
+      "step": 3534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9841508560851462e-05,
+      "loss": 0.0014,
+      "step": 3535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9835724201758446e-05,
+      "loss": 0.0004,
+      "step": 3536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9829939842665432e-05,
+      "loss": 0.0005,
+      "step": 3537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.982415548357242e-05,
+      "loss": 0.0007,
+      "step": 3538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.981837112447941e-05,
+      "loss": 0.0004,
+      "step": 3539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9812586765386396e-05,
+      "loss": 0.0007,
+      "step": 3540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9806802406293383e-05,
+      "loss": 0.0009,
+      "step": 3541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.9801018047200373e-05,
+      "loss": 0.0011,
+      "step": 3542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.979523368810736e-05,
+      "loss": 0.0007,
+      "step": 3543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9789449329014346e-05,
+      "loss": 0.0006,
+      "step": 3544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9783664969921333e-05,
+      "loss": 0.016,
+      "step": 3545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9777880610828323e-05,
+      "loss": 0.0024,
+      "step": 3546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.977209625173531e-05,
+      "loss": 0.195,
+      "step": 3547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9766311892642297e-05,
+      "loss": 0.1919,
+      "step": 3548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.976052753354928e-05,
+      "loss": 0.0003,
+      "step": 3549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9754743174456274e-05,
+      "loss": 0.1997,
+      "step": 3550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.974895881536326e-05,
+      "loss": 0.0003,
+      "step": 3551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9743174456270244e-05,
+      "loss": 0.0002,
+      "step": 3552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9737390097177237e-05,
+      "loss": 0.0004,
+      "step": 3553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9731605738084224e-05,
+      "loss": 0.0017,
+      "step": 3554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9725821378991208e-05,
+      "loss": 0.0003,
+      "step": 3555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9720037019898194e-05,
+      "loss": 0.0056,
+      "step": 3556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9714252660805188e-05,
+      "loss": 0.0009,
+      "step": 3557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.970846830171217e-05,
+      "loss": 0.0004,
+      "step": 3558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9702683942619158e-05,
+      "loss": 0.0019,
+      "step": 3559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9696899583526145e-05,
+      "loss": 0.0004,
+      "step": 3560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9691115224433135e-05,
+      "loss": 0.0021,
+      "step": 3561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.968533086534012e-05,
+      "loss": 0.0191,
+      "step": 3562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.967954650624711e-05,
+      "loss": 0.0003,
+      "step": 3563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.96737621471541e-05,
+      "loss": 0.0002,
+      "step": 3564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9667977788061085e-05,
+      "loss": 0.0009,
+      "step": 3565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9662193428968072e-05,
+      "loss": 0.015,
+      "step": 3566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.965640906987506e-05,
+      "loss": 0.0009,
+      "step": 3567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.965062471078205e-05,
+      "loss": 0.0001,
+      "step": 3568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9644840351689036e-05,
+      "loss": 0.0005,
+      "step": 3569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9639055992596022e-05,
+      "loss": 0.0003,
+      "step": 3570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9633271633503006e-05,
+      "loss": 0.265,
+      "step": 3571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.962748727441e-05,
+      "loss": 0.0003,
+      "step": 3572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9621702915316986e-05,
+      "loss": 0.0002,
+      "step": 3573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.961591855622397e-05,
+      "loss": 0.3711,
+      "step": 3574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9610134197130963e-05,
+      "loss": 0.0002,
+      "step": 3575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9604349838037946e-05,
+      "loss": 0.0002,
+      "step": 3576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9598565478944933e-05,
+      "loss": 0.0004,
+      "step": 3577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.959278111985192e-05,
+      "loss": 0.0002,
+      "step": 3578
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.958699676075891e-05,
+      "loss": 0.0003,
+      "step": 3579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9581212401665897e-05,
+      "loss": 0.0006,
+      "step": 3580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9575428042572884e-05,
+      "loss": 0.0002,
+      "step": 3581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.956964368347987e-05,
+      "loss": 0.3254,
+      "step": 3582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.956385932438686e-05,
+      "loss": 0.0009,
+      "step": 3583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9558074965293847e-05,
+      "loss": 0.001,
+      "step": 3584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.9552290606200834e-05,
+      "loss": 0.0001,
+      "step": 3585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.954650624710782e-05,
+      "loss": 0.0006,
+      "step": 3586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.954072188801481e-05,
+      "loss": 0.0018,
+      "step": 3587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9534937528921798e-05,
+      "loss": 0.0002,
+      "step": 3588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.952915316982878e-05,
+      "loss": 0.0003,
+      "step": 3589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9523368810735775e-05,
+      "loss": 0.0008,
+      "step": 3590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.951758445164276e-05,
+      "loss": 0.0002,
+      "step": 3591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9511800092549745e-05,
+      "loss": 0.0006,
+      "step": 3592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.950601573345673e-05,
+      "loss": 0.0002,
+      "step": 3593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9500231374363725e-05,
+      "loss": 0.0003,
+      "step": 3594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.949444701527071e-05,
+      "loss": 0.0002,
+      "step": 3595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9488662656177695e-05,
+      "loss": 0.0449,
+      "step": 3596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9482878297084682e-05,
+      "loss": 0.0003,
+      "step": 3597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9477093937991672e-05,
+      "loss": 0.0005,
+      "step": 3598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.947130957889866e-05,
+      "loss": 0.0003,
+      "step": 3599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9465525219805646e-05,
+      "loss": 0.0014,
+      "step": 3600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9459740860712636e-05,
+      "loss": 0.0055,
+      "step": 3601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9453956501619622e-05,
+      "loss": 0.0004,
+      "step": 3602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.944817214252661e-05,
+      "loss": 0.0006,
+      "step": 3603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9442387783433596e-05,
+      "loss": 0.0002,
+      "step": 3604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9436603424340586e-05,
+      "loss": 0.0102,
+      "step": 3605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9430819065247573e-05,
+      "loss": 0.0098,
+      "step": 3606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.942503470615456e-05,
+      "loss": 0.0008,
+      "step": 3607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9419250347061543e-05,
+      "loss": 0.0008,
+      "step": 3608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9413465987968537e-05,
+      "loss": 0.0001,
+      "step": 3609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9407681628875523e-05,
+      "loss": 0.0002,
+      "step": 3610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9401897269782507e-05,
+      "loss": 0.0016,
+      "step": 3611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.93961129106895e-05,
+      "loss": 0.0004,
+      "step": 3612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9390328551596487e-05,
+      "loss": 0.0018,
+      "step": 3613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.938454419250347e-05,
+      "loss": 0.0002,
+      "step": 3614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9378759833410457e-05,
+      "loss": 0.0005,
+      "step": 3615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.937297547431745e-05,
+      "loss": 0.0003,
+      "step": 3616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9367191115224434e-05,
+      "loss": 0.0001,
+      "step": 3617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.936140675613142e-05,
+      "loss": 0.0003,
+      "step": 3618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9355622397038407e-05,
+      "loss": 0.0017,
+      "step": 3619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9349838037945398e-05,
+      "loss": 0.0001,
+      "step": 3620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9344053678852384e-05,
+      "loss": 0.0016,
+      "step": 3621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.933826931975937e-05,
+      "loss": 0.2271,
+      "step": 3622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.933248496066636e-05,
+      "loss": 0.0002,
+      "step": 3623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9326700601573348e-05,
+      "loss": 0.0008,
+      "step": 3624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9320916242480335e-05,
+      "loss": 0.0001,
+      "step": 3625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.931513188338732e-05,
+      "loss": 0.0002,
+      "step": 3626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.930934752429431e-05,
+      "loss": 0.0002,
+      "step": 3627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.93035631652013e-05,
+      "loss": 0.2724,
+      "step": 3628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9297778806108285e-05,
+      "loss": 0.0001,
+      "step": 3629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.929199444701527e-05,
+      "loss": 0.0002,
+      "step": 3630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9286210087922262e-05,
+      "loss": 0.0004,
+      "step": 3631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.928042572882925e-05,
+      "loss": 0.0001,
+      "step": 3632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9274641369736232e-05,
+      "loss": 0.0001,
+      "step": 3633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9268857010643226e-05,
+      "loss": 0.0021,
+      "step": 3634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.926307265155021e-05,
+      "loss": 0.0001,
+      "step": 3635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9257288292457196e-05,
+      "loss": 0.0004,
+      "step": 3636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9251503933364183e-05,
+      "loss": 0.0001,
+      "step": 3637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9245719574271173e-05,
+      "loss": 0.0001,
+      "step": 3638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.923993521517816e-05,
+      "loss": 0.0001,
+      "step": 3639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9234150856085146e-05,
+      "loss": 0.0002,
+      "step": 3640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9228366496992133e-05,
+      "loss": 0.0002,
+      "step": 3641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9222582137899123e-05,
+      "loss": 0.3655,
+      "step": 3642
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.921679777880611e-05,
+      "loss": 0.0001,
+      "step": 3643
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9211013419713097e-05,
+      "loss": 0.0002,
+      "step": 3644
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9205229060620087e-05,
+      "loss": 0.0001,
+      "step": 3645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9199444701527074e-05,
+      "loss": 0.0004,
+      "step": 3646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.919366034243406e-05,
+      "loss": 0.0003,
+      "step": 3647
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9187875983341044e-05,
+      "loss": 0.0001,
+      "step": 3648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9182091624248037e-05,
+      "loss": 0.0002,
+      "step": 3649
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9176307265155024e-05,
+      "loss": 0.0001,
+      "step": 3650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9170522906062007e-05,
+      "loss": 0.0083,
+      "step": 3651
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9164738546968994e-05,
+      "loss": 0.001,
+      "step": 3652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9158954187875988e-05,
+      "loss": 0.0001,
+      "step": 3653
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.915316982878297e-05,
+      "loss": 0.0002,
+      "step": 3654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9147385469689958e-05,
+      "loss": 0.0001,
+      "step": 3655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.914160111059695e-05,
+      "loss": 0.6001,
+      "step": 3656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9135816751503935e-05,
+      "loss": 0.0014,
+      "step": 3657
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.913003239241092e-05,
+      "loss": 0.0006,
+      "step": 3658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9124248033317908e-05,
+      "loss": 0.0012,
+      "step": 3659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.91184636742249e-05,
+      "loss": 0.0001,
+      "step": 3660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9112679315131885e-05,
+      "loss": 0.0002,
+      "step": 3661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9106894956038872e-05,
+      "loss": 0.0001,
+      "step": 3662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.910111059694586e-05,
+      "loss": 0.3446,
+      "step": 3663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.909532623785285e-05,
+      "loss": 0.0002,
+      "step": 3664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9089541878759836e-05,
+      "loss": 0.0002,
+      "step": 3665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9083757519666822e-05,
+      "loss": 0.0003,
+      "step": 3666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9077973160573806e-05,
+      "loss": 0.0002,
+      "step": 3667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.90721888014808e-05,
+      "loss": 0.0003,
+      "step": 3668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9066404442387786e-05,
+      "loss": 0.0004,
+      "step": 3669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.906062008329477e-05,
+      "loss": 0.0002,
+      "step": 3670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9054835724201763e-05,
+      "loss": 0.2878,
+      "step": 3671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.904905136510875e-05,
+      "loss": 0.0067,
+      "step": 3672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9043267006015733e-05,
+      "loss": 0.0004,
+      "step": 3673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.903748264692272e-05,
+      "loss": 0.0003,
+      "step": 3674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9031698287829713e-05,
+      "loss": 0.0007,
+      "step": 3675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9025913928736697e-05,
+      "loss": 0.0002,
+      "step": 3676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9020129569643683e-05,
+      "loss": 0.0007,
+      "step": 3677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.901434521055067e-05,
+      "loss": 0.0002,
+      "step": 3678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.900856085145766e-05,
+      "loss": 0.0646,
+      "step": 3679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9002776492364647e-05,
+      "loss": 0.0003,
+      "step": 3680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8996992133271634e-05,
+      "loss": 0.1488,
+      "step": 3681
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8991207774178624e-05,
+      "loss": 0.0007,
+      "step": 3682
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.898542341508561e-05,
+      "loss": 0.0004,
+      "step": 3683
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8979639055992597e-05,
+      "loss": 0.3886,
+      "step": 3684
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8973854696899584e-05,
+      "loss": 0.0002,
+      "step": 3685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8968070337806574e-05,
+      "loss": 0.0011,
+      "step": 3686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.896228597871356e-05,
+      "loss": 0.0005,
+      "step": 3687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8956501619620548e-05,
+      "loss": 0.0006,
+      "step": 3688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.895071726052753e-05,
+      "loss": 0.0073,
+      "step": 3689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8944932901434525e-05,
+      "loss": 0.0004,
+      "step": 3690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8939148542341508e-05,
+      "loss": 0.0026,
+      "step": 3691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8933364183248495e-05,
+      "loss": 0.0009,
+      "step": 3692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.892757982415549e-05,
+      "loss": 0.0002,
+      "step": 3693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8921795465062472e-05,
+      "loss": 0.0005,
+      "step": 3694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.891601110596946e-05,
+      "loss": 0.0004,
+      "step": 3695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8910226746876445e-05,
+      "loss": 0.0003,
+      "step": 3696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8904442387783435e-05,
+      "loss": 0.0004,
+      "step": 3697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8898658028690422e-05,
+      "loss": 0.0003,
+      "step": 3698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.889287366959741e-05,
+      "loss": 0.251,
+      "step": 3699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8887089310504396e-05,
+      "loss": 0.0002,
+      "step": 3700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8881304951411386e-05,
+      "loss": 0.0005,
+      "step": 3701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8875520592318373e-05,
+      "loss": 0.0002,
+      "step": 3702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.886973623322536e-05,
+      "loss": 0.0011,
+      "step": 3703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.886395187413235e-05,
+      "loss": 0.0013,
+      "step": 3704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8858167515039336e-05,
+      "loss": 0.0008,
+      "step": 3705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8852383155946323e-05,
+      "loss": 0.0006,
+      "step": 3706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8846598796853306e-05,
+      "loss": 0.0003,
+      "step": 3707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.88408144377603e-05,
+      "loss": 0.0003,
+      "step": 3708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8835030078667287e-05,
+      "loss": 0.0007,
+      "step": 3709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.882924571957427e-05,
+      "loss": 0.0002,
+      "step": 3710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8823461360481257e-05,
+      "loss": 0.0154,
+      "step": 3711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.881767700138825e-05,
+      "loss": 0.001,
+      "step": 3712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8811892642295234e-05,
+      "loss": 0.0003,
+      "step": 3713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.880610828320222e-05,
+      "loss": 0.0009,
+      "step": 3714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8800323924109214e-05,
+      "loss": 0.0122,
+      "step": 3715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8794539565016197e-05,
+      "loss": 0.0003,
+      "step": 3716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8788755205923184e-05,
+      "loss": 0.0002,
+      "step": 3717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.878297084683017e-05,
+      "loss": 0.0005,
+      "step": 3718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.877718648773716e-05,
+      "loss": 0.0003,
+      "step": 3719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8771402128644148e-05,
+      "loss": 0.0008,
+      "step": 3720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8765617769551135e-05,
+      "loss": 0.0003,
+      "step": 3721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.875983341045812e-05,
+      "loss": 0.0011,
+      "step": 3722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.875404905136511e-05,
+      "loss": 0.0003,
+      "step": 3723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8748264692272098e-05,
+      "loss": 0.5071,
+      "step": 3724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8742480333179085e-05,
+      "loss": 0.0002,
+      "step": 3725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8736695974086075e-05,
+      "loss": 0.0001,
+      "step": 3726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8730911614993062e-05,
+      "loss": 0.0001,
+      "step": 3727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.872512725590005e-05,
+      "loss": 0.0008,
+      "step": 3728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8719342896807032e-05,
+      "loss": 0.0007,
+      "step": 3729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8713558537714026e-05,
+      "loss": 0.2856,
+      "step": 3730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8707774178621012e-05,
+      "loss": 0.0006,
+      "step": 3731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8701989819527996e-05,
+      "loss": 0.0003,
+      "step": 3732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8696205460434982e-05,
+      "loss": 0.0005,
+      "step": 3733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8690421101341976e-05,
+      "loss": 0.0007,
+      "step": 3734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.868463674224896e-05,
+      "loss": 0.0008,
+      "step": 3735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8678852383155946e-05,
+      "loss": 0.2996,
+      "step": 3736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8673068024062936e-05,
+      "loss": 0.0004,
+      "step": 3737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8667283664969923e-05,
+      "loss": 0.2518,
+      "step": 3738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.866149930587691e-05,
+      "loss": 0.0041,
+      "step": 3739
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8655714946783897e-05,
+      "loss": 0.0011,
+      "step": 3740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8649930587690887e-05,
+      "loss": 0.0005,
+      "step": 3741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8644146228597873e-05,
+      "loss": 0.0126,
+      "step": 3742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.863836186950486e-05,
+      "loss": 0.3134,
+      "step": 3743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8632577510411847e-05,
+      "loss": 0.0053,
+      "step": 3744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8626793151318837e-05,
+      "loss": 0.0006,
+      "step": 3745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8621008792225824e-05,
+      "loss": 0.0015,
+      "step": 3746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.861522443313281e-05,
+      "loss": 0.0479,
+      "step": 3747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.86094400740398e-05,
+      "loss": 0.1901,
+      "step": 3748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8603655714946788e-05,
+      "loss": 0.0004,
+      "step": 3749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.859787135585377e-05,
+      "loss": 0.001,
+      "step": 3750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8592086996760758e-05,
+      "loss": 0.0009,
+      "step": 3751
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.858630263766775e-05,
+      "loss": 0.0047,
+      "step": 3752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8580518278574735e-05,
+      "loss": 0.0016,
+      "step": 3753
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.857473391948172e-05,
+      "loss": 0.0145,
+      "step": 3754
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8568949560388708e-05,
+      "loss": 0.0015,
+      "step": 3755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8563165201295698e-05,
+      "loss": 0.0012,
+      "step": 3756
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8557380842202685e-05,
+      "loss": 0.0052,
+      "step": 3757
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8551596483109672e-05,
+      "loss": 0.0011,
+      "step": 3758
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.854581212401666e-05,
+      "loss": 0.4027,
+      "step": 3759
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.854002776492365e-05,
+      "loss": 0.1333,
+      "step": 3760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8534243405830635e-05,
+      "loss": 0.0007,
+      "step": 3761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8528459046737622e-05,
+      "loss": 0.0003,
+      "step": 3762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8522674687644612e-05,
+      "loss": 0.1149,
+      "step": 3763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.85168903285516e-05,
+      "loss": 0.0034,
+      "step": 3764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8511105969458586e-05,
+      "loss": 0.0004,
+      "step": 3765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.850532161036557e-05,
+      "loss": 0.0029,
+      "step": 3766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8499537251272563e-05,
+      "loss": 0.0004,
+      "step": 3767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.849375289217955e-05,
+      "loss": 0.0003,
+      "step": 3768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8487968533086533e-05,
+      "loss": 0.0002,
+      "step": 3769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.848218417399352e-05,
+      "loss": 0.0012,
+      "step": 3770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8476399814900513e-05,
+      "loss": 0.0884,
+      "step": 3771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8470615455807496e-05,
+      "loss": 0.0011,
+      "step": 3772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8464831096714483e-05,
+      "loss": 0.0178,
+      "step": 3773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8459046737621477e-05,
+      "loss": 0.0012,
+      "step": 3774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.845326237852846e-05,
+      "loss": 0.0005,
+      "step": 3775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8447478019435447e-05,
+      "loss": 0.0005,
+      "step": 3776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8441693660342434e-05,
+      "loss": 0.0014,
+      "step": 3777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8435909301249424e-05,
+      "loss": 0.0041,
+      "step": 3778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.843012494215641e-05,
+      "loss": 0.0005,
+      "step": 3779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8424340583063397e-05,
+      "loss": 0.0001,
+      "step": 3780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8418556223970384e-05,
+      "loss": 0.0094,
+      "step": 3781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8412771864877374e-05,
+      "loss": 0.0012,
+      "step": 3782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.840698750578436e-05,
+      "loss": 0.1326,
+      "step": 3783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8401203146691348e-05,
+      "loss": 0.002,
+      "step": 3784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8395418787598338e-05,
+      "loss": 0.0031,
+      "step": 3785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8389634428505325e-05,
+      "loss": 0.0006,
+      "step": 3786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.838385006941231e-05,
+      "loss": 0.4177,
+      "step": 3787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8378065710319295e-05,
+      "loss": 0.0002,
+      "step": 3788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8372281351226288e-05,
+      "loss": 0.241,
+      "step": 3789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8366496992133275e-05,
+      "loss": 0.0068,
+      "step": 3790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.836071263304026e-05,
+      "loss": 0.0002,
+      "step": 3791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8354928273947245e-05,
+      "loss": 0.0047,
+      "step": 3792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.834914391485424e-05,
+      "loss": 0.0025,
+      "step": 3793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8343359555761222e-05,
+      "loss": 0.0018,
+      "step": 3794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.833757519666821e-05,
+      "loss": 0.2004,
+      "step": 3795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.83317908375752e-05,
+      "loss": 0.0853,
+      "step": 3796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8326006478482186e-05,
+      "loss": 0.1513,
+      "step": 3797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8320222119389172e-05,
+      "loss": 0.0046,
+      "step": 3798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.831443776029616e-05,
+      "loss": 0.0359,
+      "step": 3799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.830865340120315e-05,
+      "loss": 0.0002,
+      "step": 3800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8302869042110136e-05,
+      "loss": 0.0029,
+      "step": 3801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8297084683017123e-05,
+      "loss": 0.001,
+      "step": 3802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.829130032392411e-05,
+      "loss": 0.0003,
+      "step": 3803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.82855159648311e-05,
+      "loss": 0.1761,
+      "step": 3804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8279731605738087e-05,
+      "loss": 0.0027,
+      "step": 3805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8273947246645073e-05,
+      "loss": 0.002,
+      "step": 3806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8268162887552063e-05,
+      "loss": 0.0029,
+      "step": 3807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.826237852845905e-05,
+      "loss": 0.0009,
+      "step": 3808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8256594169366034e-05,
+      "loss": 0.0003,
+      "step": 3809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.825080981027302e-05,
+      "loss": 0.0028,
+      "step": 3810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8245025451180014e-05,
+      "loss": 0.0008,
+      "step": 3811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8239241092086997e-05,
+      "loss": 0.0118,
+      "step": 3812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8233456732993984e-05,
+      "loss": 0.0015,
+      "step": 3813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.822767237390097e-05,
+      "loss": 0.0002,
+      "step": 3814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.822188801480796e-05,
+      "loss": 0.0217,
+      "step": 3815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8216103655714948e-05,
+      "loss": 0.0015,
+      "step": 3816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8210319296621934e-05,
+      "loss": 0.0003,
+      "step": 3817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8204534937528925e-05,
+      "loss": 0.0438,
+      "step": 3818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.819875057843591e-05,
+      "loss": 0.0006,
+      "step": 3819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8192966219342898e-05,
+      "loss": 0.0003,
+      "step": 3820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8187181860249885e-05,
+      "loss": 0.0011,
+      "step": 3821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8181397501156875e-05,
+      "loss": 0.0058,
+      "step": 3822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8175613142063862e-05,
+      "loss": 0.3457,
+      "step": 3823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.816982878297085e-05,
+      "loss": 0.0001,
+      "step": 3824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8164044423877832e-05,
+      "loss": 0.029,
+      "step": 3825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8158260064784825e-05,
+      "loss": 0.0002,
+      "step": 3826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8152475705691812e-05,
+      "loss": 0.0004,
+      "step": 3827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8146691346598796e-05,
+      "loss": 0.0002,
+      "step": 3828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.814090698750579e-05,
+      "loss": 0.0002,
+      "step": 3829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8135122628412776e-05,
+      "loss": 0.0002,
+      "step": 3830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.812933826931976e-05,
+      "loss": 0.003,
+      "step": 3831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8123553910226746e-05,
+      "loss": 0.0002,
+      "step": 3832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.811776955113374e-05,
+      "loss": 0.4232,
+      "step": 3833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8111985192040723e-05,
+      "loss": 0.2666,
+      "step": 3834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.810620083294771e-05,
+      "loss": 0.0002,
+      "step": 3835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8100416473854696e-05,
+      "loss": 0.0117,
+      "step": 3836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8094632114761686e-05,
+      "loss": 0.0003,
+      "step": 3837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8088847755668673e-05,
+      "loss": 0.0009,
+      "step": 3838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.808306339657566e-05,
+      "loss": 0.0002,
+      "step": 3839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8077279037482647e-05,
+      "loss": 0.0008,
+      "step": 3840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8071494678389637e-05,
+      "loss": 0.1421,
+      "step": 3841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8065710319296624e-05,
+      "loss": 0.0023,
+      "step": 3842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.805992596020361e-05,
+      "loss": 0.0016,
+      "step": 3843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.80541416011106e-05,
+      "loss": 0.0012,
+      "step": 3844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8048357242017587e-05,
+      "loss": 0.349,
+      "step": 3845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8042572882924574e-05,
+      "loss": 0.0032,
+      "step": 3846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8036788523831557e-05,
+      "loss": 0.0015,
+      "step": 3847
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.803100416473855e-05,
+      "loss": 0.0007,
+      "step": 3848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8025219805645538e-05,
+      "loss": 0.007,
+      "step": 3849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.801943544655252e-05,
+      "loss": 0.0121,
+      "step": 3850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8013651087459508e-05,
+      "loss": 0.2024,
+      "step": 3851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.80078667283665e-05,
+      "loss": 0.001,
+      "step": 3852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8002082369273485e-05,
+      "loss": 0.0008,
+      "step": 3853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.799629801018047e-05,
+      "loss": 0.0016,
+      "step": 3854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.799051365108746e-05,
+      "loss": 0.0008,
+      "step": 3855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.798472929199445e-05,
+      "loss": 0.0004,
+      "step": 3856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7978944932901435e-05,
+      "loss": 0.0126,
+      "step": 3857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7973160573808422e-05,
+      "loss": 0.0007,
+      "step": 3858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7967376214715412e-05,
+      "loss": 0.0002,
+      "step": 3859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.79615918556224e-05,
+      "loss": 0.0004,
+      "step": 3860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7955807496529386e-05,
+      "loss": 0.0075,
+      "step": 3861
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7950023137436372e-05,
+      "loss": 0.0002,
+      "step": 3862
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7944238778343362e-05,
+      "loss": 0.0009,
+      "step": 3863
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.793845441925035e-05,
+      "loss": 0.3162,
+      "step": 3864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7932670060157336e-05,
+      "loss": 0.0005,
+      "step": 3865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7926885701064326e-05,
+      "loss": 0.0398,
+      "step": 3866
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7921101341971313e-05,
+      "loss": 0.0063,
+      "step": 3867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7915316982878296e-05,
+      "loss": 0.0001,
+      "step": 3868
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7909532623785283e-05,
+      "loss": 0.0001,
+      "step": 3869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7903748264692277e-05,
+      "loss": 0.0001,
+      "step": 3870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.789796390559926e-05,
+      "loss": 0.0001,
+      "step": 3871
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7892179546506247e-05,
+      "loss": 0.4221,
+      "step": 3872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7886395187413233e-05,
+      "loss": 0.0034,
+      "step": 3873
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7880610828320224e-05,
+      "loss": 0.0002,
+      "step": 3874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.787482646922721e-05,
+      "loss": 0.0003,
+      "step": 3875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7869042110134197e-05,
+      "loss": 0.4368,
+      "step": 3876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7863257751041187e-05,
+      "loss": 0.0002,
+      "step": 3877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7857473391948174e-05,
+      "loss": 0.0007,
+      "step": 3878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.785168903285516e-05,
+      "loss": 0.001,
+      "step": 3879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7845904673762148e-05,
+      "loss": 0.0003,
+      "step": 3880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7840120314669138e-05,
+      "loss": 0.286,
+      "step": 3881
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7834335955576124e-05,
+      "loss": 0.003,
+      "step": 3882
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.782855159648311e-05,
+      "loss": 0.1686,
+      "step": 3883
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7822767237390095e-05,
+      "loss": 0.0005,
+      "step": 3884
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7816982878297088e-05,
+      "loss": 0.0729,
+      "step": 3885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7811198519204075e-05,
+      "loss": 0.1158,
+      "step": 3886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7805414160111058e-05,
+      "loss": 0.0076,
+      "step": 3887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7799629801018052e-05,
+      "loss": 0.3603,
+      "step": 3888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.779384544192504e-05,
+      "loss": 0.0051,
+      "step": 3889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7788061082832022e-05,
+      "loss": 0.0056,
+      "step": 3890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.778227672373901e-05,
+      "loss": 0.0044,
+      "step": 3891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7776492364646002e-05,
+      "loss": 0.1054,
+      "step": 3892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7770708005552986e-05,
+      "loss": 0.0069,
+      "step": 3893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7764923646459972e-05,
+      "loss": 0.0041,
+      "step": 3894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.775913928736696e-05,
+      "loss": 0.0065,
+      "step": 3895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.775335492827395e-05,
+      "loss": 0.1026,
+      "step": 3896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7747570569180936e-05,
+      "loss": 0.0073,
+      "step": 3897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7741786210087923e-05,
+      "loss": 0.0095,
+      "step": 3898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7736001850994913e-05,
+      "loss": 0.0039,
+      "step": 3899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.77302174919019e-05,
+      "loss": 0.0285,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7724433132808886e-05,
+      "loss": 0.0015,
+      "step": 3901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7718648773715873e-05,
+      "loss": 0.001,
+      "step": 3902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7712864414622863e-05,
+      "loss": 0.0149,
+      "step": 3903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.770708005552985e-05,
+      "loss": 0.001,
+      "step": 3904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7701295696436837e-05,
+      "loss": 0.0027,
+      "step": 3905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.769551133734382e-05,
+      "loss": 0.0005,
+      "step": 3906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7689726978250814e-05,
+      "loss": 0.2343,
+      "step": 3907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.76839426191578e-05,
+      "loss": 0.2215,
+      "step": 3908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7678158260064784e-05,
+      "loss": 0.001,
+      "step": 3909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7672373900971777e-05,
+      "loss": 0.001,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7666589541878764e-05,
+      "loss": 0.0011,
+      "step": 3911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7660805182785747e-05,
+      "loss": 0.0002,
+      "step": 3912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7655020823692734e-05,
+      "loss": 0.2176,
+      "step": 3913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7649236464599724e-05,
+      "loss": 0.0005,
+      "step": 3914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.764345210550671e-05,
+      "loss": 0.0004,
+      "step": 3915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7637667746413698e-05,
+      "loss": 0.0005,
+      "step": 3916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7631883387320685e-05,
+      "loss": 0.0004,
+      "step": 3917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7626099028227675e-05,
+      "loss": 0.0005,
+      "step": 3918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.762031466913466e-05,
+      "loss": 0.0004,
+      "step": 3919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7614530310041648e-05,
+      "loss": 0.0003,
+      "step": 3920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7608745950948635e-05,
+      "loss": 0.0006,
+      "step": 3921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7602961591855625e-05,
+      "loss": 0.0056,
+      "step": 3922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7597177232762612e-05,
+      "loss": 0.0002,
+      "step": 3923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.75913928736696e-05,
+      "loss": 0.0006,
+      "step": 3924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.758560851457659e-05,
+      "loss": 0.0024,
+      "step": 3925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7579824155483576e-05,
+      "loss": 0.0004,
+      "step": 3926
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.757403979639056e-05,
+      "loss": 0.0011,
+      "step": 3927
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7568255437297546e-05,
+      "loss": 0.001,
+      "step": 3928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.756247107820454e-05,
+      "loss": 0.0011,
+      "step": 3929
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7556686719111523e-05,
+      "loss": 0.0008,
+      "step": 3930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.755090236001851e-05,
+      "loss": 0.0003,
+      "step": 3931
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7545118000925496e-05,
+      "loss": 0.0234,
+      "step": 3932
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7539333641832486e-05,
+      "loss": 0.0011,
+      "step": 3933
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7533549282739473e-05,
+      "loss": 0.0001,
+      "step": 3934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.752776492364646e-05,
+      "loss": 0.0002,
+      "step": 3935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.752198056455345e-05,
+      "loss": 0.001,
+      "step": 3936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7516196205460437e-05,
+      "loss": 0.0074,
+      "step": 3937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7510411846367423e-05,
+      "loss": 0.0379,
+      "step": 3938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.750462748727441e-05,
+      "loss": 0.0001,
+      "step": 3939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.74988431281814e-05,
+      "loss": 0.0001,
+      "step": 3940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7493058769088387e-05,
+      "loss": 0.0024,
+      "step": 3941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7487274409995374e-05,
+      "loss": 0.0218,
+      "step": 3942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7481490050902357e-05,
+      "loss": 0.0007,
+      "step": 3943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.747570569180935e-05,
+      "loss": 0.0006,
+      "step": 3944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7469921332716338e-05,
+      "loss": 0.0016,
+      "step": 3945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.746413697362332e-05,
+      "loss": 0.0004,
+      "step": 3946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7458352614530314e-05,
+      "loss": 0.0001,
+      "step": 3947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.74525682554373e-05,
+      "loss": 0.0339,
+      "step": 3948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7446783896344285e-05,
+      "loss": 0.0002,
+      "step": 3949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.744099953725127e-05,
+      "loss": 0.0003,
+      "step": 3950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7435215178158265e-05,
+      "loss": 0.0011,
+      "step": 3951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7429430819065248e-05,
+      "loss": 0.0007,
+      "step": 3952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7423646459972235e-05,
+      "loss": 0.2927,
+      "step": 3953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7417862100879222e-05,
+      "loss": 0.0001,
+      "step": 3954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7412077741786212e-05,
+      "loss": 0.0001,
+      "step": 3955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.74062933826932e-05,
+      "loss": 0.0002,
+      "step": 3956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7400509023600185e-05,
+      "loss": 0.0001,
+      "step": 3957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7394724664507176e-05,
+      "loss": 0.0002,
+      "step": 3958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7388940305414162e-05,
+      "loss": 0.0003,
+      "step": 3959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.738315594632115e-05,
+      "loss": 0.0003,
+      "step": 3960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7377371587228136e-05,
+      "loss": 0.0002,
+      "step": 3961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7371587228135126e-05,
+      "loss": 0.0001,
+      "step": 3962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7365802869042113e-05,
+      "loss": 0.048,
+      "step": 3963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.73600185099491e-05,
+      "loss": 0.0001,
+      "step": 3964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7354234150856083e-05,
+      "loss": 0.0019,
+      "step": 3965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7348449791763076e-05,
+      "loss": 0.0001,
+      "step": 3966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7342665432670063e-05,
+      "loss": 0.0001,
+      "step": 3967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7336881073577046e-05,
+      "loss": 0.0121,
+      "step": 3968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.733109671448404e-05,
+      "loss": 0.0002,
+      "step": 3969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7325312355391023e-05,
+      "loss": 0.0001,
+      "step": 3970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.731952799629801e-05,
+      "loss": 0.0003,
+      "step": 3971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7313743637204997e-05,
+      "loss": 0.0003,
+      "step": 3972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7307959278111987e-05,
+      "loss": 0.0142,
+      "step": 3973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7302174919018974e-05,
+      "loss": 0.0065,
+      "step": 3974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.729639055992596e-05,
+      "loss": 0.0018,
+      "step": 3975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7290606200832947e-05,
+      "loss": 0.0001,
+      "step": 3976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7284821841739937e-05,
+      "loss": 0.0001,
+      "step": 3977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7279037482646924e-05,
+      "loss": 0.0001,
+      "step": 3978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.727325312355391e-05,
+      "loss": 0.0002,
+      "step": 3979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.72674687644609e-05,
+      "loss": 0.0005,
+      "step": 3980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7261684405367888e-05,
+      "loss": 0.0001,
+      "step": 3981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7255900046274875e-05,
+      "loss": 0.0001,
+      "step": 3982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7250115687181858e-05,
+      "loss": 0.0007,
+      "step": 3983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.724433132808885e-05,
+      "loss": 0.004,
+      "step": 3984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7238546968995838e-05,
+      "loss": 0.0005,
+      "step": 3985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.723276260990282e-05,
+      "loss": 0.0001,
+      "step": 3986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.722697825080981e-05,
+      "loss": 0.0001,
+      "step": 3987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7221193891716802e-05,
+      "loss": 0.0002,
+      "step": 3988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7215409532623785e-05,
+      "loss": 0.0001,
+      "step": 3989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7209625173530772e-05,
+      "loss": 0.0037,
+      "step": 3990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7203840814437766e-05,
+      "loss": 0.0001,
+      "step": 3991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.719805645534475e-05,
+      "loss": 0.0012,
+      "step": 3992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7192272096251736e-05,
+      "loss": 0.0005,
+      "step": 3993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7186487737158723e-05,
+      "loss": 0.0001,
+      "step": 3994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7180703378065713e-05,
+      "loss": 0.0001,
+      "step": 3995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.71749190189727e-05,
+      "loss": 0.0001,
+      "step": 3996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7169134659879686e-05,
+      "loss": 0.0001,
+      "step": 3997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7163350300786673e-05,
+      "loss": 0.0001,
+      "step": 3998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7157565941693663e-05,
+      "loss": 0.0007,
+      "step": 3999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.715178158260065e-05,
+      "loss": 0.116,
+      "step": 4000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7145997223507637e-05,
+      "loss": 0.0001,
+      "step": 4001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.714021286441462e-05,
+      "loss": 0.0011,
+      "step": 4002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7134428505321613e-05,
+      "loss": 0.0001,
+      "step": 4003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.71286441462286e-05,
+      "loss": 0.0,
+      "step": 4004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7122859787135584e-05,
+      "loss": 0.0018,
+      "step": 4005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7117075428042577e-05,
+      "loss": 0.0008,
+      "step": 4006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7111291068949564e-05,
+      "loss": 0.0001,
+      "step": 4007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7105506709856547e-05,
+      "loss": 0.0001,
+      "step": 4008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7099722350763534e-05,
+      "loss": 0.0006,
+      "step": 4009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7093937991670528e-05,
+      "loss": 0.0007,
+      "step": 4010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.708815363257751e-05,
+      "loss": 0.0006,
+      "step": 4011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7082369273484498e-05,
+      "loss": 0.0001,
+      "step": 4012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7076584914391484e-05,
+      "loss": 0.0006,
+      "step": 4013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7070800555298475e-05,
+      "loss": 0.0002,
+      "step": 4014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.706501619620546e-05,
+      "loss": 0.0004,
+      "step": 4015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7059231837112448e-05,
+      "loss": 0.0002,
+      "step": 4016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7053447478019438e-05,
+      "loss": 0.0001,
+      "step": 4017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7047663118926425e-05,
+      "loss": 0.0002,
+      "step": 4018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7041878759833412e-05,
+      "loss": 0.0001,
+      "step": 4019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.70360944007404e-05,
+      "loss": 0.0001,
+      "step": 4020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.703031004164739e-05,
+      "loss": 0.0001,
+      "step": 4021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7024525682554375e-05,
+      "loss": 0.2902,
+      "step": 4022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7018741323461362e-05,
+      "loss": 0.0001,
+      "step": 4023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7012956964368346e-05,
+      "loss": 0.0001,
+      "step": 4024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.700717260527534e-05,
+      "loss": 0.0001,
+      "step": 4025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7001388246182326e-05,
+      "loss": 0.0001,
+      "step": 4026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.699560388708931e-05,
+      "loss": 0.0001,
+      "step": 4027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6989819527996303e-05,
+      "loss": 0.0003,
+      "step": 4028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6984035168903286e-05,
+      "loss": 0.0003,
+      "step": 4029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6978250809810273e-05,
+      "loss": 0.0001,
+      "step": 4030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.697246645071726e-05,
+      "loss": 0.0023,
+      "step": 4031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.696668209162425e-05,
+      "loss": 0.0001,
+      "step": 4032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6960897732531237e-05,
+      "loss": 0.0005,
+      "step": 4033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6955113373438223e-05,
+      "loss": 0.0003,
+      "step": 4034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.694932901434521e-05,
+      "loss": 0.0004,
+      "step": 4035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.69435446552522e-05,
+      "loss": 0.0,
+      "step": 4036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6937760296159187e-05,
+      "loss": 0.0003,
+      "step": 4037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6931975937066174e-05,
+      "loss": 0.0001,
+      "step": 4038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6926191577973164e-05,
+      "loss": 0.0002,
+      "step": 4039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.692040721888015e-05,
+      "loss": 0.0001,
+      "step": 4040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6914622859787137e-05,
+      "loss": 0.0002,
+      "step": 4041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.690883850069412e-05,
+      "loss": 0.0001,
+      "step": 4042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6903054141601114e-05,
+      "loss": 0.0001,
+      "step": 4043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.68972697825081e-05,
+      "loss": 0.0,
+      "step": 4044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6891485423415084e-05,
+      "loss": 0.0,
+      "step": 4045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.688570106432207e-05,
+      "loss": 0.0,
+      "step": 4046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6879916705229065e-05,
+      "loss": 0.0,
+      "step": 4047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6874132346136048e-05,
+      "loss": 0.0001,
+      "step": 4048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6868347987043035e-05,
+      "loss": 0.0077,
+      "step": 4049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.686256362795003e-05,
+      "loss": 0.0,
+      "step": 4050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.685677926885701e-05,
+      "loss": 0.0,
+      "step": 4051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6850994909764e-05,
+      "loss": 0.0001,
+      "step": 4052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6845210550670985e-05,
+      "loss": 0.0,
+      "step": 4053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6839426191577975e-05,
+      "loss": 0.0001,
+      "step": 4054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6833641832484962e-05,
+      "loss": 0.0,
+      "step": 4055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.682785747339195e-05,
+      "loss": 0.0002,
+      "step": 4056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6822073114298936e-05,
+      "loss": 0.0001,
+      "step": 4057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6816288755205926e-05,
+      "loss": 0.0,
+      "step": 4058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6810504396112913e-05,
+      "loss": 0.6498,
+      "step": 4059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.68047200370199e-05,
+      "loss": 0.0002,
+      "step": 4060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.679893567792689e-05,
+      "loss": 0.0013,
+      "step": 4061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6793151318833876e-05,
+      "loss": 0.0001,
+      "step": 4062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6787366959740863e-05,
+      "loss": 0.0001,
+      "step": 4063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6781582600647846e-05,
+      "loss": 0.0005,
+      "step": 4064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.677579824155484e-05,
+      "loss": 0.0002,
+      "step": 4065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6770013882461827e-05,
+      "loss": 0.0001,
+      "step": 4066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.676422952336881e-05,
+      "loss": 0.0001,
+      "step": 4067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6758445164275797e-05,
+      "loss": 0.0002,
+      "step": 4068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.675266080518279e-05,
+      "loss": 0.0005,
+      "step": 4069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6746876446089774e-05,
+      "loss": 0.0002,
+      "step": 4070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.674109208699676e-05,
+      "loss": 0.001,
+      "step": 4071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6735307727903754e-05,
+      "loss": 0.0003,
+      "step": 4072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6729523368810737e-05,
+      "loss": 0.0006,
+      "step": 4073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6723739009717724e-05,
+      "loss": 0.0004,
+      "step": 4074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.671795465062471e-05,
+      "loss": 0.0004,
+      "step": 4075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.67121702915317e-05,
+      "loss": 0.0003,
+      "step": 4076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6706385932438688e-05,
+      "loss": 0.0004,
+      "step": 4077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6700601573345674e-05,
+      "loss": 0.0005,
+      "step": 4078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.669481721425266e-05,
+      "loss": 0.0004,
+      "step": 4079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.668903285515965e-05,
+      "loss": 0.0003,
+      "step": 4080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6683248496066638e-05,
+      "loss": 0.0004,
+      "step": 4081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6677464136973625e-05,
+      "loss": 0.0002,
+      "step": 4082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6671679777880608e-05,
+      "loss": 0.0002,
+      "step": 4083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6665895418787602e-05,
+      "loss": 0.0002,
+      "step": 4084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.666011105969459e-05,
+      "loss": 0.0003,
+      "step": 4085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6654326700601572e-05,
+      "loss": 0.0006,
+      "step": 4086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6648542341508565e-05,
+      "loss": 0.0006,
+      "step": 4087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.664275798241555e-05,
+      "loss": 0.465,
+      "step": 4088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6636973623322536e-05,
+      "loss": 0.0002,
+      "step": 4089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6631189264229522e-05,
+      "loss": 0.0003,
+      "step": 4090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6625404905136512e-05,
+      "loss": 0.0003,
+      "step": 4091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.66196205460435e-05,
+      "loss": 0.0003,
+      "step": 4092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6613836186950486e-05,
+      "loss": 0.0003,
+      "step": 4093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6608051827857473e-05,
+      "loss": 0.0005,
+      "step": 4094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6602267468764463e-05,
+      "loss": 0.0004,
+      "step": 4095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.659648310967145e-05,
+      "loss": 0.0004,
+      "step": 4096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6590698750578436e-05,
+      "loss": 0.0002,
+      "step": 4097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6584914391485427e-05,
+      "loss": 0.0003,
+      "step": 4098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6579130032392413e-05,
+      "loss": 0.2577,
+      "step": 4099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.65733456732994e-05,
+      "loss": 0.0004,
+      "step": 4100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6567561314206383e-05,
+      "loss": 0.0006,
+      "step": 4101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6561776955113377e-05,
+      "loss": 0.0011,
+      "step": 4102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6555992596020364e-05,
+      "loss": 0.0007,
+      "step": 4103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6550208236927347e-05,
+      "loss": 0.0005,
+      "step": 4104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6544423877834334e-05,
+      "loss": 0.0004,
+      "step": 4105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6538639518741327e-05,
+      "loss": 0.0003,
+      "step": 4106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.653285515964831e-05,
+      "loss": 0.0005,
+      "step": 4107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6527070800555297e-05,
+      "loss": 0.0003,
+      "step": 4108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.652128644146229e-05,
+      "loss": 0.0004,
+      "step": 4109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6515502082369274e-05,
+      "loss": 0.0001,
+      "step": 4110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.650971772327626e-05,
+      "loss": 0.0006,
+      "step": 4111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6503933364183248e-05,
+      "loss": 0.0003,
+      "step": 4112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6498149005090238e-05,
+      "loss": 0.0007,
+      "step": 4113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6492364645997225e-05,
+      "loss": 0.0004,
+      "step": 4114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.648658028690421e-05,
+      "loss": 0.0007,
+      "step": 4115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.64807959278112e-05,
+      "loss": 0.0004,
+      "step": 4116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.647501156871819e-05,
+      "loss": 0.0004,
+      "step": 4117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6469227209625175e-05,
+      "loss": 0.0002,
+      "step": 4118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6463442850532162e-05,
+      "loss": 0.0002,
+      "step": 4119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6457658491439152e-05,
+      "loss": 0.0003,
+      "step": 4120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.645187413234614e-05,
+      "loss": 0.0002,
+      "step": 4121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6446089773253126e-05,
+      "loss": 0.0002,
+      "step": 4122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.644030541416011e-05,
+      "loss": 0.0003,
+      "step": 4123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6434521055067103e-05,
+      "loss": 0.0003,
+      "step": 4124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.642873669597409e-05,
+      "loss": 0.0003,
+      "step": 4125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6422952336881073e-05,
+      "loss": 0.0003,
+      "step": 4126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.641716797778806e-05,
+      "loss": 0.0005,
+      "step": 4127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6411383618695053e-05,
+      "loss": 0.0001,
+      "step": 4128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6405599259602036e-05,
+      "loss": 0.0001,
+      "step": 4129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6399814900509023e-05,
+      "loss": 0.0003,
+      "step": 4130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6394030541416017e-05,
+      "loss": 0.0003,
+      "step": 4131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6388246182323e-05,
+      "loss": 0.0002,
+      "step": 4132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6382461823229987e-05,
+      "loss": 0.0002,
+      "step": 4133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6376677464136973e-05,
+      "loss": 0.0002,
+      "step": 4134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6370893105043964e-05,
+      "loss": 0.0002,
+      "step": 4135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.636510874595095e-05,
+      "loss": 0.0055,
+      "step": 4136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6359324386857937e-05,
+      "loss": 0.0004,
+      "step": 4137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6353540027764924e-05,
+      "loss": 0.0001,
+      "step": 4138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6347755668671914e-05,
+      "loss": 0.0751,
+      "step": 4139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.63419713095789e-05,
+      "loss": 0.0002,
+      "step": 4140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6336186950485888e-05,
+      "loss": 0.0001,
+      "step": 4141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6330402591392878e-05,
+      "loss": 0.0001,
+      "step": 4142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6324618232299864e-05,
+      "loss": 0.0002,
+      "step": 4143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.631883387320685e-05,
+      "loss": 0.0001,
+      "step": 4144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6313049514113835e-05,
+      "loss": 0.0002,
+      "step": 4145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6307265155020828e-05,
+      "loss": 0.0003,
+      "step": 4146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.630148079592781e-05,
+      "loss": 0.0026,
+      "step": 4147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6295696436834798e-05,
+      "loss": 0.2042,
+      "step": 4148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6289912077741785e-05,
+      "loss": 0.0002,
+      "step": 4149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6284127718648775e-05,
+      "loss": 0.0003,
+      "step": 4150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6278343359555762e-05,
+      "loss": 0.0006,
+      "step": 4151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.627255900046275e-05,
+      "loss": 0.0002,
+      "step": 4152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.626677464136974e-05,
+      "loss": 0.2489,
+      "step": 4153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6260990282276726e-05,
+      "loss": 0.0002,
+      "step": 4154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6255205923183712e-05,
+      "loss": 0.0005,
+      "step": 4155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.62494215640907e-05,
+      "loss": 0.0001,
+      "step": 4156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.624363720499769e-05,
+      "loss": 0.0007,
+      "step": 4157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6237852845904676e-05,
+      "loss": 0.0005,
+      "step": 4158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6232068486811663e-05,
+      "loss": 0.2587,
+      "step": 4159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6226284127718646e-05,
+      "loss": 0.0004,
+      "step": 4160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.622049976862564e-05,
+      "loss": 0.0002,
+      "step": 4161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6214715409532626e-05,
+      "loss": 0.0002,
+      "step": 4162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.620893105043961e-05,
+      "loss": 0.0001,
+      "step": 4163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6203146691346603e-05,
+      "loss": 0.0001,
+      "step": 4164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.619736233225359e-05,
+      "loss": 0.0002,
+      "step": 4165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6191577973160573e-05,
+      "loss": 0.0001,
+      "step": 4166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.618579361406756e-05,
+      "loss": 0.0004,
+      "step": 4167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6180009254974554e-05,
+      "loss": 0.0005,
+      "step": 4168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6174224895881537e-05,
+      "loss": 0.0002,
+      "step": 4169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6168440536788524e-05,
+      "loss": 0.3198,
+      "step": 4170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.616265617769551e-05,
+      "loss": 0.0001,
+      "step": 4171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.61568718186025e-05,
+      "loss": 0.0003,
+      "step": 4172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6151087459509488e-05,
+      "loss": 0.0002,
+      "step": 4173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6145303100416474e-05,
+      "loss": 0.0002,
+      "step": 4174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.613951874132346e-05,
+      "loss": 0.0002,
+      "step": 4175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.613373438223045e-05,
+      "loss": 0.0003,
+      "step": 4176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6127950023137438e-05,
+      "loss": 0.0005,
+      "step": 4177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6122165664044425e-05,
+      "loss": 0.0001,
+      "step": 4178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6116381304951415e-05,
+      "loss": 0.0009,
+      "step": 4179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.61105969458584e-05,
+      "loss": 0.0001,
+      "step": 4180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.610481258676539e-05,
+      "loss": 0.0001,
+      "step": 4181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6099028227672372e-05,
+      "loss": 0.0001,
+      "step": 4182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6093243868579365e-05,
+      "loss": 0.5013,
+      "step": 4183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6087459509486352e-05,
+      "loss": 0.0001,
+      "step": 4184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6081675150393335e-05,
+      "loss": 0.0002,
+      "step": 4185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6075890791300322e-05,
+      "loss": 0.0002,
+      "step": 4186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6070106432207316e-05,
+      "loss": 0.0005,
+      "step": 4187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.60643220731143e-05,
+      "loss": 0.2979,
+      "step": 4188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6058537714021286e-05,
+      "loss": 0.0061,
+      "step": 4189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.605275335492828e-05,
+      "loss": 0.2957,
+      "step": 4190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6046968995835263e-05,
+      "loss": 0.4609,
+      "step": 4191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.604118463674225e-05,
+      "loss": 0.0032,
+      "step": 4192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6035400277649236e-05,
+      "loss": 0.0006,
+      "step": 4193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6029615918556226e-05,
+      "loss": 0.0007,
+      "step": 4194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6023831559463213e-05,
+      "loss": 0.0003,
+      "step": 4195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.60180472003702e-05,
+      "loss": 0.0043,
+      "step": 4196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6012262841277187e-05,
+      "loss": 0.0036,
+      "step": 4197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6006478482184177e-05,
+      "loss": 0.001,
+      "step": 4198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6000694123091164e-05,
+      "loss": 0.0011,
+      "step": 4199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.599490976399815e-05,
+      "loss": 0.0142,
+      "step": 4200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.598912540490514e-05,
+      "loss": 0.0016,
+      "step": 4201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5983341045812127e-05,
+      "loss": 0.0065,
+      "step": 4202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5977556686719114e-05,
+      "loss": 0.0352,
+      "step": 4203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5971772327626097e-05,
+      "loss": 0.0204,
+      "step": 4204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.596598796853309e-05,
+      "loss": 0.0007,
+      "step": 4205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5960203609440074e-05,
+      "loss": 0.0019,
+      "step": 4206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.595441925034706e-05,
+      "loss": 0.0004,
+      "step": 4207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5948634891254048e-05,
+      "loss": 0.0046,
+      "step": 4208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5942850532161038e-05,
+      "loss": 0.0011,
+      "step": 4209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5937066173068025e-05,
+      "loss": 0.0013,
+      "step": 4210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.593128181397501e-05,
+      "loss": 0.0134,
+      "step": 4211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5925497454882e-05,
+      "loss": 0.0007,
+      "step": 4212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5919713095788988e-05,
+      "loss": 0.1458,
+      "step": 4213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5913928736695975e-05,
+      "loss": 0.0014,
+      "step": 4214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5908144377602962e-05,
+      "loss": 0.111,
+      "step": 4215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5902360018509952e-05,
+      "loss": 0.0002,
+      "step": 4216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.589657565941694e-05,
+      "loss": 0.0003,
+      "step": 4217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5890791300323925e-05,
+      "loss": 0.0002,
+      "step": 4218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.588500694123091e-05,
+      "loss": 0.0005,
+      "step": 4219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5879222582137902e-05,
+      "loss": 0.0661,
+      "step": 4220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.587343822304489e-05,
+      "loss": 0.0002,
+      "step": 4221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5867653863951872e-05,
+      "loss": 0.0005,
+      "step": 4222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5861869504858866e-05,
+      "loss": 0.0002,
+      "step": 4223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5856085145765853e-05,
+      "loss": 0.0002,
+      "step": 4224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5850300786672836e-05,
+      "loss": 0.0002,
+      "step": 4225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5844516427579823e-05,
+      "loss": 0.0003,
+      "step": 4226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5838732068486816e-05,
+      "loss": 0.0005,
+      "step": 4227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.58329477093938e-05,
+      "loss": 0.0698,
+      "step": 4228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5827163350300787e-05,
+      "loss": 0.0004,
+      "step": 4229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5821378991207773e-05,
+      "loss": 0.0003,
+      "step": 4230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5815594632114763e-05,
+      "loss": 0.0002,
+      "step": 4231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.580981027302175e-05,
+      "loss": 0.0423,
+      "step": 4232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5804025913928737e-05,
+      "loss": 0.0007,
+      "step": 4233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5798241554835727e-05,
+      "loss": 0.0002,
+      "step": 4234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5792457195742714e-05,
+      "loss": 0.0053,
+      "step": 4235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.57866728366497e-05,
+      "loss": 0.0014,
+      "step": 4236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5780888477556687e-05,
+      "loss": 0.0001,
+      "step": 4237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5775104118463678e-05,
+      "loss": 0.0001,
+      "step": 4238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5769319759370664e-05,
+      "loss": 0.0004,
+      "step": 4239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.576353540027765e-05,
+      "loss": 0.0002,
+      "step": 4240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5757751041184634e-05,
+      "loss": 0.0004,
+      "step": 4241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5751966682091628e-05,
+      "loss": 0.0019,
+      "step": 4242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5746182322998615e-05,
+      "loss": 0.0002,
+      "step": 4243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5740397963905598e-05,
+      "loss": 0.0002,
+      "step": 4244
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.573461360481259e-05,
+      "loss": 0.0007,
+      "step": 4245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.572882924571958e-05,
+      "loss": 0.0004,
+      "step": 4246
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5723044886626562e-05,
+      "loss": 0.0024,
+      "step": 4247
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.571726052753355e-05,
+      "loss": 0.4426,
+      "step": 4248
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.571147616844054e-05,
+      "loss": 0.0002,
+      "step": 4249
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5705691809347525e-05,
+      "loss": 0.1035,
+      "step": 4250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5699907450254512e-05,
+      "loss": 0.0002,
+      "step": 4251
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.56941230911615e-05,
+      "loss": 0.0004,
+      "step": 4252
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.568833873206849e-05,
+      "loss": 0.0002,
+      "step": 4253
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5682554372975476e-05,
+      "loss": 0.0002,
+      "step": 4254
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5676770013882463e-05,
+      "loss": 0.0004,
+      "step": 4255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.567098565478945e-05,
+      "loss": 0.0001,
+      "step": 4256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.566520129569644e-05,
+      "loss": 0.0007,
+      "step": 4257
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5659416936603426e-05,
+      "loss": 0.0002,
+      "step": 4258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5653632577510413e-05,
+      "loss": 0.0004,
+      "step": 4259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5647848218417403e-05,
+      "loss": 0.0002,
+      "step": 4260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.564206385932439e-05,
+      "loss": 0.0002,
+      "step": 4261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5636279500231373e-05,
+      "loss": 0.0029,
+      "step": 4262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.563049514113836e-05,
+      "loss": 0.0016,
+      "step": 4263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5624710782045354e-05,
+      "loss": 0.0014,
+      "step": 4264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5618926422952337e-05,
+      "loss": 0.2312,
+      "step": 4265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5613142063859324e-05,
+      "loss": 0.0002,
+      "step": 4266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.560735770476631e-05,
+      "loss": 0.0004,
+      "step": 4267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.56015733456733e-05,
+      "loss": 0.2737,
+      "step": 4268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5595788986580287e-05,
+      "loss": 0.0004,
+      "step": 4269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5590004627487274e-05,
+      "loss": 0.0007,
+      "step": 4270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5584220268394264e-05,
+      "loss": 0.0011,
+      "step": 4271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.557843590930125e-05,
+      "loss": 0.0002,
+      "step": 4272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5572651550208238e-05,
+      "loss": 0.0002,
+      "step": 4273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5566867191115224e-05,
+      "loss": 0.0002,
+      "step": 4274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5561082832022215e-05,
+      "loss": 0.0003,
+      "step": 4275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.55552984729292e-05,
+      "loss": 0.0005,
+      "step": 4276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5549514113836188e-05,
+      "loss": 0.0006,
+      "step": 4277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.554372975474317e-05,
+      "loss": 0.0003,
+      "step": 4278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5537945395650165e-05,
+      "loss": 0.0002,
+      "step": 4279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5532161036557152e-05,
+      "loss": 0.0002,
+      "step": 4280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5526376677464135e-05,
+      "loss": 0.0001,
+      "step": 4281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.552059231837113e-05,
+      "loss": 0.0021,
+      "step": 4282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5514807959278115e-05,
+      "loss": 0.0002,
+      "step": 4283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.55090236001851e-05,
+      "loss": 0.002,
+      "step": 4284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5503239241092086e-05,
+      "loss": 0.0012,
+      "step": 4285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.549745488199908e-05,
+      "loss": 0.0016,
+      "step": 4286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5491670522906062e-05,
+      "loss": 0.0001,
+      "step": 4287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.548588616381305e-05,
+      "loss": 0.0003,
+      "step": 4288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5480101804720036e-05,
+      "loss": 0.0003,
+      "step": 4289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5474317445627026e-05,
+      "loss": 0.0006,
+      "step": 4290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5468533086534013e-05,
+      "loss": 0.0001,
+      "step": 4291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5462748727441e-05,
+      "loss": 0.0001,
+      "step": 4292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.545696436834799e-05,
+      "loss": 0.0001,
+      "step": 4293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5451180009254977e-05,
+      "loss": 0.0003,
+      "step": 4294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5445395650161963e-05,
+      "loss": 0.0846,
+      "step": 4295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.543961129106895e-05,
+      "loss": 0.0003,
+      "step": 4296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.543382693197594e-05,
+      "loss": 0.0018,
+      "step": 4297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5428042572882927e-05,
+      "loss": 0.0001,
+      "step": 4298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5422258213789914e-05,
+      "loss": 0.0058,
+      "step": 4299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5416473854696897e-05,
+      "loss": 0.001,
+      "step": 4300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.541068949560389e-05,
+      "loss": 0.0002,
+      "step": 4301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5404905136510877e-05,
+      "loss": 0.0001,
+      "step": 4302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.539912077741786e-05,
+      "loss": 0.0002,
+      "step": 4303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5393336418324854e-05,
+      "loss": 0.0021,
+      "step": 4304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.538755205923184e-05,
+      "loss": 0.0001,
+      "step": 4305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5381767700138824e-05,
+      "loss": 0.0001,
+      "step": 4306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.537598334104581e-05,
+      "loss": 0.0001,
+      "step": 4307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.53701989819528e-05,
+      "loss": 0.1947,
+      "step": 4308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5364414622859788e-05,
+      "loss": 0.0016,
+      "step": 4309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5358630263766775e-05,
+      "loss": 0.0032,
+      "step": 4310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.535284590467376e-05,
+      "loss": 0.0001,
+      "step": 4311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5347061545580752e-05,
+      "loss": 0.0002,
+      "step": 4312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.534127718648774e-05,
+      "loss": 0.0001,
+      "step": 4313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5335492827394725e-05,
+      "loss": 0.0027,
+      "step": 4314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5329708468301715e-05,
+      "loss": 0.0001,
+      "step": 4315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5323924109208702e-05,
+      "loss": 0.0281,
+      "step": 4316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.531813975011569e-05,
+      "loss": 0.0094,
+      "step": 4317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5312355391022676e-05,
+      "loss": 0.0001,
+      "step": 4318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5306571031929666e-05,
+      "loss": 0.0003,
+      "step": 4319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5300786672836653e-05,
+      "loss": 0.0001,
+      "step": 4320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5295002313743636e-05,
+      "loss": 0.0001,
+      "step": 4321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5289217954650623e-05,
+      "loss": 0.0001,
+      "step": 4322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5283433595557616e-05,
+      "loss": 0.1821,
+      "step": 4323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.52776492364646e-05,
+      "loss": 0.0001,
+      "step": 4324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5271864877371586e-05,
+      "loss": 0.3964,
+      "step": 4325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.526608051827858e-05,
+      "loss": 0.0001,
+      "step": 4326
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5260296159185563e-05,
+      "loss": 0.0002,
+      "step": 4327
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.525451180009255e-05,
+      "loss": 0.0001,
+      "step": 4328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5248727440999537e-05,
+      "loss": 0.0011,
+      "step": 4329
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5242943081906527e-05,
+      "loss": 0.0003,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5237158722813514e-05,
+      "loss": 0.0001,
+      "step": 4331
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.52313743637205e-05,
+      "loss": 0.001,
+      "step": 4332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5225590004627487e-05,
+      "loss": 0.0001,
+      "step": 4333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5219805645534477e-05,
+      "loss": 0.0003,
+      "step": 4334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5214021286441464e-05,
+      "loss": 0.4725,
+      "step": 4335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.520823692734845e-05,
+      "loss": 0.0003,
+      "step": 4336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5202452568255434e-05,
+      "loss": 0.0004,
+      "step": 4337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5196668209162428e-05,
+      "loss": 0.0001,
+      "step": 4338
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5190883850069415e-05,
+      "loss": 0.0001,
+      "step": 4339
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5185099490976398e-05,
+      "loss": 0.0002,
+      "step": 4340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.517931513188339e-05,
+      "loss": 0.0001,
+      "step": 4341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5173530772790378e-05,
+      "loss": 0.3178,
+      "step": 4342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.516774641369736e-05,
+      "loss": 0.0001,
+      "step": 4343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5161962054604348e-05,
+      "loss": 0.0002,
+      "step": 4344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5156177695511342e-05,
+      "loss": 0.0002,
+      "step": 4345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5150393336418325e-05,
+      "loss": 0.0005,
+      "step": 4346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5144608977325312e-05,
+      "loss": 0.0003,
+      "step": 4347
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.993658599715284,
+      "eval_loss": 0.028007030487060547,
+      "eval_runtime": 71.2957,
+      "eval_samples_per_second": 108.38,
+      "eval_steps_per_second": 6.775,
+      "step": 4347
+    }
+  ],
+  "max_steps": 8694,
+  "num_train_epochs": 2,
+  "total_flos": 9198970821324480.0,
+  "trial_name": null,
+  "trial_params": null
+}